P1993-9

WORDT MET IIITGELEEND TECHNISCHE UNIVERSITET

Laboratorium voor
Scheepshydromechanlca
Archlet
Mekelweg 2,2628 CD Delft
Tel.: 015-785873- Fax 015-781838

Proceedings of the

Twenty-Third
American
Towing Tank
Conference
 TECHNISCHE UNIVERSITET
Laboratorium voor
Scheepshydromechanloa
Archief
Mekelweg 2, 2628 CD Delft
Tel.: 015- 786873- Fax: 015- 781838
Proceedings of the

Twenty-Third
American
Towing Tank
Conference

edited by
Robert G. Latorre

Sponsored jointly by:

Naval Studies Board of the U.S. National Research Council
School of Naval Architecture and Marine EngineeringUNO
Society of Naval Architects and Marine Engineers
Office of Naval Research
Bollinger Machine and Shipyards
Dantec Electronics, Inc.
Waldemar Nelson and Co.
Newport News Shipbuilding & Drydock Company
M. Rosenblatt & Sons
Trinity Marine

NATIONAL ACADEMY PRESS

Washington, D.C. 1993
 The National Research Council serves as an independent advisor to the federal
government on scientific and technical questions of national importance. Established in 1916
under the congressional charter of the private, nonprofit National Academy of Sciences, the
Research Council brings the resources of the entire scientific and technical community to
bear on national problems through its volunteer advisory committees. Today the Research
Council stands as the principal operating agency of both the National Academy of Sciences
and the National Academy of Engineering and is administered jointly by the two academies
and the Institute of Medicine. The National Academy of Engineering and the Institute of
Medicine were established in 1964 and 1970, respectively, under the charter of the National
Academy of Sciences.
The National Research Council has numerous operating units. One of these is the
Naval Studies Board, which is charged with conducting and reporting upon surveys and
studies in the field of scientific research and development applicable to the operation and
function of the Navy.
A portion of the work done to prepare this document was pei fumied under
Department of Navy Contract N00014-87-C-0018 issued by the Office of Naval Research
under contract authority NR 201-124. However, the content does not necessarily reflect the
position or the policy of the Department of the Navy or the government, and no official
endorsement should be inferred.
The United States Government has at least a royalty-free, nonexclusive, and
irrevocable license throughout the world for government purposes to publish, translate,
reproduce, deliver, perform, and dispose of all or any of this work, and to authorize others
so to do.

Copies available from:

The 23rd American Towing Tank Conference

Dr. R. Latorre, Chairman
911 Engineering Building, University of New Orleans
New Orleans, LA 70148

Telephone: 504-286-7180
Fax: 504-286-7413

Printed in the United States of America

 SPONSORS

The Executive Committee of the 23rd American Towing Tank Conference extends
its thanks to those organizations whose financial contributions made this Conference
possible. Because of their generous contributions, copies of the Proceedings will be
distributed to all member organizations of the International Towing Tank Conference.

Naval Studies Board of the U.S. National Research Council

School of Naval Architecture and Marine EngineeringUNO

Society of Naval Architects and Marine Engineers

Office of Naval Research

Bollinger Machine and Shipyards

Dantec Electronics, Inc.

Waldemar Nelson and Co.

Newport News Shipbuilding & Drydock Company

M. Rosenblatt & Sons

Trinity Marine

111
 PREFACE

The Twenty-Third American The conference was well supported

Towing Tank Conference (ATTC) was by the University of New Orleans, the
hosted by the School of Naval College of Engineering, and the Office of
Architecture and Marine Engineering, Naval Research. We also received
University of New Orleans (UNO). The financial contributions from Bollinger
two-day meeting was held in the UNO Machine and Shipyards, Dantec
engineering auditorium on June 11-12, Electronics, Inc., Waldemar Nelson and
1992. Co., Newport News Shipbuilding &
Drydock Co., M. Rosenblatt & Sons, and
Fifty-nine delegates from Canada Trinity Marine. These bound proceedings
and the United States and one delegate were prepared through the kind assistance
from Israel attended the meeting. They of Mr. Lee Hunt, Naval Studies Board.
represented government laboratories,
commercial firms, and universities that use As chairman I am very grateful to
towing tanks for hydrodynamic testing of the executive committee and technical
ships, marine vehicles, and offshore committee chairs for an excellent technical
structures. The ATTC meets every three program. In closing, I would like to
years prior to the International Towing acknowledge four persons who helped with
Tank Conference (ITTC). The meeting the meeting, Mrs. Latapie, Mr. Fox, Mr.
goal is to improve test methods and Petitfaux, and my wife, Erika.
techniques by a timely exchange of
technical information and experience. Dr. Robert Latorre
During the two days 41 papers were Chairman, 23rd ATTC
presented in 11 sessions. The paper
length was set at 7 pages and the
presentations kept to 10 minutes with 5-
minute discussions. This made for a very
good exchange of technical information.

The opening day also included a

demonstration of the UNO tank's fiber-
optic laser doppler velocimetry (LDV)
system and a slide presentation by Mr.
Clifford Wiseman, Dantec Electronics, Inc.
After a bus tour of the lakefront, a
banquet of Louisiana seafood and
specialties was held at the Marriott Hotel.
The banquet speaker, Mr. Paul
Versowsky, Chevron USA, presented an
overview of offshore platform design and
the development of Chevron's compliant
tower for deep ocean production.

V
 EXECUTIVE COMMIllEE

Dr. Robert Latorre, Chairman Dr. J. Randolph Paulling

Department of Naval Architecture Department of Naval Architecture
School of Naval Architecture University of California
and Marine Engineering Berkeley, California
911 Engineering Building U.S.A. 94720
University of New Orleans
New Orleans, Louisiana Dr. Edwin P. Rood
U.S.A. 70148 Fluid Dynamics Program, Code 1132F
TEL: (504) 286-7180 Office of Naval Research
FAX: (504) 286-7413 800 N. Quincy Street
Arlington, Virginia
Mr. Robert F. Beck U.S.A. 22017-5000
Department of Naval Architecture TEL: (703) 696-4305
and Ocean Engineering FAX: (703) 696-0934
The University of Michigan
2600 Draper Dr. Daniel Savitsky
Ann Arbor, Michigan Davidson Laboratory
U.S.A. 48109-2145 Stevens Institute of Technology
TEL: (313) 764-0282 711 Hudson Street
FAX: (313) 936-8820 Hoboken, New Jersey
U.S.A. 07030
Dr. David C. Murdey TEL: (201) 420-5307
Institute for Marine Dynamics
Kerwin Place
P.O. Box 12093, Postal Station A
St. John's, Newfoundland
CANADA A1B 3T5
TEL: (709) 772-2481

Dr. William B. Morgan

David Taylor Research Center
Department of the Navy
Bethesda, Maryland
U.S.A. 20084-5000
TEL: (202) 227-1578

vi
 23rd AITC (JUNE 9-12, 1992) TECHNICAL COMMITTEE CHAIRMEN

Cavitation Propulsion

Dr. Robert Latorre Dr. Ki-Han Kim

911 Engineering Building Propulsor Technology Branch
University of New Orleans Code 1544
New Orleans, Louisiana, U.S.A. 70148 David Taylor Research Center
TEL: (504) 286-7180 Bethesda, Maryland, U.S.A. 20084
FAX: (504) 286-7413 TEL: (301) 227-2633/1450
FAX: (301) 227-5764

High Speed Craft

Resistance and Flow
Mr. John Hoyt III
David Taylor Research Center Mr. Gabor Karafiath
Code 1562 Design Evolution Branch
Bethesda, Maryland, U.S.A. 20084-5000 Code 1521
1EL: (301) 227-1216 David Taylor Research Center
FAX: (301) 227-5442 Bethesda, Maryland, U.S.A. 20084
TEL: (301) 227-1596
FAX: (301) 227-3883
New Facilities, Techniques,
Ice Testing and Validation
Steering and Maneuvering
Mr. John Zseleczky
U.S. Naval Academy Mr. Lewis Motter
Hydrodynamics Laboratory, Stop 11H Surface Ship Dynamics
Annapolis, Maryland, U.S.A. 21402 Code 1561
TEL: (410) 267-3361 David Taylor Research Center
FAX: (410) 267-2591 Bethesda, Maryland, U.S.A. 20084
TEL: (301) 227-1692
FAX: (301) 227-5442
Ocean Engineering

Mr. Ian Collins Seakeeping

Arctec Offshore Corp.
578 Enterprise Street Dr. Armin Troesch
Escondido, California, U.S.A. 92029 Department of Naval Architecture
TEL: (619) 746-5511 and Marine Engineering
FAX: (619) 746-3035 University of Michigan
Ann Arbor, Michigan, U.S.A. 48103
TEL: (313) 764-6470
FAX: (313) 936-8820

vii
 CONTENTS

SESSION I-A STEERING AND MANEUVERING

23rd ATTC Steering and Maneuvering Committee Report,

L. Motter, Chairman 3

Directional Stability Tests and Maneuvering Predictions

for a High-Speed Patrol Boat, by E.M. Lewandowski 11

Parameter Identification of Ships Maneuvering, by G. Zilman 19

Modeling Realistic Ship Behavior in Highly Restricted Waterways,

by L.L. Daggett 25

Interactions Between the Hull and Rudder During Ship

Maneuvering, by P.J. Kopp, N. Fuller, and R. Sedat 33

On the Full Scale Maneuvering Characteristics of the SWATH

Fishing Vessel Ali, by R.C. McGregor and A.F. Miller 41

SESSION I-B SEAKEEPING

The Seakeeping Committee Report, A.W. Troesch, Chairman 51

SHIPMO Seakeeping Predictions and Correlations, by B. Menon

and J. Vienneau 55

Initial Seakeeping Tests on an IACC Yacht Hull Model:

Two Ballast Conditions, by S.B. Cohen and R.F. Beck 65

Run Length and Statistical Error Estimation for Seakeeping Tests

and Trials, by R.D. Pierce 73

SESSION I-C OCEAN ENGINEERING

The Ocean Engineering Committee Report, J.I. Collins, Chairman 87

Comparison of Wave Reflection Equations with Experimental Wave Basin Data,

by D.C. Cotter and S.K. Chakrabarti 89

ix
 SESSION II-A RESISTANCE AND FLOW

The Resistance and Flow Committee Report, G. Karafiath, Chairman 101

Experiments and Predictions of the Resistance Characteristics of

a Wave Cancellation Multihull Ship Concept,
by M.B. Wilson, C.C. Hsu, and D.S. Jenkins 103

Computation of the Hydrodynamic Forces and Moments Developed on a

Body of Revolution at Combined Angles of Attack and Angles of
Drift and Yawing Angular Velocities, by Y.S. Hong 113

Resistance Tests with UBC Series Fishing Vessels,

by S.M. Calisal, J. Mikkelsen, and D. McGreer 121

Comparison of Wake Flow Measurements between 5-Hole Pitot Probes and

Laser Doppler Velocimetry in the Ship Propeller Plane,
by M. Pirrone and W. Lindenmuth 129

SESSION II-B PROPULSION

The Propulsion Committee Report, K-H. Kim, Chairman 139

Development of a Waterjet Propulsor for the Marine Corps High Water

Speed Landing Craft Application, by J. Stricker, A.J. Becnel,
and J.G. Purnell 147

Experience with Hot-Film Anemometry for Ship Model Nominal Wake Survey
in a Towing Tank, by S. Wu and N. Bose 157

Numerical vs. Experimental Cavitation Tunnel, by S.A. Kinnas and C.H. Mazel 165

Design Method and Application of an Asymmetric Stator Upstream of an

Inclined Shaft Propeller, by S.K. Neely, J. McMahon, and B.Y-H. Chen 175

Improvement of Surface Ship Propeller Cavitation Performance Using

Advanced Blade Sections, by J.W. Bailar, S.D. Jessup, and Y.T. Shen 185

Propeller Parametric Study for Mid-Term Fast Sealift Ships,

195
by K-H. Kim and J.F. Fraas
 SESSION II-C CAVITATION

The Cavitation Committee Report, R. Latorre, Chairman 205

An Expression of Inception Cavitation Number, by S. Qin 207

Cavitation Scaling Experiments with Axisymmetric Bodies,

by Y.P. Kuhn de Chizelle, S.L. Ceccio, CE. Brennan, and Y. Shen 211

Propeller Erosion Reduction with an Asymmetric Preswirl Stator,

by T.B. Smith and K.D. Remmers 219

Propeller Tip Cavitation Suppression Using Selective Polymer Injections,

by G.L. Chahine, G.F. Frederick, and R.D. Baseman 227

SESSION III-A NEW FACILITIES, T'ECHNIQUES, ICE TESTING,

AND VALIDATIONI

The New Facilities, Techniques, Ice Testing, and Validation

Committee Report, J. Zseleczky 237

The Large Cavitation Channel, by R.J. Etter and M.B. Wilson 243

Flow Field Measurements at a Free Surface Using Digital Particle

Image Velocimetry, by J. Hamilton, S. Fish, and D. Anthony 253

Calibration of the Wavemaker System in the Offshore Engineering Basin

at the Institute for Marine Dynamics, by J.J. Murray, L.M. Mak,
and M.A. Sullivan 261

Two Novel Free Surface Measurement Techniques Developed at DTMB,

by P.A. Chang, III, T.J. Ratcliffe, J. Rice, and S. McGuigan 269

Developments at University of New Orleans Towing Tank, 1989-1992,

by R. Latorre 279

SESSION III-B NEVV FACILITIES, TECHNIQUES, ICE TESTING,

AND VALIDATIONII

Development Studies for a Six Degree-of-Freedom Marine Dynamic Test

Facility, by M. Mackay, G. Unsworth, G. Panton, and Y. Thériault 289

A New Apparatus for Captive Model Testing of Submerged Bodies in

Coning Motion, by E.M. Lewandowski 295

xi
A Standard Method for the Conduct and Analysis of Ice Resistance Model
Tests, by D. Spencer 301

CD Model Ice - Preliminary Observations on the Effect of Density

on Tank Testing, by D. Spencer 309

The Deep Water Multi-Directional Wave Basin at the Offshore Technology

Research Center, by R.P. Johnson 317

Radar Imaging of Surface Ship Wake Flows, by K.R. Nicolas, C.S. Weller,
W.T. Lindenmuth, and B.H. Cheng 321

SESSION IV-A NEW FACILITIES, TECHNIQUES, ICE TESTING,

AND VAL/DAT/ONI//

Repeatability of Resistance Experiments in Ice with a 1:20 Scale Model

of the Canadian R-Class Icebreaker, by S. Newbury 327

2D Tests with One and Two Cylinders in Waves, by M. de Conti 335

A Precision, High Capacity, High Speed, Right Angle Drive Fully

Submerged Propulsor Dynamometry System (FSPD), by K.J. Anderson
and G.A. Hampton 339

SESSION IV-B HIGH SPEED CRAFTI

The High Speed Craft Committee Report, J.G. Hoyt, III, Chairman 349

353
A Note on the ATTC Friction Coefficient, by P.W. Brown (Contribution)

Discussion of the Origins of the Frigate and the Sloop

Constellation, by K.M. Lynaugh 355

A New Systematic Series of Planing Hulls, by G. Delgado, G.E. Lacy,

365
and C.C. Hsiung
373
Mathematical Simulation of Hovercraft Maneuvering, by G. Zilman and T. Miloh

Scaling of Main Spray Generated by Planing Surfaces, by S. Ryan and R. Latorre

383

xii
 SESSION IV-C HIGH SPEED CRAITII
Correlation of the USMC High Speed Technology Demonstrator and Propulsion
System Demonstrator, by J.G. Hoyt III, B.D. Abramson, and D.D. Hayden 391

Scaling SES [Surface Effect Ship] Motions, by D.R. Lavis 401

A Simplified Power Estimation Method for Monohull High-Speed Small Craft,

by C.M. Lee 407

23rd ATTC Group Photograph, June 12, 1992, at UNO 415

23rd ATTC Business Meeting Minutes 416

Aims and Organization of the American

Towing Tank Conference 417

Past Meetings of the American Towing Tank Conference 419

 SESSION I-A

STEERING AND NUNEUVERING

 5 May 92

Major Contributers to the ATTO

Steering and Maneuvering Committee Report

Dr. Rod Barr

Hydronautic Research Incorporated

Mr. Ash Chatterjee

U.S. Coast Guard Head Quarters

Dr. Jerome Feldman

Naval Surface Warfare Center
Carderock Division

Dr. Edward Lewandowski

Stevens Institute of Technology
Davidson Laboratory

Mr. Lewis Motter

Naval Surface Warfare Center
Carderock Division
Mr. Robert Sedat
U.S. Coast Guard R & D Center

2
 23rd ATTC Steering and Maneuvering
Committee Report
L. Motter
(Carderock Division, Naval Surface Warfare Center, USA)

The Steering and Maneuvering

Sub-committee, and the ATTO Steering
Committee would like to dedicate this
session to the memory of Professor
Martin A. Abkowitz from MIT. Professor
Abkowitz succumbed to cancer on Jan
26th of this year.
Dr. Abkowitz was professor of
Ocean Engineering, and an
international authority on ocean
vehicle hydrodynamics and ship
control. He was a regular participant
in the ATTO as a committee member, as
an author, and as a commenter for many
years.
Professor Abkowitz, who for many
years was the director of the MIT Ship
Model Towing Tank, was born in Revere,
Mass. in 1918. He received a SB in
naval architecture in 1940 and a MS
degree in 1949, both from MIT. He
received a Ph.d in physics from
Harvard in 1953.
From 1940 to 1942, he was a
naval architect with the David Taylor
Model Basin. During World War II he
served with the US Army's He was a Life Member of the
Transportation Corps and had the rank Society of Naval Architects and Marine
of Captain. He was released from the Engineers. He was awarded the
Army in 1946 and returned to the David Fulbright fellowship to the Technical
Taylor Model Basin. He joined the MIT University of Denmark (1962) and the
faculty in 1949 and worked in the University of Nantes,France(1971-
Department of Naval Architecture and 1972), and a von Humbolt fellowship to
Marine Engineering. He the University of Duisburg, Germany
became
associate professor in 1954 and full (1990-1991).
professor in 1959. One of Dr. Abkowitz's best known
His teaching and research maneuvering publications is "Lectures
focused on ship hydrodynamics,
on Ship Hydrodynamics, Steering and
particularly ship motions in waves and Maneuverability" published in 1964
the stability of ships and submarines. while he was working in Denmark [1].
He was responsible for the design, The lectures were included in Ref 2
construction, and operation of the MIT which was reprinted twice. This work
Ship Model Towing Tank. He taught and was one of the first complete
inspired many students, several of documentations on many of the theories
whom are currently on the faculty of used today by the industry.
what is now the Department of Ocean More recently, Dr. Abkowitz is
Engineering at MIT and other faculties known for his work on a simplified
throughout the world. He travelled approach to ship maneuvering system
and lectured extensively throughout identification [ 3 - He has
the world. demonstrated that simple maneuvering
His formal retirement occurred equations can be identified using data
in 1988, but he continued to be active collected with little more than a stop
in professional activities, including watch and installed on board
the 1989 ATTO, until he became ill instrumentation.
last year. His presence will be missed at
the ATTC.

3
UPDATE ON FACILITIES hydrodynamic forces and moments are
not modeled adequately in current
The hydraulically actuated twin motion simulators.
strut horizontal oscillator built at
the Taylor Model Basin in the 1950's, Maneuverincr of Submersibles at
Ref 4, has been scraped. It was Davidson Laboratory
possibly the strongest horizontal The Davidson Laboratory Coning
planar motion mechanism, HPMM, in the Motion Apparatus significantly expands
world, but the mechanical control the matrix of possible test conditions
by permitting the examination of
system was no longer functional and
the system was not in demand. The simultaneous yaw, pitch and roll
newer single strut wide amplitude HPMM angular velocities in captive model
is still available at the Carderock tests. Such devices have been used
Division, Naval Surface Warfare Center for many years in wind tunnel studies
(CARDEROCKDIV, NSWC). of rockets and aircraft; however, the
The HPMM and towing basin from CMA is the first to be used
what used to be Tracor Hydronautics underwater. The apparatus, consisting
The of a body of revolution model, six
Inc. are still available for use.
basin facilities are being operated by component force and moment transducer
Hydronautics Research Inc. and associated signal conditioning
Horizontal planar motion equipment, yaw adjustment sector and
mechanisms are also available at Artec drive shaft connector, was designed
Offshore Engineering in Escondido, the and built at the Davidson Laboratory
US Naval Academy and the University of in 1988. An extensive series of tests
Michigan. was completed in 1991.
The rotating arm basins at both
Laboratory and Maneuvering of Submersibles at
the Davidson
CARDEROCKDIV are operational. There CARDEROCKDIV
are papers from each of these The Defense Research Advanced
facilities in the session. Also there Projects Agency (DARPA) is sponsoring
is a brief description of a new a program of experiments and analyses
submersible model support and at the CARDEROCKDIV to (I) improve the
computer codes used to make
measurement system for each rotating
arm later in this report. predictions of submarine motions, (2)
improve existing experimental methods
IN THE METHODS FOR and (3) develop new methods to measure
IMPROVEMENTS
PREDICTING MOTIONS OF SUBMERSIBLES accurately the hydrodynamic forces and
moments and the flow around a
The motions of a maneuvering submarine performing severe maneuvers.
can consist of The current methods of performing
submerged body rotating arm
rolling, pitching and straight line and
simultaneous experiments are
yawing; indeed, it would be surprising captive-model
discussed in Ref 5 and 6. The
if the motions were restricted to a
plane even during routine equations of motion that are used for
single
maneuvers, due to coupling that exists predicting the motions of submarines
among the various motions. To predict are provided in Ref 7 and discussed in
these motions, knowledge of the Ref 5.
forces and moments on A new sting support system for
hydrodynamics arm, figure I, was
the body is essential. However, there the rotating
practical designed, fabricated and evaluated.
exist at present no
experiments
theoretical means of determining the An extensive program of conducted
using the new system was
forces and moments that arise from with a 17-foot model of the SSN 688
these combined rotations.
test submarine. The purpose of the
Furthermore, until recently the
experiments is to investigate
facilities could not duplicate these from
In combined rolling, trajectories of the vortices shed
conditions: and the appendages, the
and yawing, the crossflow the hull
pitching strengths of these vortices, and the
velocity varies along the body in both
magnitude and direction, whereas in location of the flow separation from
maneuvering tests employing rotating the hull for various conditions of
combined angles of attack and drift
arms or planar motion mechanisms, the pitching and yawing angular
velocity vectors are and
crossflow This work could not be
coplanar along the length of the body. velocities. L-shaped
performed with the existing
As a consequence, the roll-dependent

4
twin strut system because
there are large interference
effects between the struts and
the model when the model is lic.1.0 AT_
----]OTOIC

set at combined angle

a of
attack and angle of drift. ST I TABLE
The measurements
included the hydrodynamic
forces and moments developed ATER LIAIE
on the hull and appendages,
the hydrodynamic forces and
moments developed on only the
appendages, and the flow field
9.2'Dm
in the vicinity of the model 4"DMI

using particle displacement

velocimatry.
Particle displacement Ze
velocimatry involves the
illumination of sections of Figure 1 - Model and Sting Support System from CARDEROCKDIV Rotating Arrn Facility
the flow field with laser
sheets while seeding the water
with microscopic, neutrally
buoyant particles containing a small velocities (the ratio of the length of
amount of fluorescent dye. By pulsing the model to the radial location of
the laser more than once while the model on the rotating arm). The
recording a single photographic frame, yaw and pitch angles and radial
each particle leaves multiple traces position were set by using the
on the same film. Particle tracking positioning apparatus of the rotating
velocimetry involves matching traces arm. The roll angle was set using push
of the very same particle. The rods attached at the lower end to a
velocity is determined by measuring crank arm at the aft end of the sting.
the distance between these traces. The A preliminary analysis suggested
direction of the flow is identified by that in severe maneuvers there was a
recording the first trace longer than combination of flow separation and
the second trace. asy.matrical vortex shedding from the
The sting support system was unappended hull and from the sail both
designed to fit into the dry dock in of which induced circulation around
the Rotating Arm Facility. Hence the the hull. The two phenomena would be
overall length between the nose of the difficult, if not impossible, to
model and the aft end of the sting separate out from measurements on the
could be no longer than 22 feet. The fully appended configuration. Hence,
system was also designed so that (1) experiments were also performed on the
the maximum deflection angle of the unappended hull, the hull plus the
model would be no greater than 0.2 sail, and the hull plus the stern
degree, (2) the diameter of the sting appendages.
as it entered the model would result The results of the force and
in a minimum amount of area removed moment measurements, as well as the
from the stern planes and rudders, and photographs and video tapes of the
(3) the vertical support for the sting
would be located far enough away from
flow are being analyzed. The data
improve
will
our understanding of the
the model so that there would be no complex physical phenomena associated
significant interference effects. with severe submarine maneuvers. For
The sting support system example, the effect of pitch angle on
attaches to the rotating arm at the the pitching moment at relatively
same location as the existing vertical large yaw angles and yawing angular
struts. The arrangement of the block velocities appears to be significantly
gages and their attachment to the different than currently modeled.
model is the same as it is when struts
are used.
The test program included a MODULAR MANEUVERING MATHEMATICAL
range of roll angles (-15 to 30 MODELS
degrees), yaw angles (0 to 16
degrees), pitch angles (0 to -8 The ultimate goal in ship
degrees), and non-dimensional angular maneuvering simulation is a

5
 successful. A number
large of
standardized prediction technique with
of motion that can be maneuvers were performed. The raw
equations
determined from hull, and appendage time history data will be published as
Realistic maneuvering soon as it can be assembled. Further
geometry.
predictions could be made in early reduction of the data is necessary and
ship design without model experiments planned for the future.
in much the same way as seakeeping A series of 14 ship hulls has
predictions are done. Several been designed by the U.S. Coast Guard
researchers have prediction techniques and constructed with Coast Guard
based on empirical data, and other support. The hulls are approximately
have made maneuvering performance 6 ft long and include systematic
little or no variations in beam/draft ratio,
prediction with
experimental data [8], [9], [10]. length/beam ratio, block coefficient,
Most techniques are not fully and plan view projected area of the
documented and require an expert to bow and stern. Rotating arm
There are no standards. experiments are being planned for this
The data obtained will be used
run.
It is the current thinking of year.
many maneuvering experts that the in the development of the modular
problem of predicting the forces and prediction methods.
on a complete ship with Theoretical work has also been
moments this cooperative
simultaneously is too supported by
appendages of the
A better approach to the agreement. The objective
difficult.
problem may be to develop an current work is "to determine if a
understanding of the individual parts standard modular approach can allow
ship i.e. hull, rudder, bow simple data exchange between
of the
dome, propellers, etc. and to develop facilities, as well as establish some
prediction techniques for the force on degree of uniformity of simulated
between results that will facilitate
each and the interaction force basis of
on a wider
components [11]. The total force and acceptance conclusions obtained by use of ship
moment on the ship would be the
of all the modular simulation models." The many tasks
summation this work are not
components. associated with
This idea of a modular scheduled for completion until the end
maneuvering math model is gaining of the year.
support. The Maritime Administration,
MANEUVERING OF PLANING HULLS
the U. S. Coast Guard, and SNAME have
agreed to support work on the
In support of the US Coast
development of a modular prediction Guard's pursuit of R&D projects that
technique for U. S. researchers,
The work will enable it to evaluate advanced
simulator facilities, etc. the Davidson
being carried out under the marine vehicles
is Laboratory has completed an extensive
direction of SNAME Panel H-10, Ship
study of the directional stability of
Controllability. using rotating arm
References [12] and [13] present planning boats The model
of the first model tests of five models.
the results of
experiments to be conducted under this series is generally representative
experiments were a 100 ton, 100 ft hard-chine patrol
agreement. The
conducted at CARDEROCKDIV on a MARINER boat. Two of the models were notional
The other three have
Class ship and focused on the detailed WPB designs. of and form a
interaction effects between the hull length-beam ratio 5
having
and rudder as well as the hydrodynamic prismatic planing hull series degrees and
Of particular interest deadrise of 10, 20 and 30
hull forces. shapes representative of those
was the effect of in flow angle to the bow currently in service. The models were
rudder.
Following these model fixed in trim, roll and yaw and tested
at turning diameters of 8+ and 17 boat
experiments, full scale trials were lengths and on straight course. Test
conducted on the OBSERVATION ISLAND speeds correspond to 22.5, 45 and 60
which has a MARINER Class hull similar Underwater pictures of the
to the model. The trials were knots.
conducted by CARDEROCKDIV and wetted area of the models were taken
all conditions, and video
supported by the U.S. Coast Guard and in recordings of all runs were made.
the U. S. Air Force, who operated the
provided tracking The seakeeping behavior of the
ship and WPB hulls in head seas up to 15 ft
information. The trials were very

6
significant wave height was --

determined, and their beam sea (ABLE 1 - SHIP MANEUVERABILITY STANDARDS (PROVISIONAL) ESTABLISHED
rolling response at zero speed BY THE INTERNATIONAL MARITIME ORGANIZATION (IMO)
was measured using a new free-to 1 TURNING ABILITY (TURNING CIRCLE), ADVANCE 545 L: TACTICAL DIA 55.0 L
sway apparatus.
A new free oscillation
2 YAW CHECKING ABILITY (Z MANEtJVER)
apparatus was used to measure the
roll extinction characteristics 10/10 Z: 1ST 0/S s 15 DEG; 2, 3 8, 4 0/S 520 DEC
at high speed of the prismatic 20/20 Z 1ST 0/S 525 DEG;

hulls on straight course at fixed 3. INITIAL TURNING ABILITY:

trim and yaw. Empirical W/10 DEG RUDDER (PIS) THE SHIP SHOULD TRAVE 525 L
BY THE TIME HEADING HAS CHANGED 10 DEG
expressions for the added inertia
and roll damping were obtained. 4 COURSEKEEPING ABILITY (USING 10/10 Z MANEUVER)
The FOR UV 510 SEC. 1ST 0/S 5 10 DEG (V = SHIP SPEED, L = LBP)
results of the FOR LN 5 30 SEC 1ST 0/ 5 20 DEG
stability tests are presented in FOR LN IN-BETVVEEN, INTERPOLATE FOR ALLOWABLE 1ST 0/S
a database of well over 5,000 2ND 0/8 5 25 DEG, FOR ALL UV
runs, including tabulated values
NOTE: The Spiral Test is encouraged but not required
of forces, moments, drafts and
wetted lengths in both CONDITIONS FOR APPLICABILITY OF STANDARDS, The standards apply to deep.
unrestricted water, calm conditions, full load, even keel. Where trials are conducted in
dimensional and non-dimensional other than full toad, maneuvering performance in t he full load condition shall he predicted
form. based on trial information The predictive method used mar be model testing,

ICEBREAKER MANEUVERING DATABASE \414.

The Standards and Explanatory
Notes could effect the types of
The geometric configuration of maneuvering experiments and
icebreaker hull forms are considerably predictions required for future ships.
different than those of conventional One example is that in the draft
ships, particularly in the bow and Standards "The spiral test is
stern regions. As a result, the encouraged but not required." The
existing database for conventional working group concluded that the
ships cannot be applied to these spiral test results were not worth the
unique hull forms with confidence. A extra effort to obtain.
research program is underway at the
Davidson Laboratory to provide a ACTIVITIES OF MARINE BOARD COMMITTEE
hydrodynamic database which will ON ASSESSMENT OF THE USE OF SHIP
include existing icebreaker hulls and HANDLING SIMULATORS FOR WATERWAY
proposed designs with the Thyssen-WAAS DESIGN
bow design.
The Marine Board of the National
INTERNATIONAL MARITIME ORGANIZATION Research Council has completed an
MANEUVERING STANDARDS assessment of the application of
computer-based shiphandling
The International Maritime simulations to waterway design. The
Organization (LMO) created an ad hoc final report, Shiphandling Simulation,
working group under the Sub-committee Application to Waterway Design, is
on Ship Design and Equipment to scheduled for release in late June and
develop ship maneuvering standards. will be available from the National
The working group has considered Academy Press. Professor William C.
recommendations from all member Webster, University of California at
nations at the 1992 meeting and has Berkeley, chaired the committee that
submitted a draft for the standards to conducted the
assessment. Its
the sub-committee. The draft of the findings, conclusions and
standards is shown in Table 1. recommendations will
be of interest to
It is the intention of the those concerned with the ship
working group to develop "Explanatory maneuverability and controllability
Notes" to clarify the standards, give and to waterway project sponsors.
guidance for application of the The report addresses whether or
Standards and give guidance for not simulation should be used, and how
trials, model tests and predictions by simulation can be enhanced as a design
the 1993 meeting. Final approval of aid. The committee found that while
the Standards and Explanatory Notes is scientific, quantitative validation of
expected at the 1994 meeting. the results of simulations is not yet
available, reasonable success can

7
nevertheless be achieved in modeling
waterway complexities and the physical REFERENCES
environment. Case studies of
simulation application revealed that 1 Abkowitz, M. A., "Lectures on Ship
the technique can effectively aid in Hydrodynamics Steering and
decision making by providing unique Maneuverability", Report No. Hy-5,
quantitative information for answering Hydro-Og Aerodynamish
waterway design questions associated Laboratorium, Denmark, May 1964.
with channel depth, width, geometry,
dredging requirements, aids to A., "Stability and
2 Abkowitz, M.
navigation requirements, and tugboat Motion Control of Ocean Vehicles",
assistance. The cou.attee concluded The MIT Press, 1969.
that computer-based simulations
provide an improved means to assess 3 Abkowitz, M. A., "Measurement of
the operability of a proposed waterway Characteristics from
Hydrodynamic
improvement by approximating vessel Ship Maneuvering Trials by System
behavior in the full waterway SNAME Annual
thereby Identification",
operating environment, Meeting, Nov 1980.
offsetting the traditional reliance on
rules of thumb to provide adequate 4 Gertler, Morton, "The DTMB
margins of safety. The committee also Mechanism System,"
recommended a modest research program Planar-Motion
Proceedings of Symposium on Towing
to fill existing gaps in technical Tank Facilities, Instrumentation
knowledge and to make the use of and Measuring techniques, Zagreb,
simulations more attractive to
Yugoslovia, Sept 1959.
waterway project sponsors.
In concluding the assessment, Huang T. and J. Feldman, "Methods
5
committee members visited laboratories the Motions of
and simulator facilities in the United of Predicting
Submarines," Paper 88-WA/DSC-22
States, Japan, and Europe to determine
the state of practice and to identify Presented at the Winter Annual
Meeting of ASME (November -
primary shortcoming of existing
December 1988).
technology, including the underlying
mathematical maneuvering models. The and
a detailed 6 Feldman, J., "Straightline
report contains Arm Captive-Model
reference list. Rotating
mathematical model Experiments to Investigate the
Off recurring concern was the poor Stability and Control
existing ability to measure or predict Characteristics of Submarines and
hydrodynamic coefficients and
Other Submerged Vehicles," DTRC
resulting maneuvering performance with SHD-0393-20 (November
underkeel clearances of 10 percent or Report
less of draft, a situation frequently 1987).
encountered in existing and planned J., "DTNSRDC Revised
waterways and harbors. Research needs 7 Feldman,
include the establishment of fidelity, Standard Submarine Equations of
Motion," DTNSRDC Report
identification and validation
SPD-0393-09 (June 1979).
requirements for mathematical models.
also prepared a
The committee K., and V. Ankudinov,
comparisons of 8 Jakobsen B.
background systematic "Operator's Manual of Maneuvering
maneuvering models for the Esso Osaka Tracor
its assessment. Evaluation Program"
to support Hydronautics Tech Report 87078-4.
Comparisons were made of model test
for
data-based mathematical models and V. Poerio,
deep water and shallow water operation 9 Capurro, G.
sources, "Maneuvering Predictions for
from 12 and 7 different
Preliminary Ship and Port Design",
respectively. Joint International Conference on
Marine Simulation and Ship
Manoeuvrability in Japan, June
1990.

10.Kijima, K., et. al. "Prediction

Method of Ship Manoeuvrability in

8
 Deep and Shallow Waters", Joint
International Conference on Marine
Simulation and Ship
Manoeuvrability in Japan, June
1990.

11.SNAME Panel H-10 International

Work Shop on Modular Maneuvering
Simulation, New York Hilton, NY,
Nov 1991.

12.Kopp, P., R. Bishop, and L.

Motter, "Experimental Study of the
Rudder Flow Field and Its Effect
on Hull Forces", David Taylor
Research Center Report
DTRC/SHD-1341-01, August 1990.

13.Kopp, P., "Captive Model Rotating

Arm Test on the Bare Hull T-AG(S)
38 Represented by Model 4414,
David Taylor Research Center
Report DTRC/SHD-1341-02, August
1991.

9
 Directional Stability Tests and Maneuvering Predictions
for a High-Speed Patrol Boat
E.M. Lewandowski
(Davidson Laboratory, Stevens Institute of Technology, USA)

ABSTRACT solutions of the stability problems of

small high-speed craft discussed by
In order to assess the maneuvering Cohen and Blount (5).
qualities of a 120 ft USCG patrol boat A key difference between the
design, straight-course and rotating-arm turning of a high-speed planing craft
captive model tests were conducted. and a displacement ship is that the roll
Tests were carried out at two speeds, motion of the craft must be considered
two turning radii (plus straight- in the former case whereas rolling is
course), at a range of drift and roll generally neglected in the latter case.
angles, for a single load and LCG Thus to properly model the behavior of
combination. planing hulls it is essential to acquire
The data were analyzed to obtain data at a range of roll angles and to
simple expressions for hydrodynamic include the roll equation in the
forces and moments in terms of the test mathematical model of the vessel. The
parameters. These expressions were used data included in References 1-4 permits
in conjunction with the equations of this to be done.
motion of the vessel in four degrees of This paper presents turning
freedom to obtain trajectory predictions performance predictions for the U.S.
(including roll behavior) for the patrol Coast Guard 120 ft WPB Notional Design
boat. Hull based on the data of Reference 3.
This particular set of data was chosen
INTRODUCTION because it was obtained with the model
towed free to trim and heave whereas the
A series of experimental studies other data was obtained at a range of
on planing hulls is being conducted at fixed trim angles. In the latter case
the Davidson Laboratory in support of there is more curve-fitting to do and
the efforts of the U.S. Coast Guard to the pitch equation must be incorporated
improve its ability to design and in the mathematical model; the advantage
evaluate advanced marine vehicle of this approach is that any LCG
concepts (References 1-4). The location can be accommodated whereas the
experimental studies involve rotating- free-to-trim data applies only to a
arm and straight-course captive model single LCG location. Thus the present
tests to measure the hydrodynamic forces results only scratch the surface of
and moments which would be generated in potential uses of database.
turning and other maneuvers, on a series
of planing hull models. The results MODEL TESTS
constitute a data base which can be used
to assess the directional stability and Model and Apparatus
maneuvering qualities of existing A 1/18 scale model of the 120 ft
designs and of designs being considered WPB was used in the tests. The model
for future service. For example, it is lines are shown on Figure 1 and
expected that the data will provide principal characteristics are given in
insight into the causes and possible Table 1. Tests were conducted with and

11
without the rudders shown on Figure 1. ble; what was done in order to make use
Tests were carried out in the Davidson of the data was to fit the data for each
Laboratory High Speed Test Facility and of the two test speeds independently,
in the Rotating Arm Facility. The model and then fit the resulting coefficients
was towed free to trim and heave but to a quadratic equation in the velocity
fixed in roll, yaw, surge and sway. which was forced through the origin
Hydrodynamic axial force, side force, (i.e., zero force at zero velocity).
roll moment and yaw moment were measured The roll equation was handled different-
using a four component dynamometer; ly; with the exception of some roll
heave and trim were also measured. restoring moment terms, which were
independent of speed, the coefficients
Test Matrix were observed to vary linearly with
Tests were conducted at a single speed in the range of parameters covered
LCG location at two displacements (135 by the tests. Roll damping and added
and 155 long tons) and included para- inertia were determined using the
metric variations in drift angle (-6 deg empirical expressions in Reference 4.
to 12 deg), roll angle (-10 deg to 20
deg), speed (12.5 and 35 knóts) and Rudder Force and Moment
turning radius (4.8 and 2.4 boat Much of the data acquired in the
lengths, and straight-course). Tabula- experiments described above was obtained
tions of the data can be found in Refer- without rudders; this is the data which
ence 3. Typical results are shown on was analyzed in this study. In order to
Figures 2 and 3. examine the effects of the rudder size
and location on the turning performance
ANALYSIS of the craft, the semi-empirical
expression for fin lift presented in
In order to make use of the data Reference 6 was used:
in maneuvering simulations, a regression
analysis was performed to express the dCL/db = 1.8T/(1+2.8/Ae)
hydrodynamic axial force, side force,
roll moment and yaw moment as functions where A is the effective aspect ratio
of the test parameters. It was found to (equal eto twice the geometric aspect
be convenient to adapt a "modified wind ratio, for a fin on a flat surface such
axes" coordinate system centered at the as a planing hull). This equation was
CG in which the x-axis is longitudinal, applied in conjunction with the local
the y-axis is horizontal and positive to angle of attack at the rudder, which is
starboard, and the z-axis is vertical, a function of the yaw and roll angular
positive downward. The development of velocities as well as the geometric
the equations of motion and details of rudder angle. The moments induced by
the regression analysis will appear in a the rudders were taken to be equal to
subsequent paper which is now in
the lift force on the rudder multiplied
preparation. by the appropriate coordinate of the
The dependence of the hydrodynamic center of rudder force", which was
forces and moments on speed is more assumed to be at mid span on the
complicated for a planing hull than for quarter-chord line. The effects of the
a displacement craft, in part because propeller slipstream were not considered
the underwater shape of the hull changes in these initial calculations (see
markedly with speed. Thus normalization Appendix).
of the forces and moments using the
free-stream dynamic pressure did not RESULTS
prove to be satisfactory. Because only
two speeds were run, an in-depth analy- Predicted turning performance of
sis of the speed effect was not possi- the craft with the rudders as designed

12
 is summarized in Table 2. Calculations ACKNOWLEDGEMENT
were carried out at speeds of 12.5, 20
and 35 knots, under the assumption of The experimental program described
constant thrust. The table lists in this paper was sponsored by the U.S.
approach speed; rudder deflection; Coast Guard under Contract No.
dimensionless yaw angular velocity; DTCG23-85-C-20060. The advice and
steady roll angle; maximum (transient) support of Mr. James White of the U.S.
roll angle; final speed; and steady Coast Guard R&D Center, New London,
turning diameter in boat lengths. It is Connecticut, is gratefully acknowledged.
noted that at a speed of 35 kt, at
rudder angles greater than 20.5 degrees, REFERENCES
calculations indicate that the boat
rolls over; however in these cases the Klosinski, W.E., and Brown, P.W.,
predicted roll angles clearly fall "Directional Stability Tests of Two
outside of the range of the test data, Prismatic Planing Hulls", Davidson
so these results should be interpreted Laboratory Draft Report 2614,
with caution. March 1990.
Steady turning results are
summarized in Figures 4-7. The effects Brown, P.W., and Klosinski, W.E.,
of speed on turning performance are I,
Directional Stability Tests of a
apparent. Typical time histories of the 30 Degree Deadrise Prismatic
roll angle in turns at 20 degree rudder Planing Hull", Davidson Laboratory
angle are shown on Figure 8; the Draft Report 2658, July 1991.
pronounced roll overshoot at 35 knots
can be clearly seen. Klosinski, W.E., and Brown, P.W.,
Computer simulation is a useful "Course keeping Stability Tests
tool to quickly examine the effects of of a USCG 120 ft Notional Design
changes in appendage design. Figures 9, Model" Davidson Laboratory Draft
10 and 11 show the effects of variations Report 2570, January 1988.
of rudder span, chord and longitudinal
location on turning performance; it may Brown, P.W., and Klosinski, W.E.,
be seen that a 207. reduction in rudder "Experimental Determination of the
chord reduces the roll overshoot by 4 Added Inertia and Damping of
degrees (167.) without a significant Planing Boats in Roll", Davidson
change in the turning diameter. Figure Laboratory Draft Report 2632,
12 shows that the roll overshoot can be March 1991.
reduced by lowering the VCG of the
vessel. Cohen, S.M., and Blount, D.L.,
"Research Plan for the Inves-
CONCLUDING REMARKS tigation of Dynamic Instability
of Small High-Speed Craft", SNAME
This study of the turning Transactions, Vol. 94, 1986.
performance of a U.S. Coast Guard 120 ft
Notional Design Hull illustrates one Lewandowski, E.M., "The Effects
possible use of the extensive database of Reynolds Number, Section Shape,
for such hulls developed at the Davidson and Turbulence Stimulation on the
Laboratory under sponsorship of the Lift of a Series of Model Control
Coast Guard. This work could readily be Surfaces", Proceedings of the
extended to provide a more general 22nd American Towing Tank
design tool; for example, the effects of Conference, August 1989.
a skeg could be examined quite easily.

13
 TABLE 1 LOA 118.79
P 110'

VESSEL PARTICULARS

120 ft WFB

AP FP
/0 5 8 7 t. S 4. 5 Z.
o
Displacement, LT 135

Length overall, ft 118.79

Length between perpendicular, ft 110.0
LCG, forward AP, ft 42.90
VCG, above baseline, ft 7.66
rummmwmovniardviii
passumwsproin
21.1
Maximum beam at upper chine, ft

Rudder characteristics,
imonsurgerazu
Root chord, ft 2.33
Tip chord, ft 1.67

Span, ft 3.50
Area, sq ft 8.021
FIGURE 1 PROFILE AND BODY PLAN, 120 FT WPB
Rudder stock to LCG, ft 39.817
Model scale 1:18

Sida force
hrnirq rate: 1.4 boat leek
0.130

0.103

TABLE 2 12

Q-
PREDICTED TURNING PERFORMANCE M 0.0I0

Approach Rudder SD* 1* 0 max U **

dég deg kngts > IME
Speed, kt deg L

12.5 15.99 1.54 1.54 12.4

5 12.4
10 9.81 2.50 2.52
20 6.11 3.93 4.01 12.3 -0.050
1.0 13.0
4.50 4.88 5.18 12.1 0.0 3.0 6.0
30
3.87 5.07 5.67 11.8 Drift angle, der's
35

20 5 23.73 1.94 1.97 19.5

10 12.84 3.58 3.75 18.9 Yaw MOROI) t
20 7.17 6.44 7.41 17.8 UMW rode: 1.4 boat fames
30 5.12 8.82 11.53 17.0 0.001

35 4.44 9.85 13.88 16.6

-01119

5 31.16 3.88 4.15 33.3

35 30.9
10 15.53 7.06 8.49
20 7.78 12.91 24.68 26.2 1.(110

20.5 7.60 13.28 27.39 26.1

-0.260

a.,
\
rQ 400

-0.100

* STD: Steady turning diameter -0.120

*1 0s, Steady roll angle; Us: Final speed (constant thrust)

-0.140

1.0 12.0 0.0

00 3.0 i.0
Drift angle. deg

FIGURE 2 TYPICAL DATA, V = 12.5 KNOTS

14
 Rol 1 mantra Rol 1 nonent
Torok@ radios: 2.4 toe Tunas ralba: 2.4 Coat lengths
0.200 0.015

0.150 0.010

0.100
0.035
'.114 }ROLL -1o°
0.050
ID 0.00)
CO 003J to
9 0°
-0.005
-0.050
X10.
-0.010
-0.103

2
-0.150 -0,015

1.200 -0.020
00 3.0 6.0 1.0 U.0 15.0 00 3.0 6.0 9.0 12.0 11.0
Drift angle, deg Drift angle, deg
FIGURE 2 (concluded) FIGURE 3 (concluded)

Sida force STD us rudder angle

Tango ode: 2.4 loot loath% Origaanna
0.150

0.100

IV 0.00

0.000

-0.050
00 3.0 CO to 15.0
Drift angle, des CO LO MA HA 210 HA HA MA NA OA
Rudder angle. deg
FIGURE 4 STEADY TURNING DIAMETER cSTD) vs RUDDER ANGLE
AT VARIOUS SPEEDS
Yaw nomant
Tuning rota: 2.4 Init loves
0.003

Steadg rol I angla va rudder angla

Original ntlig
20.0

ROLL 20°
LO'
-0.1991
o°
11.0
ro

14.0
-0.060
35 knots
o° 12.0'

-0.010 110
e 20 knots
LO.

1.106 LO'
00 LO CO LO 12.0 15.0
Drift angle. deg 12.5 knots

FIGURE 3 TYPICAL DATA, V = 35 KNOTS

LO.

LO
/11r7V7.ir
00 LO 10.0 15.0 MA MA MA 35.0 0.0
Rudder deflection, deg
FIGURE 5 STEADY ROLL ANGLE vs RUDDER ANGLE AT VARIOUS SPEEDS

15
 Max roll angla us rudder angla Effect of Rudder Span
°Nadine* :Mt 35 kt Wit 20 deg
30.0 MA
MA
25.0
35 knots MA
20.0
MO
15.0
15.0
o 20 knots
10.0
104
O
5.0 12.5 knots 5.0
10 0.0
00 5.0 MA ox MA MA 304 35.0 410 20 2.2 2.4 2.6 2.8 3.0 3.2 3,4
Rudder mg le, des Fludder Sean, ft
FIGURE 6 MAxIMUM ROLL ANGLE vs RUDDER ANGLE AT VARIOUS SPEEDS FIGURE 9 EFFECT OF RUDDER SPAN ON TURNING DUALITIES
Spend reduction in turn Effect of Rudder Chord
Origearses !35kt Rutifer 29 tin
35.0
30.0
20 knOts
35 knots 15.0
Steady roll
10.0
E-
ST L
5.0
15 0.0
ZA VA LI 2.1 2.2 1.3 2.4
00 5.0 OA nx MA AA 304 15 1.6 1.7 1,8
Rudder neon chord. f t
2.0
Rudder mg le. deg
FIGURE 7 SPEED REDUCTION IN TURN FIGURE 10 EFFECT OF RUDDER CHORD ON TURNING QUALITIES
Roll mg Le us tine Effect of Rudder Location
Imd: 35 kt Mae 211 deg
MA
MA
Max roll
MA
Rudder deflectlon
MA
35 knots
nx Steady roll
WA STD/L
20 knots
5.0
5.0
12.5 knots
10
0.0
MA WA nx nx MA MA IA MI MA NA
00 OA MA MA MA Distance from rudder stock te LeG. ft
Tine. sec
FIGURE 11 EFFECT OF RUDDER LOCATION ON TURNING QUALITIES
FIGURE e -/ME HISTORY OF ROLL ANGLE IN TURNS. RUDDER ANGLE: 20 DEG
16
 30.0
Used: 35 ktkr: 213 tag
TABLE Al

23.0 PREDICTED TURNING PERFORMANCE

WITH EFFECTS OF PROPELLER RACE

max roll
Approach Rudder FrD* ** 0 max U **
Speed, kt deg L dgg deg kngts
12.5 5 15.39 1.58 1.60 12.3
10 8.95 2.72 2.76 12.3
Steady roll
15 6.51 3.70 3.79 12.3
20 5.13 4.56 4.73 12.3
25 4.13 5.25 5.57 12.2
5.0 20 5 22.13 2.07 2.12 19.4
10 11.96 3.84 4.07 18.8
15 8.43 5.47 6.10 18.2
0.0 20 6.61
00 6.98 8.27 17.7
LO 3.0 CO tO CO LO CO tO 110 25 5.46 8.41 10.66
KG, ft 17.2

FIGURE 12
35 5 29.64 4.05 4.36 33.2
EFFECT OF vcs ON ROLL BEHAVIOR IN STEADY TURN 10 14.75 7.33 8.98 30.6
15 9.78 10.37 15.16 28.0
19 7.75 12.90 25.49 26.3

APPENDIX * STD: Steady turning diameter

*s Ifs; Steady roll angle; Us: Final speed (constant thrust)
EFFECTS OF PROPELLER RACE
ON TURNING QUALITIES

For designs in which the rudders

are located in the slipstream of the
propellers, the propeller wash may have
a significant effect on the rudder
performance. The location of the
propellers for the 120 ft WPB Notional
Design is shown on Figure Al. Calcula-
tions were performed to obtain the
average axial inflow over the rudder
using simple actuator disk theory
(Reference Al). Based on Figure Al it
was assumed that 507 of the rudder area
is in the slipstream; the wake fraction FIGURE Ai
and thrust deduction were both taken to Propellers and rudderS, 120 ft WPB Notional Design

be zero.
Results are summarized in Table
Al. Figure A2 shows that the rudder STD vs rudder angle
Effect of PratIlitr Rae
effectiveness is enhanced by the
35.0

propeller wash, as expected; the largest

30.0'
effect occurs at high speed at small
Speet 33 knots
rudder deflections where a decrease in
turning diameter of 1.5 boat lengths due
to the propeller wash is noted.
15.0' Without propeller effect
REFERENCE o

10.0

ci
Al. Glauert, H., The Elements of speed 12.5 knots
m 5.1
Aerofoil and Airscrew Theory,
With propeller effect
Cambridge University Press, 1959. m
m m MO MO MD 25.0
Rudder angle, deg
FIGURE A2
17 Effect of proDelle, wesh on teady turning diameter
 141490.,,,,ow

Parameter Identification of Ships Maneuvering

(An 111-Posed Problem Formulation)
G. Zilman (University of Tel-Aviv, Israel)

Abstract Flowever, quite often the mathematical mod-

Attem.pts to reduce the difference between the results of el suffers from scale effects or improper selection of the
mathematical simulations and ship maneuvering trials by mathematical representation of hydrodynamical forces.
means of solving the inverse problem of dynamics gener- The International Maritime Organization (IMO) is
ally encounter considerable difficulties. May be, the main currently considering the establishment of maneuvering
obstacle for system idenizfiabiliiy problems is their insta- performance standard. In a.ccordance with the Resolution
bility, which results in the nonrealistic values of the hy- IMO A.601(13) it. is necessary to predict. some standard
drodyn amical coefficients. maneuvers of the ship with a. proper accura.cy. Thus, if
A practical method for estiinating parameters of the the mathematical model of a particular ship is not suffi-
mathematical model of the ship by means of solving the ciently accurate, one need to improve t,his model in order
consequence of direct problems is here proposed. The to provide the required precision of the sirnulation. In
method is based 071 minimazing the vector of cost func- this case maneuvers, corresponding inputs and the accu-
racy of the simulation are given. It, should be stressed,
tions, which represents the "di.stance" between the object
and it's mathematical 1110dei. that these conditions are essential a,nd impose additional
This method allows to take into account the physical requirements on the algorithm of identification.
restrictions imposed 071 the hydrodyn amical coefficients Sy..stem identification technique is an alternative way
for estimating the hydrodynamical coefficients. Never-
and to select the Pareto-optimal solution, which satisfy
all required functional constraints and, thus, to avoid the theless, an emplia.sis on working directly wit,h trials data
occurrence of "cancellation effects." often gives a gross disagreement between the towing tank
and the identified coefficients.
Despite of existing considerable works ( [1] 1, [2]),
INTRODUCTION why are there apparently so few examples of successful
identification of a maneuvering ship?
Although extensive time and effort has been expended
in recent years in the development of more exact ship Mar be, the limited number of examples is mainly due
maneuvering identification technique, the extra.ction of to the most obvious symptom of identifiability problems,
hydrodynamical coefficients from maneuvering trials data i.e. to the physically nonrealiza.ble parameter estimates.
still remain a. difficult, and time-consuming problem. This symptoms may arise from a. number of different rea-
SOns, e.g.
Before exa.mining the problem of identification of a
maneuvering ship, the following question should be an-
the mathematical particular model chosen for iden-
swered: what, are the main reasons for identifying ? tification is inadequate;
The growing demands for mathematical simulation
of ship maneuvering and the necessity of implementa- the input, sequence does not excite the considering
tion this simulation using ship maneuvering simulators output, the output. is too wea.k or the input choice
(including ship maneuvering board stations), lea.c1 to a is poor, so that some of the ship response variables
general recognition of the significance to predict, the mo- are linearly. correlated;
tion of the ship with a relatively high a.ccura.cy. The there are large, una.ccounted for instrumentation er-
ship performance prediction requires t,o identify the hy- rors and una.ccounted input, noise.
drodyna.mic coefficients of the mathematical model. The
hy/drodynarnical forces which are needed for the numeri- However, let. us assume that. the mathematical mod-
cal sinnilation can be obtained by means of constrained el is accurate as far as possible, the input sequence is
model tests in the towing tank and by theoretical cal- 'see also the surycy, ibid.

19
chosen properly and both, the instrumentation error and THE OUTLINE OF THE PROBLEM
the input noise are small. Is there any guaranty, that the A system of ordinary nonlinear differential equations is
identification of the system will be "successful" ? The usually used as a structural mathematical model of ship
general answer should be negative. Why is it so ? maneuvering. These equations are \veil known and under
As a mathematical problem, the system identification some assumptions can be written as follows [11:
problem must satisfy some conditions. One of the impor-
tant. properties of mathematical problems is the stability = f[x(t), a, u]; x(0) = x° (2)
of their solution to small changes in the initial data. If
where x is the n-dimensional state vector, u is the m-
the initial data are known only approximately (it con-
dimensional input control vector and a is M dimensional
cerns the ina.dequa.cy of the mathematical model also),
then the instability of solutions leads to a nonuniqueness vector of hydrodynamic coefficients.
of the classically-derived approximate solutions and to The components of the state vector are:
serious difficulties in their physical interpretation. Snch
surge velocity,
problems are said to be ill-posed [3].
sway velocit,y,
All available methods for solving ill-posed problems
are based on using the supplementary a'priory informa- yaw angular velocity,
tion, but all of them. e.g. equation error estimates, weight- propeller rate of rotation,
ed least squares estimates, maximum likelihood estimates, hea.ding angle,
extended Kalman filter, reduce to minimazing (maximaz-
coordinates of the center of gravity in the laborato-
ing) of only one cost function [4].
ry coordinate system.
The way of introducing the supplementary informa-
tion depends on the formulation of the problem. Just to We consider two components of the input vector u i.e.,
give a simple exa.mple let, us consider the weighted least the rudder deflection 6(0 and the torque moment of the
square method. In this case the cost function (the objec- sh ip engine.
tive) has the form of The solution of the Ca.uchy problem (2) can be written
as

X = x(i,a,x°) (3)
= Y()]T1V.[z(li) Y(1)1 (1)
The classical method of estimating the hydrodynam-
ical coefficients is based on measurements of hydrody-
namical forces , which are involved in (2), by using con-
where z is an (n x 1) vector of measurement, y is an strained tests of a model in the towing ta.nk. In this
(n x 1) model response vector. W is a positive definite
weighting matrix and N is the number of data points. case the problem eventually reduces to the selection of
an appropriate mathematical method of parametric ap-
Thus, all the information which we possess about ev- proximation of these forces
ery component zn, reduces to only one objective J. which
as a matter of fact, has no clear physical meaning. f = f(a) (4)
The mathematical preferences of minimazing of only Identification should give some para.meters A, such
one cost function are almost obvious. However, the first that the beha.vior of the object and of the model be as
problem which we have to tackle concerns the question close as attainable. This similitude is to be determined
whether it is pra.ctically expedient and theoretically fea- mathematically, consequently it is necessa.ry to introduce
sible to employ a number of cost functions with a clea.r the definition of the distan ce R(0, M) between the ob-
physica.1 sense in order to provide the simulation of prac- ject 0 and the model M or, in other words, the metric,
ticallv essential maneuvers with a desirable or attainable
which is usually determined by formulation of the prob-
accura.cy. lem.
Herein we a.lso deal with the problem of estimating Generally speaking, the distance satisfies the condi-
the hydrodynamical coefficients, which are involved in tion
the ship maneuvering equations, grounding on the theo-
retical a.ncl experimental a'priory information about. each R(0. M) = e (5)
of the components of the response vector, bounds of each is equal
hydrodyna.mical coefficient and the functional relations where the cost function e is always positive and
between them. to zero only in one exclusively. case

20
 M (6)
= xi"2(t., a. u ) xi (t ) (p = 1, 2, - -) (8)
Practically the distance can not be reduced to zero
for the reasons, we stated above. Even in the theoretical Ivliere -= 1,2 ... u. j = 1, 2, , I, and lis the number of
-
case, where the characterization is perfect, the measure- inputs.
ment noise generally results in a. minimum R = e which With the selected metric pr(i-cr) we have
is not zero. On the other hand, even if the noise and the
error in the measurements are not presented, the adequa- pr = e(Kr)
cy between 0 an M. especially for such a complex craft (9)
like a. ship, is unattainable. 2. The aim of multiobjective identification is to reduce all
Mathematically speaking. the system identification pro- components of the vector cost function B. = p, .

blem reduces to the minimizing of the functional (5), or, However, as a rule, this can not be done for all compo-
in other words, we are required to find an element, â E A, nents simultaneously.
at which the given functional (5) attains its smallest, value A point in the objective variable space is ( rough.ly
on the set A speaking ) Pareto-optimal if one can not find another
point which is better with respect io at lea.st one objective
= aril mi.?? R(a) (7) without sacrificing with respect to at ¡casi other objective.
We need to distinguish between two kinds of such pro-
This means that a different point Unproves some ob-
jectives but deteriorates at least one of the rest. The
aim of the decision making is to select, one point among
The problem of minimizing R under assumptions
that it is not important which vector a provide the sought all the Pareto-optimal points, which can be accepted as
771 in i717 U771 .
a compromise solution. Obviously, the decision maker
The problem in. wh.ich we need to find ¡he vector
should be able to select. the better point, thus, he has to
a that minimize 1/te functional R ( minimiza/ion with understand the physical nature of the object. However,
respect to the argument). the better point can be selected by using one preferable
objective. Such a preferable objective is a question of
Concerning systems identification, the instability- of
problem B is encountered fairly frequently and can induce agreement between experts (for example, it can be the
drastic fluctuations of the hydrodynamic coefficients. error of simulation of the minimal stea.dy turn radius).
Hence, after utilization system identification methods one
ca.n obtain the numerical values incompatible with their TEST SIGNALS AND COST FUNCTIONS
physical significance. Since the estimated parameter val- Every particular input ui excites the correspondent re-
ues may be quite different, from what hydrodynamics the- sponse of the ship. The choice of the test signal in
ory may indicate, therefore, the ability of the model to identification experiments has a. significant bearing up-
predict the behavior of the vessel under inputs, different. on the a.chievable accura.cy. However, the precise nature
from those with which the system vas identified, can be of the rudder perturbations is not critical as long as the
insufficient. Thus, when the parameters are estimated. perturbation excite the ship motion sufficiently-. Special
there is no guarantee that the chosen vector a is actually clyna.mics experiments are made during delivery tests in
s.ufficiently good for the intended purpose. order to evaluate the steering and maneuvering abilities.
In other words, the class of approximate solutions A Typical of them are the spiral and zig-zag tests. In the
can be too broa.d. A rule for selecting the possible solu- spiral test the rudder angle is set to
a fixed value. This
tions is thus necessary. For this purpose ve should use results in spiral motion which converges to
a steady state
supplementary a 'priory information. This information circular motion. The experiments is then repeated for d-
concerns the choice of metrics, the error of the measure- ifferent. values of the rudder angle. The spiral test yields
ments, the functional relation between the hydrodynam- the following cost function:
ical coefficients and their physical constraints.
pi, =17'07 i'' oil 1
MULTIOBJECTIVE IDENTIFICATION (10)
pup = IU,71 I

When the parametric model is selected, the error ni be- where (6 = ; i = 1,2, I), r', denote the the steady
tween the measured component ein and it's estimate ya.w velocity and U.; is the surge velocity in the steady
can be written as follows state circular motion. These criterions have clear phys-
'In the author's opinion the sufficient proof of this a.ssertion ical sense and are practically essential. One can hardly
is the variety of hydrodynamical models in the theory of ship imagine that the expert, (a. master or a pilot) is able to
maneuvering
make a decision about the adequacy of the mathematical

21
model without. verifying the steady state characteristics. where //,(a) are such functional constraints, which reflect
Since the requirements for the accuracy of simulation of some basic features of ship dyna.mics. Herein, among a
these characteristics are quite definite, thus they have to set of possible restrictions, we consider only one of them.
satisfy the following conditions Under some a.ssumptions the linearized equations (2)

Pr,u < prna(11) with a zero right-hand side (the rudder(s) fixed at neutral
angle) can be written as a pair of decoupled second order
equations:
where the superscript mar denotes the maximal error of
simulation. T1T24+ (T1 + T2)4 + q = 0 (15)
In the zig-zag test the rudder is changed from star-
board to port in such a way that a stable zig-zag motion where q denotes non-dimensional yaw rate or sway veloc-
is obtain. Since this motion is periodical and all responses ity. It renders the characteristic stability equation for the
for different, ships are similar , this maneuver can be de- yaw-sway mode. i.e.
fined by two quantities i.e, by the period T. of oscillations
k= + (1/7-1 + 1/T2)k + 1/(7'12'2) = 0 (16)
and by the amplitude A. of the resulting heading angle
0 (yaw velocity or resulting trajectory). Consequently, it For a merchant ship the roots of the characteristic
is reasonable to introduce in addition the following cost. stability equation (stability indices) k1.9 1/Ti,, have
functions to satisfy the following conditions:

P.4]
both roots are real quantities;
(12)
PT j Tv,i1 the larger by the absolute value root k1 is always
= }- I
negative;
where (j = 1,..., J) and 6i (t) represents the sequence of if the ship is dyna.mically nonstable then the smaller
inputs signals 100 100, 200 20`'. 30" 30°, 20° 100.
by absolute value root. k2 > 0, if the ship is dynam-
etc. ically stable then k2 < O;
The values of PA] and pi,' also physically clear and
ha.ve to satisfy the inequality similar to (11). Certainly, at 1k1 1k21

the sa.me time, it is possible to get a.nother cost functions,Practically, the last inequality can be expressed more def-
for example the overshoot. heasding angle, functional (1). initely as lk1/k21 > 2. Usually it is very difficult to ob-
etc. tain a sufficiently accurate mea.surement of sway velocity..
Consequently, all terms oil the right-side of (2), where the
FUNCTIONAL CONSTRAINTS sway velocity is involved, can not be satisfactory estimat-
ed . However, it is well known, that for a conventional
For identification algorithms the specific particular math- merchant ship. stea.dily moving with a rudder fixed at a
ematical form of the expressions of hydrodynamical forces given angle. the following inequalities should be satisfied
is not really important.. More critical is the feasibility to
evaluate hydrodyna.mical coefficient as a function of the /mi"() < 1 < 1m"(6) (6 20° 30°) (17)
ship, rudder and propeller geometry. Evidently-, some
hydrodynamical coefficient depend on the kinematics of where / 30/7.' and 3 is the drift angle. Usually lmin'
the ship motion and the propeller frequency of rotation. 0.25 and 1"i" 0.5 Constraints (17) are useful even in
Thus, such a hy:drodynamical model is preferable. which the case when measurements of the drift angle (or surge
correctly accounts for fluid physics. isolates the forces ex- and sway velocities respectively) can be done.
erted on each of the separate components, accounts the
interaction between the various components and allow THE ALGORITHM FOR MULTIOBJECTIVE
the mathematical model to be studied without perform- MINIMIZATION
ing model tests. The main difficulty of multiobjective optimization is to
"Modular" mathematical models (see for example [5]),
scan the Pareto surface Dr in a small number of steps.
[6], [7], [8]) satisfies these conditions and allow to estab- The small number is important beca.use the determina-
lish some physical constraints on ea.ch component of the tion and evaluation of one point of this surface can be
vector a as follows implemented only by solving numerically the state equa-
tions (2) and masy cost a lot of computer time.
a'stnin < < at"'" (t 1, 2. ) (13)
In the space of para.meters A the components of the
vector a with constraints (13) fill the M-dimensional hy-
hTi" < (a) < 1.17":" (s = 1,2, ) (14) per cube V. The strategy of distribution of M test points

22
in the hyper cube D depends on the available information Table 2 Part of the table of satisfactory points
about the location of the domain Dr E D which satisfies (zig-zag 10°/10°)
all constraints.
Just, to give an exa.mple. let us consider the function Point No. Ttr,',' - Tt,l, sec. T,., sec.
z = F(ai, a.,) where O < a1 < 1. The minimization of 34 1.3 202.3
F with respect to the argument a can be done by eval- 269 35 197.5
uation of F(aii,a,;), with the subsequent search 24 4.2 205.2
of the minimal element, of the array Certainly, the 221L 4.7 196.3
sequences falil and may be defined by uniform l91L 4.9 20:5.9
subdividing the sides of the square 0 < a12 < 1 If .

z is a weakly dependent function of one of the argu- Table 3. Part of the table of satisfactory points
ment al , this algorithm can not be effective, beca.use (turning circle 6 = -20°)
all computations with respect, to al practically yield in-
significance additional information about the surface .-..- Point. No. 17)(7: -
i'l o o
F(ai, a2). More efficient is the method based on filling :37 0.003 0.334
the the square by uniformly distributed points (so called
4:3 0.007 0.338
LP, sequences [9]).
221L 0.027 0.304
The LP, method can be used in M dimensional space 312 ' 0.031 0.300
of parameters a and P dimensional space of objectives. 191L 0.041 0.372
Concerning the present, problem it, can be realized on a. 54 0.049 0.380
desktop computer IBM PC 486 with a coprocessor, if the
number of parameters do not exceed 40 and the number
of cost functions do not exceed 10. After showing the Pareto-optimal solution (in the ex-
a.mple they are marked by triangular), the decision maker
The algorithm for constructing the Pareto-surface can is only requested to indicate the desirable point.
be done in accordance with the following steps It follows from tables 1-3, that only two satisfa,ctorv
Step 1. The sequence of N vector aN . which are points (No. 191 and 221) are in common.
uniformly distributed in the hyper cube D. is generated. The additional prompt relates to the explicit prefer-
Srep 2. The selection of points which satisfy all con- a.ble objective, for exa.mple, to the the minimal error in
straint (satisfactory points) is implemented. estimating the turning circle nondimensional angular ve-
Step 3. The table of Q first, satisfactory points, where locity. Under this a.ssumption the point No. 221 is opti-
the cost functions follow in ascending order, is construct- mal. If the explicit, preferable cost function is known, the
ed. procedure of identification can be implemented by default
Step 4. Pareto-optimal points are found. (a.utornatically), but. usually the interactive procedure is
more obvious and instructive.

Example
SUMMARY AND CONCLUSIONS
Herein we present some examples of estimating the pa-
Multiobjective systeni identification procedure have been
rameters of the mathematical model [8] using trial results
successfully applied to 10 ship forms [il], [12]. Notably,
[10]) (Mari 71e r).
in accordance with the author's experience, after finding
Pareto-optimal points the decision can be ma.de rather
easily because usually only few Pareto-optimal points ex-
ist.
Table 1 Part of the table of satisfactory points
(zig-zag 10°/10') However, some cases were encountered, for which the
set of Pareto-optimal points is empty. It reflects the fact
of possible incompatibility of the trials and the adopted
Point No. 1.47,7: - .4.,,,,I, deg. .4,,, deg. mathematical model. Certainly, that in this situation the
221,L 1.73 26.27 problem has to be reconsidered.
147 1.84 29.84
483 2.63 25.37
101 2.8
Acknowledgments
30.08
191.L 3.84 :31.84 The a.uthor is indebted to Prof. T.Miloh for useful dis-
14 3.96 31.96 cussions.

23
References
Abkowitz, M.A. (1980). Measurement of Hydrody-
namic Characteristics from Ship Maneuvering Trial-
s by System Identification. SNAME Trans.,, v. 88,
pp.283-318.
Aström, K.J., Kallström, C.G. (1973) Application
of system identification techniques to the determina-
tion of ship dynamics. Proc. of the 3rd IFAC Sympo-
sium, Hague/Delft. Netherlands, American Elsevier
Pub!. Comp.,Inc., N.Y.
Tikhonov, A.N., Arsenin, V.Y (1977) Solution of ill-
posed problems. V.H. Winston k. Sons, Washington.
D.C.
Sage, A.P, Melsa J.L. (1971). Systen2 identification
Academic Press, N.Y.
Katsuto Kijima et. all (1990). On a numerical sim-
ulation for predicting of ship maneuvering perfor-
mance. Proceedings 1 9th ITTC, vol.2.
Khristov, P.. Zilman, G. (1988). Development of Cd
models and their application in ship maneuverability
studies. Proc. 1 5-th International Scientific Seminar
On Ship Hydrodynamics, Varna, v.2, pp.II-16, II-25.
Sobolev, G.V. (1976) Ship maneuvering and au-
tomation of navigation. (Russian). Sudostroenie,
Leningrad.
Ship theory h.andbook (Russian). (1988). Ed. by
Y.I.Voitkunski, Vol. 3, Manoeuverability of conven-
tional ships. Sudostroenie, Leningrad
Sobol, Statnikov. R.B. (1083). Selection of op-
timal parameters in multiobjeclive problems (Rus-
sian). Nauka, Moskow.
Morse, R., Price D. (1961). 'Maneuvering Charac-
teristics of Mariner Type Ship (Compass Island) in
Calm Seas. Sperry Gyroscope Co., Pub. No CL-2233.
Zilman. G., Ter-Zakhariants, A. (1987). Parame-
ter identification of ship dynamics as a problem
of multiobjective Optimuzation. Ingenie.urhochsliule
fiir Seefahrt 14'arnemiinde/Wustrow , Heft 4/87,
Rostock, pp.48-57.
Zilman. G., Ter-Zakhariants, A. (1987). Methods of
constructing the library of mathematical models in
Leningrad Trainig Research Center (Russian).
Trans. of Leningrad Central Maritime In.stitute,
Vodny Transport., Moscow, pp.14-22.

24
 Modeling Realistic Ship Behavior
in Highly Restricted Waterways
L.L. Daggett (U.S. Army Engineer Waterways Experiment Station, USA)

ABSTRACT effects. This is typical of almost all

presentday simulator models.
Recent simulation testing of
existing and proposed improvements to the PROJECT DESCRIPTION
Houston Ship Channel (HSC) in Galveston
Bay has demonstrated the need for The Port of Houston is the third
additional information on the behavior of largest in the United States in terms of
ship maneuvering and passing in very cargo tonnage. The HSC project consists
restricted waterways. Modeling based on of about 65 miles of improved deepdraft
towing tank tests for bank/ship and channels leading from the Gulf of Mexico
ship/ship interactions had to be seriously to the Houston Turning Basin at the head
modified for the ships to respond as the of navigation Fig. 1. The HSC shares the
pilots indicate ships behave in the HSC.
Comparison with scale physical models
indicates that these modifications do
produce ship motions similar to those
measured with the freerunning physical
models. Prototype measurements also
indicate the behavior is being reproduced
with the revised simulator model. This
suggests the need for new and better
methods for determining ship behavior in
highly restricted waterways that can
account for the free motion of the ship.

BACKGROUND

The USArmy Engineer Waterways

Experiment Station (WES) has been using
ship simulation for about 10 years to im-
prove the design of US waterways during
port improvement project studies. This
simulation used a model based upon the
basic ship hydrodynamic model developed by
Tracor Hydronautics (1). To account for
factors such as shallow water, bank ef-
fects, ship interaction, etc., a linear Fig. 1. Houston Ship Channel,
superposition of the forces and moments is Galveston Bay, Texas
used to develop the external effective
forces and moments acting on the ship.
The algorithms for these factors are typi- lower portion of which is shown in the
cally developed from towing tank mea- entrance channels with the Ports of Galve-
surements and limited prototype data. ston and Texas City to Bolivar Roads and
These forces and moments are usually
continues across Galveston Bay to Morgans
developed independently of any other Point and continues up Buffalo Bayou into

25
the City of Houston. The present channel oil tankers currently operating in the
dimensions for the lower 26 miles in the world fleet. Larger ships loaded to a
bay section are 400 ft wide and 40 ft deep greater draft would substantially reduce
at mean low tide (MLW). Tides in the bay commodity shipping costs. Proposed pro-
are predominantly diurnal; the mean tide ject improvements would deepen the chan-
is about 1.6 ft at the entrance, decreas- nels from the present 40 ft to 50 ft and
ing to 1.0 ft or less near Baytown. widen the channels from 400 ft to 600 ft
Currents are normally about 2 knots at over most of the distance of the HSC. The
peak flood and ebb near the entrance, but HSC part of the project will be con-
decrease significantly away from the structed in two phases:
entrance to about 0.5 knot near Baytown.
Ebb tide currents are stronger than flood Phase I provides for deepening to
by about 50 percent. Depths beyond the 45 ft with an intermediate width of
navigation channel limits are quite small, 530 ft.
typically between 5 ft and 10 ft; over
large areas of the bay, average water Phase II would deepen to 50 ft and
depth is about 6 ft. The bay hydrography provide the full width of 600 ft.
has been substantially modified by dredged
material disposal islands on the eastern CONCERN
side of the navigation channel in the
upper part of the bay. Generally, cur- During the planning of tests for the
rents are not a major navigation problem HSC in Galveston Bay, it was anticipated
in the HSC, except in the Bar Channel and that the conditions in this channel might
Bolivar Roads area. Cross currents in the make this approach of modeling ineffec-
bay segment are noticeable near Redfish tive. For example, the limitations of the
Bar and farther up the bay south of the assumption of superposition could be
dredged disposal islands. violated to such a degree that the nonlin-
The navigation channels in the study ear effects and any interaction between
area are operated primarily with two-way the factors could become significant. One
traffic. The restricted channel width of of the most critical and difficult aspects
of the ship simulations in HSC is the ship
400 ft and the typically low overbank
depths of less than 10 ft cause consider- meeting and passing ("The Texas Chicken"
able bank effects on ships maneuvering or, as the Houston pilots prefer to call
away from the channel center line. Pilot it, "The Texas Two Step") maneuver. The
strategy in the HSC has been adapted to narrow channel, the high bank forces,
moments on both ships, and the strong
the existing situation and provides for
ships to travel on the channel center line interaction forces involved are extremely
demanding on any simulation tests
normally. A bow-on meeting and passing
maneuver (popularly called "The Texas conducted.
Chicken") is performed using both bank and As far as could be determined, data
The size of required to confirm the capability to do
ship interaction effects.
ships and environmental circumstances of this type of simulation with the confi-
meeting in the HSC are subject to careful dence needed to provide accurate channel
scrutiny by the local pilots and governed design information are not available.
Recently the Therefore, physical scale model tests were
on a case-by-case basis.
pilots have placed a limitation on the included in the study plans to be per-
formed at WES. These tests were to pro-
meeting of ships with a combined beam
width over 251 ft, or 62 percent of the vide information to confirm or adjust
With approximately 1,000 simulation models for this critical maneu-
channel width.
ver. They involved two 1:100-scale ship
ship movements per month, meeting and
models; one ship was 805 ft long with a
passing occur frequently.
90-ft beam and the second ship was 840 ft
Existing HSC channel dimensions Both ships
limit the use and efficient loading of the
long with a beam of 126 ft.
were drafted to 36 ft in a 40-ft-deep
larger dry bulk cargo vessels and crude

26
channel. The channel was formed in a bow-out moment from the time the vessels
compacted sandbed model with a bottom are abeam to when the sterns approach each
width of 400 ft toe to toe, side slopes of other. This is the result of force and
1 on 3, and an overbank depth of 10 ft. moment measurements as ships are towed
In addition, field measurements of parallel to the model bank line or on
two ships meeting and passing in the bay parallel courses (Fig. 3).
portion of the channel were undertaken.
These measurements would allow the compar-
ison of model test results with a full- 1 2 3
ouxell, EMI! ---,
scale maneuver. 4-7( 1 ( 1

, /---- TRAVEL i
1 c 5",
I
,---' DincTIEN
I

VALIDATION OF THE NAVIGATION STUDY

O MOMENT SMALL BOW-OUT MOMENT LARGE BOW-N MOMENT
0 LATERAL FORCE SMALL RE,LSION FORCE NEAR 0 LATERAL FORCE

One of the most important milestones

in the simulation process is the valida- 4 5 6
tion exercise for existing conditions.
cc___:__J r(
- 1 --( 1

The purpose of validation is to use local ,

I )) I
1

)-) )-
pilot expertise (i.e.,
E

knowledge and
LARGE BOW-OUT MOMENT LARGE BOW-N MOLENT 0 MOMENT
experience) to ensure that the simulation LARGE ATTRACT/CN FORCE SIMAU_ REPULSION FORCE 0 LATERAL FORCE
is as realistic as possible. While con-
Fig. 2. Modeling of ships meeting and
ducting validation tests the pilots pay
close attention to ship handling, external passing based on towing tank tests
force effects, and objects in the visual
scene and make comments and recommenda-
tions for improvement. Normally one or
two pilots from the study area come to the
simulator for validation; however, for the 200
bay segment of the HSC study, no less than ECPERLIBMTAL
1141 PFEDIC11ON
eight pilots were required for checkout 150 Ili, SSA. MTH PROP B.FECTS
because of the complicated nature of the
hydrodynamic phenomena being modeled. 2 100

For the HSC navigation study, the E

'1

focus of validation was the modeling of

§
the hydrodynamic interaction between ships o
during meeting and passing. Ship/ship
interaction forces and moments on the 3
<
-50
,
flI
simulator are calculated with a numerical
model developed by Tracor Hydronautics, -100

Inc. (2). This model generates vessel 2 -150

interaction forces and moments for the WATER DEFTH - 115 T
case of two ships meeting and passing on SEF'ARAT1ON - 15 8
-200
constant parallel (but opposite) courses. -1200 600 /200
The calculation is based on inviscid Se.ARATION (FT)

potential flow and includes semiempirical Fig. 3. Model measurements and simu-
corrections for additional factors that lation model
affect interaction, for
computations,
example, hull (Reference 2)
appendages and shallow and confined
channels.
Figure 2 shows a typical meet-
ing/passing operation as was modeled by In contrast, Fig. 4 shows a typical
the simulator prior to validation exercis- meeting/passing operation as experienced
es. The most notable characteristic is and described by the Houston pilots. The
the strong bowin moment after the bows most significant difference from Fig. 2 is
pass each other, followed by a strong the lack of the strong bowout moment when

27
 move to the side prior to meeting,
2 3 after which they return to the chan-
rime
nel center line. It was apparent
through physical model tests that the
c===:::)-)
orientation of ships just before
APPROX. 3/4 UAL APART SW SWINGS BOW-CAJT BAIX FORCES BOW N
BOW-OUT RUDDER RUDOER MOVED TO MIDS.'S ROW-OUT RUDDER use, meeting had an important impact on
the sequence of events in the
4 5 6 meeting/passing maneuver.

c.<=3 b. In the simulation, only one of the

ships was controlled by a human pilot
HARD BOW-ti SWNG STRAIGHTEN VESSEL
SNIPS STEADY MOLENTARLY
BOW-NI SWI43 NCAEASES BOW-OUT RUDD69 USED USE RUDDER TO CONTROL
while the other was controlled by a
linefollowing numerical autopilot.
Fig. 4. Modeling of ships meeting and The behavior of the autopiloted ship
passing based on pilot experience had significant impact on the stra-
tegy of the human pilot in the simu-
the ships are abeam (frame 4). As indi-
lation. The actual meeting and pass-
cated in Fig. 4, according to the pilots, ing operation is very much a joint
once the bows have passed each other effort between two human pilots,
during the meeting, there is a constant particularly in such a confined
tendency for the ship to rotate away from waterway.
the closest (in this case starboard) bank
and toward the other ship. A large amount C. It was evident that the simulation
of countering rudder is usually required model underestimated the effect of
to regain a straight course in the center the extremely confined channel on the
of the channel. This countering rudder meeting operation. The most signifi-
begins to be required when the ship ap- cant aspect of this appears to be the
proaches the bank from the intitial move role that the very close submerged
from the center line or from the initial banks play in the behavior of the
bowout movement and is never removed. ships. Pilot experience indicates
The countering rudder is applied to con- that the bowin moment imparted to
trol the naturally occurring rotation and the ship by the closest bank seems to
movement of the ship during the remaining cancel the bowout moment (Fig. 2,
passing action. frame 4) measured in the restrained
It should be noted that in Fig. 4 model tests, conducted in less con-
the pilots were referring directly to ship fined water, on which the simulator
movement rather than moment and forces as model is based.
referred to in Fig. 2. However, the two
descriptions are essentially synonymous d. Another aspect of a highly restricted
since ship movement closely paralleled the waterway is the impact caused by low
moment during the simulated operation. It underkeel clearance. The simulator
became evident from validation testing model limits ship underkeel clearance
that the existing simulator model of the to no less than 5 percent of the
meeting/passing operation needed modifica- water depth. In short, the simulator
tion to achieve an acceptable level of model maintains water under the keel
realism. The following differences were at all times. This is standard prac-
noted between the simulator modeling tice in simulation modeling at pres-
method and the actual meeting operation: ent because of the lack of reliable
data on the maneuvering of vessels
a. The simulator assumed the two ships whose hulls are very close to the
were on constant parallel courses in bottom. In the HSC, ships with less
their own halfchannel lanes. In the than 5 percent of channel depth under
actual operation the two ships travel their keel frequently meet and pass
down the center of the channel and in the channel.

28
 All of these factors indicate that the Numerous tests were conducted with
meeting/passing operation in the HSC is a different setups and strategies using WES
very complicated, dynamic process that personnel and Houston pilots to obtain a
does not lend itself easily to simulation. qualitative understanding of the hydrody-
Due to a combination of time and cost namic processes of the meeting and passing
restraints and lack of reliable data, maneuvers. Generally, the professional
model improvements for most of the factors pilots who conducted these tests were
listed in the previous paragraph were not pleased with the level of accuracy demon-
possible prior to pilot testing. Model strated in the scale model maneuvers. The
adjustment focused on modification of the modifications performed on the simulator
ship/ship interaction movement until a numerical model were based partly on
sequence of events similar to that shown knowledge gained through observation of
in Fig. 4 was produced during meeting and these scale model tests, in addition to
passing on the simulator. This required the pilots' description of the events and
continuing the bowin moment (Fig. 2, control measures required during meeting
frame 3) throughout the entire maneuver and passing situations.
and increasing its magnitude, in effect, Because a combination of hydrodynamic
truncating any bowout moment calculated and human factors is involved in meeting
by the original computer model. This and passing maneuvers, a basis of compari-
modification forced the vessel to rotate son between simulator and scale model
toward the other ship throughout the results was difficult to choose. For the
meeting to simulate the ship behavior that scale model, accurate or exact replication
the pilots expected. Also, some modifica- is a problem, while on the simulator the
tions were made in the way the model tests can be replicated but only with the
calculates bank effects by assigning use of a numerical autopilot. It was
higher weights to the forecs produced by decided that the preferred method of com-
the closer bank and reducing those from parison was to remove, to the extent
the opposite bank. These modifications possible, the human influence in the tests
were carried out with the help of a con- and compare only the hydrodynamic behavior
sultant (3) and additional validation of the ships. For the comparison tests
pilots. both ships were placed in their respective
halfchannel lanes on parallel (but oppo-
PHYSICAL SIMULATION MODEL COMPARISON site) heading. The autopilot kept the
ships on a straight course until just
noted previously, part of the
As prior to the bows meeting, at which time
navigation study involved a 1:100scale the autopilot rudder was returned to
model of a generic section of the bay midships and the tests were completed
segment of the existing HSC constructed at without any further control. In the
WES. Two radiocontrolled, freerunning physical model tests the setup was the
models, each independently controlled by same except the ship had human control
a technician, were operated in the chan- (via telemetry) until just prior to the
nel. The position and orientation of meeting when, again, the rudder was moved
these models were measured using a video to midships and the test was allowed to
tracking system developed at WES. From continue without additional human input.
these data, ship speed, heading, and posi- For these validation tests, ship sizes on
tion in the channel can be determined and the simulator were matched as closely as
the motions of the ship during meeting and possible to those in the physical model.
passing determined. These factors could Table 1 lists these ship dimensions.
also be compared against the reports of
the pilots and the behavior of the
simulation model.

29
Table 1. Ship Dimensions similar conditions. It can be seen that
the general behavior of the models was
Ship LOA (ft) Beam (ft) Draft (ft) similar. The simulated model experienced
about the same rate of heading change
Physical Model during the period that the ships actually
passed each other. The physical model
1 805 90 36 began to experience stronger bank effects
2 840 126 36 as it moved closer to the bank line and
turned away from the bank, while the
Simulation Model simulator did not. This may also indicate
physical contact with the bank slope.
1 810 106 36
2 840 138 36 FULLSCALE MEASUREMENTS

Note: LOA Length overall A differential Global Positioning

System (GPS) survey was conducted in which
5 compares physical model tests
Fig. two receivers were positioned at the bow
with simulator results for 6.5 knots. and stern of two ships, one at the en
Similar results were obtained for 9 knots. trance to the bay and one at dock in
The shaded regions indicate variability Houston (4). The ships passed near Red-
ranges for physical model ship position fish Island as shown by the tracklines
(bow and stern), speed, and heading. (Fig. 6). Rudder settings and engine
These plots result from five separate revolutions per minute (rpm's) were video
runs. Human control during the first half
taped during the transit and transcribed.
of each of the physical model tests re- While it appears that one of the ships
sulted in much of the variability. The went out of the channel, it did not ground
single lines show simulator results for

Fig. 6. Ships meeting and passing as

measured by GPS in Houston Ship

Fig. 5. Model comparison, physical and Channel

simulator

30
because its draft was shallow enough to has been made using the same techniques of
clear the bank at this point as confirmed standard maneuvers with a 40ft lobster
by the crosssectional view (Fig. 7). boat (Fig. 9).
Similar results were obtained during the
simulation tests near the same location
(Fig. 8); however, the large ship was on
the opposite side of the channel. As an
example of the level of accuracy that can
be obtained, a recent set of measurements

ECGE OF CHANNEL EDGE CF CHANEL

400 FT

WR FT

- 136.5 FT --- - 84 FT
AgAt FT 22F1
NEW VEAL ATLANDC
28,5 FT

42 FT

E,C1INO 8070.4
LOOKING NORTH

Fig. 7. Ship positions in channel at the

time of passing

Fig. 9. GPS tracking of lobster boat

during steady turn maneuver

CONCLUSIONS

Improvements are needed in the de-

scription of the forces and moments on
ships passing in a highly restricted
waterway. The determination of these
forces must account for the changing
orientation and position of the ships in
the channel and relative to one another.
This will require innovative approaches or
possibly a numerical approach.
The use of differential GPS will
provide an excellent means for more accu-
rately measuring ship tracks in realworld
situations to provide data for verifying
simulation models. The increasing use of
intergrated electronic bridge systems with
logging capabilities will further enhance
Fig. 8. Simulated ship passing in the possibilities for more accurate and
Houston Ship Channel near Redfish Is- scientifically sound verifications.
land Careful planning and a lot of luck are
required for completely successful

31
results; the ships measured in the HSC did Ankudinov, V. (1988)."Mathematical
not have the desired draft and were and computer models for predicting ship-
smaller than desired. Therefore, direct ship interaction forces for use on WES
comparisons for validation are not ship maneuvering simulator." Technical
possible. Report 87018.0124, prepared for U.S. Army
Engineer Waterways Experiment Station,
ACKNOWLEDGMENTS Vicksburg, MS, by Tracor Hydronautics,
Inc., Laurel, MD.
The tests described and the resulting
data presented herein, unless otherwise Ankudinov, V. (1991). "Improvements
noted, were obtained from research spon- to ship handling model for meeting and
sored by the US Army Engineer District, passing situations in the Houston Ship
Galveston, by the US Army Engineer Water- Channel (HSC) navigation study." Report
ways Experiment Station, Vicksburg, MS. No. 510-91-007, Document No. 5101007R.D12,
Permission was granted by the Chief of prepared for U.S. Army Engineer Waterways
Engineers to Publish this information. Experiment Station, Vicksburg, MS, by
British Maritime Technology, Inc.,

REFERENCES Columbia, MD.

1. Goodman, A. et al (1976). "Experi- Chance, J. E. and Associates, Inc.

mental Techniques and Methods of Analysis (1990). "GPS survey report, Houston Ship
for Surface Ship Channel, EN-PS Project C-91-012."
Used at Hydronautics
Maneuvering Predictions," Hydronautics Galveston, TX.
Incorporated Technical Report 7600-1.
Laurel, MD.

32
 Interactions Between the Hull and Rudder
During Ship Maneuvering
P.J. Kopp (Carderock Division, Naval Surface Warfare Center, USA),
N. Fuller (U.S. Coast Guard Headquarters, USA)
R. Sedat (U.S. Coast Guard Research Center, USA)

ABSTRACT BACKGROUND
Maneuvering experiments, sponsored The model test described herein
by the U.S. Coast Guard, were performed are part of an overall plan to develop
in the rotating arm and linear towing the capability to accurately predict the
tank facilities at
the Carderock directional stability and
Division of the Naval Surface Warfare maneuverability characteristics of ships
Center. The experiments investigated the during all stages of design. This
effect of the hull on the inflow to the information is not only useful in
rudder, and the effect of the rudder on determining whether or not the ship will
the forces on the hull. A 1/24th scale satisfy the requirements of the
model of a MARINER Class ship was used. customer, but also will indicate if the
The average inflow angle to the vessel satisfies the proposed
rudder was determined, at various yaw International Maritime Organization
rates and drift angles, by an unbalanced (IMO) Maneuverability Standards.
flat plate rudder that was free to seek Several years ago, Dr. Volf
a neutral angle. In separate tests, the Asinovsky began publishing papers
rudder inflow velocity and angle were dealing with maneuverability measures
determined by pitot tubes mounted at and indices [1]. He was also an avid
four spanwise locations on the leading proponent of the modular approach to
edge of a rectangular rudder. In other deriving the equations of motion. The
tests, hull forces and moments were modular approach describes the process
measured with the rudder stock passing by which the various forces and moments
through the hull but not attached, and on the hull and appendages are
compared to similar data taken without calculated separately as a kinetics and
the rudder. These tests were used to kinematics physics problem. These
quantify the hull force differential due nonlinear forces and moments are then
to the presence of the rudder. combined to provide a time domain
representation of the ship trajectory.
NOMENCLATURE See the appendix for a discussion of the
modular maneuvering model.
PP length between A quasi-static simplification of
perpendiculars these equations results in the ability
L, to calculate the Diagram of Steering.
length of the waterline
This representation of the
Cp prismatic coefficient maneuverability and stability
Cb block coefficient characteristics of ships, while popular
in the Soviet Union and utilized in
scale ratio Europe and the Far East, has been
EAR expanded area ratio largely ignored in this country. Dr.
Asinovsky, using an empirical procedure
MWR mean width ratio from Fedyayevskiy and Sobolev [2], was
BTF blade thickness fraction instrumental in developing the computer
code to analytically determine the
cg center of gravity Diagram of Steering [3].
Vcg Velocity at the cg Shortly after Dr. Asinovsky was
water density employed by the Coast Guard, he proposed
various criteria which could be used to
measure the stability and
The opinions expressed herein are those of the authors and do not represent the
views of the Department of the Navy or the United States Coast Guard.

33
 characteristics of a the same propeller used in the
maneuverability
vessel [4]. He also formulated an previously mentioned powering tests
analytical and experimental program to (DTRC Prop number 3221). Characteristics
develop tools to calculate indices of of the propeller used are listed in
these criteria which would enable Table 2.
designers and regulatory bodies to A body plan of the model is shown
determine the acceptability of a given in Figure 1. Figure 2 shows a photograph
design. Dr. Asinovsky enlisted the aid of the model attached to the rotating
of the authors to accomplish these arm subcarriage.
objectives. Mr. Sedat was to develop the
time variant nonlinear maneuverability Length, L, (ft/m) 21.84/6.66
equations and subsequent computer code. Length, L, (ft/m) 21.65/6.60
Mr. Fuller was to develop the
quasi-static nonlinear equations and Beam (ft/m) 3.14/0.95
subsequent computer code for the Diagram Draft (ft/m) 1.12/0.34
of Steering. Mr. Kopp was to perform a
series of model and full scale tests to Displacement 1.28/1.30
determine the force and moment (LT/Tonnes)
relationships to the ship trajectory, in Prismatic
addition to the hull/appendage Coefficient, Cp 0.620
interaction effects. Block Coefficient,
The model test results described C, 0.608
herein have been incorporated into a
completely rewritten nonlinear Diagram Scale Ratio, k 24.18
of Steering computer program which is
solved iteratively through direct Tab. 1. Model Characteristics
numerical integration. The full scale
trials required for correlation have Diameter (ft/m) 0.91/0.28
been performed, however the data has yet Pitch/Diam. Ratio 1.05
to be analyzed. The equations of motion
based on the modular approach have been EAR 0.56
developed and funds are being programmed MWR 0.28
for the encoding of these equations into 0.05
A fourteen model BTF
a computer code.
Asinovsky Maneuvering Series for Tab. 2. Model Propeller Characteristics
investigating the effects of various
hull characteristics on maneuverability
have been

Utilizing
developed, built, and will
probably be tested in early 1993.
The program laid down by Dr.
Asinovsky prior to his death in 1990 is
well on the way to being accomplished.
the data from these and
possibly future tests, we can expect to
Nit It
be able to predict ship stability and
maneuverability characteristics during
all stages of design.

MODEL DESCRIPTION
Model 4414 (USNS HESS, T-AG(S) 38)
11!U
Fig. 1. Body Plan
was selected from the existing DTRC
collection. This model was selected Prior to the start of the test,
because it represents the Mariner class the model was modified to restore the
of commercial vessels as originally hull to the original stock lines.
built [5]. The loading conditions of the Platforms for mounting the block gauges
test were determined by a previous were also installed. In addition, the
series of powering tests. The selection semi-balanced horn rudder and rudder
of an existing model also eliminated the post bearings were removed, and the
construction cost from the test program. rudder stock hole enlarged from one to
Characteristics of the model as tested three inches in diameter. The rudder
are listed in Table 1. post hole was then fitted with a
The propeller selected for use was removeable sleeve that was subsequently
a four bladed, right hand prop and was

34
fitted with bearings that could be used pressure tubes freely hung from the
with a standard rudder post. trailing edge of the rudder and were led
up to the subcarriage. The rudder stock
passed up through the three inch
diameter rudder stock hole and was
attached only to a servo motor hung from
the subcarriage. Therefore, no rudder
forces would pass to the hull and be
measured by the block gauges. Figure 3
shows a photograph of the pitot tube
rudder [7].

Fig. 2. Photograph of Model Attached to

the Rotating Arm Subcarriage

TEST SETUP AND PROCEDURE

Throughout the test, the model was

fully constrained in surge, sway, heave,
pitch, yaw, and roll. Horizontal plane
hull forces were measured using block Fig. 3. Photograph of the Pitot Tube
gauges mounted equidistant from Rudder
midships. Non-dimensional sway forces
and yaw moments (referenced to the model
cg) were computed from averages of the Neutral Rudder Angles
block gauge time histories. Tares and With the flat plate rudder
centripetal force effects were also attached to the model, several series of
subtracted from the measured forces. All runs were made using the rotating arm
force and moment data were facility and one of the linear (straight
non-dimensionalized by 1/2pV21,,12 and line) towing tanks at the same model
1/2pveg2L13 respectively. speed. In one series, the propeller was
As part of the post test analysis, removed from the model. In the other
a general uncertainty analysis was series, the propeller was operated near
performed for the force and moment data the self propulsion point as determined
collected [6]. This analysis provided by the powering tests.
95% and 99% confidence intervals for The model drift angle was varied
each block gauge. from plus to minus 30 degrees. The drift
angle at the rudder was determined from
FLOW STRAIGHTENING EFFECTS the model geometry after the tests were
completed. Hull forces were not measured
In order to investigate the flow during this series of tests.
straightening effects of the hull on the The mean position of the rudder,
rudder, two separate rudder free to rotate in the flow field, is the
configurations were used on the model. angle at which it experiences zero mean
The first rudder was a rectangular flat lift (the neutral angle). The difference
plate rudder mounted to a stock on the between the neutral angle and the
leading edge. The rudder geometric drift angle at the rudder
stock was
indicates the amount of flow
allowed to rotate freely in the bearings
of the removable sleeve. A sensor was straightening. Figures 4 and 5
attached to the top of the stock to illustrate the flow straightening
measure the rotational position of the results at five different
rudder. This rudder was non-dimensional yaw rates with and
used to
determine the neutral rudder angle. without the propellers operating.
The second rudder was also of
rectangular planform but was a constant
25% thickness airfoil section. Four
five-hole pitot tubes were extended from
the leading edge of the rudder. The

35
 20

o
10 ...

o
o 0.44

-
0 0.32

-20 -

_
d .02
W iii.(4,

0
0.20

0.00 vw
I Til
0 20 40 NOTES:
-40 -20
Rudder angles ere mint,* to ship centerline.
Geometric Drif) Angle at the Rudder (dog) Pitot tube Inflow angles ere relative to rudder centerline.

Fig. 4. Flow Straightening Effect for Fig. 6. Pitot Tube Inflow Angle Sign
Various Non-Dimensional Yaw Rates, No Conventions
Props
The geometry of the flow and
rudder angles made interpretation of the
20
pitot tube data somewhat difficult.
10 O 0.63
After much discussion, it was decided
o that the graphical presentation the
O 0.44 pitot tube data would be made with two
o

_
u ,5>
0a 0.32 related plots at each drift angle/arm
-10
radius condition. The first plot would

'
0.20
-20
e_
o give the inflow velocity magnitude
0.00 expressed as a ratio to the ship speed
-30 _ versus the rudder angle setting. A
-10
sample is shown in Figure 8. The second
-40 -20 0 20 40
plot gives the pitot tube measured flow
Geometric Drift Angie at the Rudder (dog) angle, also versus the rudder angle
setting. A sample plot is shown in
Figure 9. These two sample plots
Fig. 5. Flow Straightening Effect for 8 and 9), represent the case
Various Non-Dimensional Yaw Rates, Props (figures
without propellers at a drift angle of
Operating 30 degrees and zero non-dimensional yaw
It is worth noting that the flow rate (straight line).
For comparison to the neutral
straightening is insensitive to the
rudder angle data, a line with minus
turning radius (non-dimensional yaw
rate). The effect of the propeller race unity slope was placed on the second
plot. The intercept of the line (at zero
is also relatively uniform over drift to the ship
rudder angle relative
angle and essentially moves the zero rpm
down slightly and centerline) is the neutral rudder angle
curve (Fig. 5)
at the same conditions.
somewhat flattens the curve.

Pitot Tube Flow Angles

The pitot tube fitted rudder MolTubo3
measured flow angles with respect to the
rudder centerline, in the horizontal
plane. The rudder was positioned at MorNbeet
various angles relative to the ship
CAnter Ins
centerline. The flow angle geometry is NolTube5
shown in Figure 6. A schematic drawing
of the pitot tube rudder, showing the NotTubeEt
pitot tube locations and numbering
scheme, is shown in Figure 7.
Rucholermaonits25%th.ress

Fig. 7. Schematic of Pitot Tube Rudder

36
 tM
1

0.98 Tube 3
Tube 4
0.96
Tube 5
z. 0.94 Tube 6
0.92
! 0.9
0.8.9
o
0.86
0.84
0.82

-30 .20 -10 0 10 20 30

Rudder Angle Relative to Ship (deg)

Fig. 8. Sample Measured Inflow Fig. 10. Difference in Non-Dimensional

Velocities Sway Force on Hull, With Rudder, No
Props
15
10

5
0
-5 .,
2 ,
,

-
°
Tube 3
Tube 4
Tube 5
,
a Tube 6 ,
-3° ,
-35 - flag Rudder
-40
-30 -20 -10 0 10 20 30
Rudder Angle Relative to Ship (deg)

Fig. 9. Sample Measured Inflow Angles Fig. 11. Difference in Non-Dimensional

Sway Force on Hull, No Rudder, Props
The pitot tube data indicates the Operating
general trend of reduced flow velocity
and increase flow straightening near the
top of the rudder. This is due to the
combined effects of the hull boundry
layer and hull blockage. However, the
flow field around the stern region is
extremely complicated and the results
show cases where these trends are
reversed.

DIFFERENTIAL HULL FORCES

Figures 10 through 15 show the
difference in non-dimensional sway
forces and yaw moments as compared to
the bare hull data. These plots were Fig. 12.
developed from Difference in Non-Dimensional
simple regression Sway Force on Hull, With Rudder, Props
equations derived for each set of data Operating
[7, 8]. The difference from the bare
hull regression equation was used to
plot contours. The zero difference
contour line is indicated on each plot.
Contour lines represent non-dimensional
force/moment increments of 0.5X10-3, with
the darker bands representing negative
differences and the lighter bands
representing positive differences.

37
 regression equations). This result was
expected. As the drift angle increases
at zero non-dimensional yaw rate, the
relative sway force decreases and the
relative yaw moment increases. This
indicates that the center of lateral
force on the hull has moved forward far
enough to offset the decrease in side
force. Conversely, at zero drift angle,
as the turn radius becomes tighter, the
relative side force increases and the
relative yaw moment decreases.
With the propeller operating, but
no rudder, the behavior of the relative
force and moment is quite different. The
Fig. 13. Difference in Non-Dimensional lines approach
zero contour
Yaw Moment on Hull, With Rudder, No approximately 20 dearees drift for zero
Props non-dimensional yaw rate. This is due to
the asymmetric flow caused by the
rotation of the propeller. It is
interesting though, that the relative
sway force contours are nearly
perpendicular to the no props/no rudder
case.
When the propeller is operating
and the rudder is used, the results are
a composite form of the two other
conditions. The yaw moment, in this
case, at zero non-dimensional yaw rate,
has moved towards the zero drift angle
point. This shift is due solely to the
upstream hydrodynamic effect of the
rudder on the propeller/hull system.
Fig. 14. Difference in Non-Dimensional
Yaw Moment on Hull, No Rudder, Props CONCLUSIONS
Operating
These experiments provide certain
elements necessary for the development
of a truly "modular" maneuvering model,
le. one where the forces are computed
separately on the hull, propeller, and
rudder. This computation includes the
interference effects, ie. the forces on
one part due to the presence of the
other parts.
For example, the flat plate rudder
results show how the average inflow
direction to the rudder is different
from the geometric drift angle due to
the presence of the hull. The pitot tube
results give additional insight on the
Fig. 15. Difference in Non-Dimensional spanwise distribution of the angle, and
Yaw Moment on Hull, With Rudder, Props the associated inflow velocities.
The differential hull force
Operating
experiments demonstrated that the forces
The cases shown with the rudder and moments acting on the hull are
(pitot tube) are for zero degrees rudder affected by the deflected rudder, just
angle. as the pressures on the forward part of
the no propeller, with
Consider a flapped wing are affected by the
rudder case in Figures 10 and 13. At orientation of the flap.
and several
zero non-dimensional yaw rate and zero These interactions,
drift angle, the difference in force and others, must be known before the full
moment tends to zero (not exactly for potential of "modular" modeling can be
Figure 13, but these are differences of realized.

38
REFERENCES APPENDIX - THE MODULAR MANEUVERING MODEL
Asinovsky, Volf, "On Maneuverability The general unsteady equations of
Criterion for Determining Ship motion are written,
Response to Rudder Angle Change", idu du,
presented before the Chesapeake - 111 t m22vr-km26r2±Fx
m(%.-n4dt dt -r
Section of Society of Naval f dv dv ,
Architects and Marine Engineers, May 134--ur)=-m22 1-m11 ur-i-m26 --rr Y
17, 1983.
Fedyayevskiy, K.K. and G.V. Sobolev, L=-m66 7-m26(2-r; -f-ur)-(m22 -m11 )uv+Mz 1.
"Control and Stability in Ship
Design", State Union Shipbuilding where the first terms on the left hand
Industry Publishing House, side represent mass times acceleration
Leningrad, 1963, English translation and the second terms result from the
by U.S. Department of Commerce, moving coordinate system. The right hand
1964. side contains the hydrodynamic force
Asinovsky, Volf, Kiang-Ning Huang, terms, including added mass and damping
and Mark C. Oakes, "Ship effects. The F, F, and 1.4 terms
Maneuverability Analysis Using represent the viscous forces resulting
Steady State Techniques", presented from frictional drag, vortex shedding,
before the Chesapeake Section of the and propeller lift and drag.
Society of Naval Architects and The "standard" treatment for the
Marine Engineers, June 14, 1988. viscous terms is to expand them in a
Asinovsky, Volf, "Review and Taylor series. In doing so, the force
Analysis of Ship Maneuverability and moment contributions from the bare
Criteria", Naval Engineers Journal, hull, rudder, propeller, etc. are lumped
1989. together in the series expansion
Russo, Vito L. and E. Kemper coefficients.
Sullivan, "Design of the In the modular approach, each of
Mariner-Type Ship", Transactions of the viscous terms is considered to be a
the Society of Naval Architects and sum of the contributions from the hull,
Marine Engineers, 1953. appendages, propellers, etc. as in
Coleman, Hugh W. and W. Glen Steele,
"Experimentation and Uncertainty Fx =F x-rudder+F x-propeller+
Analysis for Engineers", John Wiley 2.
& Sons, Inc., 1989. Each of these terms include the force
Kopp, Paul J., Richard C. Bishop, due to that component alone and the
and Lewis E. Motter, "Experimental interactions resulting from the other
Study of the Rudder Flow Field And components.
Its Effect on Hull Forces", In order to determine each of the
DTRC/SHD-1341-01, August 1990. "force modules" in equation 2, either
Kopp, Paul J., "Captive Model analytical, empirical, or experimental
Rotating Arm Test on the Bare Hull methods are used. SNAME Panel H-10 (Ship
T-AG(S) 38 Represented by Model Controllability), as well as the USCG
4414", DTRC/SHD-1341-02, February and CDNSWC, is actively engaged in
1991. developing analytical and empirical
methods.
ACKNOWLEDGMENTS
Experiments for determining all of
the interaction effects requires a
The authors would like to thank sufficiently large test matrix
Mr. Richard Bishop for his efforts in and
several separate tests. First, bare hull
performing the experiments reported, and measurements are made
Mr. Richard Bell for collecting over a
the sufficiently large range of drift angles
pitot tube data. The support and and turning radii (also roll angles and
suggestions from Mr. Lewis Motter are
also greatly appreciated. speeds). Free stream characteristics of
the rudder and propeller must also be
measured. The hull/propeller combination
is then tested thru the same matrix with
forces measured on both hull and
propeller. The difference between the
bare hull and free stream propeller
forces and those measured with the
combination are the hull-prop and
prop-hull interaction forces

39
respectively. The same test method is
used on the hull/rudder combination to
measure the hull-rudder and rudder-hull
interaction forces. The final test is
with the hull/propeller/rudder
combination to yield the three
hull-prop-rudder interaction forces.
The experimental program discussed
in this paper, including the Asinovsky
Model Series and full scale trials on
the Mariner vessel OBSERVATION ISLAND,
provide necessary insight into the
interaction effects and experimental
validation of the numerical procedures
being developed.

40
 On the Full Scale Maneuvering Characteristics
of the SWATH Fishing Vessel Ali
R.C. McGregor (University of Glasgow, United Kingdom)
A.F. Miller (Vickers Shipbuilding & Engineering Ltd., United Kingdom)

ABSTRACT expected to change as the number of

SWATH vessels in service worldwide,
This paper describes a series of under trials and under construction
experiments relating to the manoeuvring increases in the coming years. Despite
of the 22 tonne SWATH fishing vessel its paucity, study of the published
ALI. The results are related to the literature does allows several
output of a manoeuvring prediction interesting conclusions to be drawn and
program resulting from the adaptation thee have been summarised in refs[5,6].
of monohull manoeuvring techniques to
SWATH ships. The trials reported in this paper
are part of an extensive series of
INTRODUCT ION tests into the resistance, propulsive
characteristics, seakeeping and
SWATH ships are a modified form of manoeuvring of the SWATH fishing vessel
catamaran in which the underwater form ALI. The main characteristics of ALI
has been adapted to move the supporting are set out in refs[7,8] and summarised
buoyancy well below the sea surface of here in table 1.
to mitigate the effects of wave action.
A typical SWATH vessel consists of two Table 1
totally submerged slender torpedo-like Length of Hull (m) 10.5
hulls upon which an above water cross- Hull Diameter (m) 1.0
structure, or box, is supported by Length of Strut (m) 8.9
means of a pair of long narrow surface Max Strut Thickness (m) 0.6
piercing struts. The resulting vessels Draught (m) 1.6
have demonstrated dramatically better Centreline Separation (m) 4.0
seakeeping behaviour relative to Waterplane Area Coefficient 0.92
conventional monohulls and catamarans Hull Prismatic Coefficient 0.82
in both model and full scale trials. Displacement (tonnes) 20 0
Design Speed (knots) 8.0
Much effort has been gone into Installed Power (kW) 104.0
improving understanding of the
hydrodynamic forces and moments acting SWATH Mano.uvrina Factors
upon these hullforms. Since the
primary justification for the very In order that any predictions on
existence of SWATH ships is the the manoeuvring capability of a ship
excellent seakeeping behaviour inherent may be made, the acceleration and
in the concept, most of this effort has velocity derivatives must be known or
been aimed at predicting and evaluating approximated. Direct evaluation of
ship motions in waves. these derivatives is not at present
considered a practical proposition
In comparison, relatively little especially at the design stage, and
effort has been devoted to the study since copious model testing is not
and prediction of SWATH ship motions in always feasible in the early stages of
the horizontal plane, and in particular development. It was decided to examine
their manoeuvring characteristics.
the possibility of modifying existing
Most of available literature originates
semi-empirical techniques for finding
from the DTRC and largely concerns monohull manoeuvring characteristics
model tests performed in rotating arm (Clarke et al [9]) to provide guidance
facilities. for the naval architect engaged in the
design of rudders and steering gear for
There is very little full scale SWATH vessels.
operational information [1-4] available
at present although this may be

41
 SWATH vessels are as a tactical diameter of the order of 20%
consequence of their geometry using this method. Similarly
inherently directionally stable. The deployment of a retractable "turning
centroid of the projected area of the foil" has been found to yield benefits.
struts is generally aft of the centre For a rudder which is surface piercing,
of gravity of the vessel, and it trim has an important influence on the
therefore takes a substantial side performance of manoeuvres.
force to initiate a turn when the
vessel is travelling at speed. This Turning performance is dependent
implies that larger rudders and heavier on speed. At 13 knots the tactical
steering gear must be employed relative diameter of the SSC SEAGULL was found
to those on comparable monohulls. to be half that at 23 knots [2]. This
However, for missions requiring a is a result of the flow patterns along
steady course to be kept, directional the hull changing considerably as
stability is an advantage, especially Froude Number is varied. Tank tests
in oblique seas. It is a particular [11] with a long strut design reveal
advantageous if towing a sonar array that the manoeuvring derivatives are
since the signals from the array will less speed dependent with that
be less confused and thus easier to geometry.
interpret reliably. Because of this
inherent directional stability and the The turning performance of SWATH
problem of rudder location an SWATH ships can be made comparable to that of
vessels, careful attention must be paid equivalent monohulls provided
to the design of the steering mechanism appreciable care is exercised in the
in order to ensure that the ship design of the control surfaces and
possesses adequate turning performance their positioning. The often quoted
in terms of turning and ability to ratio of tactical diameter/ship length
initiate a turn. is misleading for comparison purposes
since the length of an "equivalent"
The SWATH geometry presents a SWATH is about 40% less than its
number of unique problems when it comes monohull counterpart (MacGregor [12]).
to siting rudders and steering gear.
This has led to the development of Rudder Design for MV
several innovative configurations of
control surfaces. A description of ALI has strut rudders in which a
these can be found in refs[5,6]. movable section is incorporated into
the trailing edge of the struts
The transverse separation of controlled by steering gear in the
propellers leads to excellent low speed cross deck.
manoeuvrability on SWATHs through the
use of differential thrust. For This is the simplest and cheapest
example SSP KAIMALINO turns within her solution which also adds least drag to
own length at very low speeds (Fein the vessel. However, the configuration
[10]). Vessels equipped with bow is ineffective at both low and high
thrusters should be able to turn speeds, since at low speeds there is no
effectively on the spot. Using bow benefit gained from the locally
thrusters linked to differential thrust increased flow velocity due to
from controllable pitch propellers propellers, and at high Froude No's the
under active automatic control, the waterline dips towards the stern of the
possibility exists for providing vessel, reducing wetted-effective
exceptional stationkeeping or docking rudder area. Consequently, greatly
performance. Controllable pitch increased rudder areas are required to
propellers and/or electric transmission ensure adequate turning performance.
is to be recommended for applications In addition, it is impossible to
requiring good low speed manoeuvring. balance strut rudders, and so the
Unacceptable strains, caused by the steering gear must be large, powerful,
frequent forward/reverse shifts, would heavy and relatively costly.
arise if fixed pitch propellers and a
conventional drivetrain were used.
MANOEUVRING THEORY
At higher speeds, it is possible
to improve turning performance by The SWATH manoeuvring prediction
employing canards to bank the vessel program, SWATHMAN, is based on a
into the turn. The asymmetric drag monohull manoeuvring theory which is
produced, sharpens the turn. SSP described in Clarke et al [9] and
KAIMALINO has reported reductions in Miller [5]. This theory is founded

42
 upon multi-variate linear regression errors when determining side force from
and semi-empirical expressions higher aspect ratio surface piercing
developed from analysis of experimental strut and spade rudders. This effect
data. Because of the symmetrical is counteracted by the incorporation of
nature of SWATH geometry however it is a mirror imaging factor into the
reasonable to expect that it will calculation. This is primarily
readily adapt to the study of SWATHs. designed to model the increase in
effective aspect ratio which occurs as
Lstimation of Ship Derivatives a result of the lower end of the rudder
effectively butting on to the hull.
Several attempts have been made Rudder fences will have a similar
to derive empirical expressions effect in the case of surface piercing
relating the velocity derivatives to control surfaces.
ship geometry. These formulae were
derived after analysis of experimental For the above cases the standard
results obtained on planar motion value of lift curve slope coefficient
mechanism and rotating arm devices. is modified according to the following
formula
A significant difference arises
in the treatment of the side force Y (.9 acaL)mod=ldac3r:std[1+(i+:)(1 dciii rd5I1
created by the rudder. This is
calculated on the basis that the rudder
acts like a low aspect ratio wing,
giving the non-dimensional side where K is the imaging factor for the
force/helm angle coefficient as rudder.

A\ )( )(E)2 A manoeuvring prediction tool for

SWATH vessels allows the user to
.E.TAL) dc5 u
determine the size of rudder-required
for a given vessel in order to provide
adequate manoeuvring performance. The
where c is the flow velocity over program can cope with rudders fore and
rudder, A is the rudder area, T is the aft of the propellers and will check
draught and C, is the lift coefficient
the directional stability for a given
for the rudder section. design in addition to estimating the
likely turning performance and heel
The flow velocity ratio term is angles for specified degrees of helm.
dependent on whether the rudder is
subject to propeller induced This program, SWATHMAN3, has been
accelerated flow. For rudders subject run for a number of designs at model
to propeller accelerated flow the ratio and full scale, including the fishing
becomes SWATH ALI. Similarly the program has
been used to evaluate designs with
-
( VA (ACL)\2 rudders fore and aft of the screws in
(2
E) = {1 + flare. 1 I. an attempt to quantify the effect of
rudder type/location upon turning
performance.
where Pa-ea is the proportion of rudder
area subject to the accelerated flow, FULLSCALE TRIALS
ACL the flow acceleration factor due to
the propeller, Vs is the ship speed and The most obvious manoeuvring
Va is the flow speed into rudder. characteristic of the MV ALI was the
directional stability which she
possessed. Despite noticeable yawing
The lift curve slope coefficient
in bow quartering seas the overall
for the rudder is less easy to define.
After trying classical approach,
a course remained straight, with little
1.301 rad-1 was
chosen following or no correction to the helm necessary.
extensive analysis of full scale and
model test data along with published zup_er.inaQat.
comparisons and predictions utilizing
Clarke's regression routines. The experiment itself was a
This
eliminates the effect of rudder aspect simple as could be devised and required
ratio from the calculation. However, little equipment. Consequently these
this is unlikely to effect results for experiments were less disrupted by any
low aspect ratio rudders system failures.
behind
propellers, but may introduce small

43
 The instruments used were the terms of sea-room by the SWATH being
compass, two stopwatches and the appreciably shorter for the same
trailing log. The procedure adopted displacement.
was for one experimenter in the cabin
to initiate the turn with the helmsman
in such a way that a second researcher 5000
at the stern could simultaneously start
his /her stopwatch. Every 10 seconds,
the helmsman reported the compass 4000"
bearing to the first experimenter who
recorded both the time and the heading.
Meanwhile the second experimenter (1) 3000
recorded the speed and logged distance
at the same times. In practice the o
o
speed varied very little in the turns -1 2000
and no systematic speed loss, as would o
have been expected with conventional
ships, was observed and the speed was 1000'
only recorded when there was a change
to be noted. The logged distance
travelled was only a few hundredths on o
a nautical mile each ten seconds and 0 1000 2000 3000 4000 5000
consequently each distance measurement Measured
was somewhat coarse being only ± 0.01
of a nautical mile. However the Figure 1: Cross-plot between measured
accumulated distances were reliable and calculated values
since the total systematic error was of
a similar magnitude. 5000
In Measured
Wherever possible the SWATH was Calculated
turned through two complete turns to
get 3 half turns at a supposedly steady
radius and also to allow the effects of
winds and currents to be observed.

Results
2000 -
The calculated values of distance
travelled were in general in very good
agreement with the measured values as
shown on figs 1 and 2. The best fit 1000 -
straight line show that there is a
negligible bias of only 0.5% and
indicates that there is negligible
drift in the turning of this SWATH.
O
1 2 3 4 5 6 8
b
9 101 12131415161718192 2.
Experiments
The initial yaw rate (over the
first 60 seconds), the average yaw rate Figure 2: Comparison of predicted and
over the full 720 degree turn, the measured distances on a run by run
effective turning radius for the basis
second, third and fourth half (180 Some of the paths are shown in
degree) turns (in reverse order) were
figs 3 and 4). There was a persistent
calculated. This allows the effects of
predominant direction (ie influence of tendency for the second and fourth half
current and wind effects etc) to be turns to be tighter ie there was
frequently an obvious drift to the east
seen. This was
during the full manoeuvre.
The absence of drift is not too the case even when the initial turn was
surprising since the ratio of the to port (west in the case of most of
tactical diameter to the hull length is these trials). Indeed it was the turns
seen to have ranged from about 8 to 40 to port that most apt to run away east
and linear manoeuvring theory is particularly at the higher rev
expected to be valid. The best of settings.
these values are about two to three
times the value for conventional
monohulls but this must be off-set in
44
 300

a
200 5 -
E
a. o .
X
uo 100
s
N. 00
*
. . X
X x
t :
co x
a
E
° a 4 -

m-100 5
. o.

o .

.
4,-200
,
o, . . 3 X
o 16 degs
o
4-300 24 degs
0 36 degs

-400
2
-100 0 100 200 300 400 500
T000 1200 T400 /600 1800 2000
Lateral Distance (m)
Figure 3: Effect of rudder angle on the Engine Revolutions (rpm)
turning path at an engine speed of 1400 Figure 5: Speed achieved at different
rpm engine settings

300- 1400 rpm Using the perceived port/

1700 rpm starboard symmetry it was decided to
E plot the data against the modulus of
1900 rpm 13 0 the rudder angle.
200- This resulted in fig
0
6 from which the following relations
0 may be deduced

at 1400 rpm dW/dt = 0.053 8

4: 13,40' 0

o
at 1700 rpm dW/dt = 0.066 5
- 0 11, at 1900 rpm dW/dt = 0.072 5
to0 or approximately dWidt = N5/26000
e 1,
lb
. e

0
0

13

-200 1400 rpm o o

-100 0 100 200 300 400 1700 rpm
Lateral Movement (m) 1900 rpm
Figure 4: Effect of engine speed on the o
turning path with the rudder 32 degs to
starboard

Figure 5 shows how the average

speed of the SWATH during the turns
varied with engine revolutions. The
SWATH's design speed of 8 knots was
comfortably exceeded at the higher
engine settings of 1900 and 1950 rpm
but the speeds achieved at 1400 and 0 rpm port/1400 rpm stbd
1700 rpm were very similar. This was 0 rpm port/1900 rpm stbd
not what was indicated by the speed
against revolutions curve obtained
during the self-propulsion tests [14] o 10 20 30 40
and meant that the graphs where speed Rudder Angle
is used as an axis have only two sets Figure 6: Linear relationships between
of data points rather than 3. yaw rate and helm angle
45
 The results with differential to the rudders produce an increase in
thrust indicate that it does improve rudder efficiency by increasing
turning performance by producing a effective aspect ratio.
higher yaw rate and a lower steady
turning radius but the trial with 1900 Kaimalino Prediction
rpm against zero was no more effective
than that with 1400 rpm against zero. Kaimalino Trials
In both cases the effectiveness may be Patria Prediction
attributed, in part, to the lower speed 700 - Patria Trials
of the SWATH in these trials. Ê
Ali Prediction
600 -
When the data is plotted against Ali Trials
speed it is the steady turning radius
which is well behaved (fig 7). Over 500 -
the speed range tested a linear fit is
quite good with the gradient being 400-
roughly inversely proportional to the
rudder angle suggesting R a V/S. This 300 -
simple fit can only have a limited
range of application.
200-

500 100 - .4
+ 16 degs rudder
E o
24 degs rudder
o 400 o 2 3 4 5
a X 32 degs rudder Rudder Area (sq m)
m
r:4
Figure 8: Predicted and fullscale
turning for KAIMALINO, PATRIA and ALI
300

Fullscale @ 1700rpm
a
200 Fullscale @ 1400 rpm

7:5 Prediction (K=1.5)

m
o -le Prediction (K=1)
100 500

O
400
3 4 5

Speed in m/s 300

Figure 7: Turning radius against speed
for different rudder angles
200
COMPARISON WITH SWATHMAN

Figure 8 presents SWATHMAN

together with trials 100
predictions
results for minimum turning diameters/ 0 10 20 30 40

rudder area for the vessels PATRIA, Rudder Angle (degrees)

ALI, and KAIMALINO. It may be seen
that the predictions agree well with Fig 9: Turning performance for ALI
values measured on full scale trials.
Figures 9 present SWATHMAN showed that the
Miller [8]
predictions of turning performance manoeuvring characteristics are quite
(Turn Diameter / Helm Angle) for the sensitive to the value assumed for the
vessels ALI. Reasonable agreement with imaging factor, the acceleration of the
measured values is again observed. Fig
flow due to the propeller and the
4 also demonstrates the influence upon proportion of the rudder that is
manoeuvring performance of rudder subjected to that accelerated flow.
imaging effects. This phenomenon
occurs where the proximity of large
flat areas close to and perpendicular
46
CONCLUSIONS Narita, H., and Mabuchi, Y.,
"Design and Full Scale Test Results of
SWATH ALI used the rudder flap Semi-Submerged Catamaran (SSC)
arrangement and so this data set is Vessels", Proceedings 1st International
unique and peculiar to ALI. Marine Systems Design Conference
(I.MSDC) , London 1982.
The main findings are that Coe, T., 'A Technical Evaluation
of the 60 Foot SWATH Ship HALCYON to
a simple experiment such as the Determine Utility in Coast Guard
one reported here can yield valuable Operations', Intersocity Advanced
information; Marine Vehicles Conference, Arlington,
this SWATH did not lose speed or USA, June 1989
drift in a turn to any noticeable Warren, N., Private Communication
degree: regarding manoeuvring performance on
predictions of path length were trials of the MV "PATRIA". Nov 1990.
very accurate; Miller, A.F.,"Aspects of SWATH
wind and current drift during Design and Evaluation", MSc Thesis,
turns can be appreciable particularly Glasgow University. September 1991
if the turn is gentle; Miller, A.F. and McGregor, R.C.,
(4) yaw rate is proportional to 'Development of SWATHMAN, a Manoeuvring
rudder angle and the gradient depends Simulation Tool for Small Waterplane
on the engine revolutions and steady Area Twin Hull Ships', Proceedings of
turning radius is inversely 2nd Motion Control and Modelling
proportional to rudder angle which is Conference, Southampton, UK, July 1992
as would be expected from linear MacGregor,J.R., Bose, N. and
manoeuvring theory; Small, G.'Design and Construction of a
steady turning radius increases Small Waterplane Area Twin Hull (SWATH)
with speed as indicated in ref[2] for Fishing Vessel', Proc. World Symp. on
SEAGULL; Fishing Gear and Fishing Vessel Design,
The manoeuvring prediction St John's, Newfoundland, Canada, Nov
program may be applied to SWATH 1988
vessels, operating in the low to medium Djatmiko, E.B., Chun, H.H.,
speed range, fitted with rudders both McGregor, R.C., and MacGregor,
in and out of the propeller slipstream J.R.,"Hydrodvnamic Behaviour of SWATH
and possessing widely different Fishing Vessel', CSME Mechanical
resistance and propulsive Engineering Forum, Toronto, June 1990
characteristics. Clarke, D., Gedling, P., and
Hine,G., 'The Application of
ACKNOWLEDGEMENTS Manoeuvring Criteria in Hull Design
Using Linear Theory', TRINA, Vol 125,
In addition to acknowledging the 1982.
financial support of the Science and Fein, J.A., "Vertical and
Engineering Research Council (through Horizontal Plane Control of SWATH
mTD Ltd), Vickers Shipbuilding and Ships", Proceedings of 7th Ship Control
Engineering Ltd, Yarrow Shipbuilders Symposium.
Ltd and YARD Ltd, the authors wishes to Hart, C.J. et al, "Rotating Arm
recognise the contributions made by Experiment for an Extended Strut SWATH
Jacqueline McKay and Graham Small (as Ship as Represented by SWATH 6E",
skipper and helmsman) to the conduct of DTNSRDC SPD Report 0698-03,Sept 1983.
the trials. MacGregor, J.R., "A Computer
Aided Method for Preliminary Design of
REFERENCES SWATH Ships", PhD Thesis, Glasgow
University, May 1989.
1. Nethercote, W.C.E. et al, McGregor, R.C., Miller, N.S. and
"Manoeuvring of SWATH ships", Byrne, J.B.,'Self-Propulsion Trials of
Proceedings of 20th American Towing Fishing SWATH ALI on Skelmorlie
Tank Conference, Hoboken, New Jersey, Measured Mile', Univ of Glasgow Marine
July 1983.
Technology Centre Report 92-03, 1992

47
SESSION I-B

SEAICEEPING
 The Seakeeping Committee Report
A.W. Troesch (University of Michigan, USA)

The Seakeeping Committee of the Twenty- In addition to large amplitude motion

third American Towing Tank Conference is projects, specialized experiments
comprised of the following members: relating to the seakeeping qualities of
offshore structures, sailing vessels,
Armin W. Troesch (Chairman) and planing hulls are reported. These
Department of Naval Architecture and unconventional experiments require a
Marine Engineering variety of innovative testing equipment
University of Michigan and testing techniques.

Stuart B. Cohen Individual reports are given by the

Ship Hydrodynamics Laboratory following member institutions:
University of Michigan

Lewis Motter SHIP HYDRODYNAMICS LABORATORY

CARDEROCKDIV UNIVERSITY OF MICHIGAN
Bethesda, MD
Stuart B. Cohen
John J. Zseleczky
Hydromechanics Laboratory Since last ATTC, the Ship Hydrodynamics
U. S. Naval Academy Laboratory at the University of Michigan
has engaged in a major effort to improve
every aspect of seakeeping testing to an
A significant reduction in new Navy accuracy and repeatability level with an
designs has meant a drop in the type of error of 0.1% of the maximum value of a
seakeeping experiments as reported in measurement. Since most transducers and
previous ATTC meetings. The thrust of instruments are rated according to their
the activity among various ATTC members linearity fullscale, this effort
appears to be directed away from requires predicting the maximum values
standard seakeeping tests and towards during an experiment and adjusting
more one-of-a-kind research projects. outputs, digitizers, filters and
displacements to provide fullscale
Typical of the these research oriented output for the maximum measurement.
projects is the study of the
hydrodynamics and dynamics associated During recent tests of an International
with large amplitude motions. Nonlinear America's Cup Class (IACC) yacht model,
loads and responses in severe seas are these levels of accuracy were checked.
becoming important in the design process Moments of inertia were found for hull,
as structural damage due to impact and keel, cradles and swing table
bow flare slamming become known. The separately, and found to agree with
corresponding members report experiments total model values. Measurements of
involving flexible PVC models wavemaker input signal to output wave
(CADREROCKDIV), steep wave motion tests form were checked throughout the tank.
(US NAHL), and segmented model tests Carriage speed was measured three ways
with video verification of wetted and found to agree. Errors due to cross
surface (UM SHL). The results of these coupling in dynamometers was eliminated
tests are expected to appear in the by four dimensional surface curve
literature before the next ATTO meeting. fitting. Capacitance wave probes were
developed that give good static
51
measurements with corrections for to determining primary bending moments
meniscus effects. Even with corrections between the cross structure and the
for frequency and speed, measurements at struts, local slam impact loads on the
forward speed contain 1% uncertainty. cross structure, and wave slap loads on
We cannot get better accuracy with non- the side shell.
contact (ultrasonic) wave probes. This
area still needs more work. The Navy is always interested in
reducing the weight of ship structure to
The recent IACC yacht tests include allow for increased payload without
increasing total ship displacement. In
heave, pitch, surge, roll moment, yaw
moment, yaw angle and added resistance support of that objective, a flexible
measurements. Using three different PVC ship model was constructed to the
models, the conditions tested were at hydrodynamic and the structural detail
two drafts, at three wave slopes, at for the CG-47. Test were conducted in
three speeds, for six wave lengths, two severe sea conditions. Structural
moments of inertia, two beams and one details were changed twice and the model
heel and yaw angle besides the baseline re-tested. One configuration represented
upright condition. Not all conditions the modified design used in later ships
were tested for all models. Results are in the Class. The second configuration
expected shortly. represented a new configuration change
being considered for the Class.
Longitudinal shear force, longitudinal
CARDEROCKDIV bending moment and torsional moments
Bethesda, MD were measured in addition to local plate
loads, and the rigid body motions. The
Lewis Motter model results agreed well with ship
experiments and were correlated with
There has been a reduction in the number computer prediction techniques.
of new Navy ship designs and major ship
modifications. As a result there have The Navy is re-considering the survival
been only a few significant seakeeping criteria limits of damaged ships.
experiments at CARDEROCKDIV that may be Currently Navy survival criteria is
In three of based on static stability analysis. In
of interest for the ATTO.
the four recent experiments, order to identify the important
hydrodynamic structural loads were the parameters effecting capsize and dynamic
main objective. stability limits for existing ships,
several models of existing ships were
The extensive model experiments for the tested. models were constructed so that
first large Navy SWATH, T-AGOS 19 were internal compartments in the damaged
completed in 1990. Ship hydrodynamic area could be modeled and allowed to
loads and structural responses were one free flood. The models were tested in
of the primary objectives of the severe waves and winds equivalent to 100
experiments. The ship is complete and knots full scale. See Figure 1. No
has commenced dedicated full scale
capsizing occurred during any of the
trials in February of this year. The experiments.
manned trials will continue through the
end of this year. A response monitoring
system will remain on board for several Hydromechanics Laboratory
more years to collect statistical data U. S. Naval Academy
on 85 channels of seakeeping and loads
information. Analysis of available John J. Zseleczky
trials data indicates that seakeeping is
as good as expected with model study The U.S. Naval Academy Hydromechanics
results being confirmed. Laboratory has been involved in an
experimental study of nonlinearities in
Experiments have been conducted on a new seakeeping. A 1:50 scale model of the
T AGOS-23 SWATH concept. During these ITTC standard SF-175 hull was tested in
experiments, special emphasis was given regular waves with varying steepness, to

52
 Figure 1. Dynamic stability tests on a Navy vessel.

investigate the point at which pitch, waves will be compared with the results
heave and acceleration motions were no obtained in random seas (Reference B).
longer linear functions of wave
elevation. This work was performed as a
cooperative effort with Dr. John O'Dea REFERENCES
of DTRC, as an extension of his earlier
joint project with the University of O'Dea, J., Troesch, A., "Comparative
Michigan (Reference A). Seakeeping Model Experiments," 21st
ATTO, Washington, D.C., August,
Experiments were also conducted with the 1987.
SF-175 model in random head seas. The
intent was to obtain a large sample of O'Dea, J., Powers, E. and Zseleczky,
model motions data in waves with a J., "Experimental Determination of
nominally constant wave energy spectrum. Nonlinearities in Vertical Plane
Ten minutes of run time were accumulated Ship Motions," ONR 19th Symposium on
for each of three sea states. This is Naval Hydrodynamics, Seoul, Korea,
equivalent to over an hour of full scale August, 1992.
run time per sea state. The data are
being analyzed by Dr. Edward Powers of
the University of Texas, using state-of-
the-art signal analysis techniques
developed in the electronics industry.
The results from the tests in regular

53
 SHIPMO Seakeeping Predictions and Correlations
B. Menon, J. Vienneau (Fleet Technology Limited, Canada)

ABSTRACT

Experience with the strip theory-based numerical seakeeping performance prediction

model SHIPMO is described with applications ranging from fast surface vessels to
unconventional hull forms such as icebreakers. A brief history of the model's development, its
salient features and examples of validations are given. Four comparisons between predictions
and measurements from scale model tests are illustrated, along with limited comparisons against
predictions of other standard seakeeping models_ These include pitch and heave motions in
head seas, unstabilized roll motions in beam seas, as well as motions in stern quartering seas.
Correlations have been made both for regular as well as irregular waves. It is shown that
regular wave pitch, heave and roll RAO's correlate well, while motions in irregular seas
correlate to a lesser degree. Comparisons with other models show that SHIPMO predictions are
as good as those from other established programs based on strip theory.
1. INTRODUCTION

Interest in the seakeeping performance of non-naval vessels has increased dramatically

in recent years. The reasons for this range from varying roles of purpose-built vessels, to
conversions that necessitate re-examination of operational limits. Traditionally, strip theory-
based numerical seakeeping prediction codes were used only for naval ships and commercial
vessels with large length to beam ratios. The applications ranged from ship design and
evaluation to operational seakeeping assessments and research. Many of the numerical
seakeeping assessments currently undertaken involving smaller vessels and vessels with
unconventional bow or stern forms, are not supported with scale model testing and correlation.
Hence, it is essential that the tools used for the assessments be validated. DREA's strip theory-
based, linear frequency domain code SHIPMO has been extensively used in Canada for such
applications. A number of examples have been chosen mostly from hitherto unpublished
sources, that demonstrate the validity of the numerical predictions for a variety of hull forms.
Over the past fifteen years, several comparisons of strip theory predictions have been
made with model experimental data for head seas. In the early eighties, attempts to correlate
strip theory predictions with model test data were encouraging. Murdey [1] demonstrated the
accuracy of predictions through comparison with data from tests with the NRC hull form series
for fast surface ships. It was concluded that, for Froude numbers below 0.35, early versions
of strip theory models predicted heave and pitch motions in regular waves to within 20% of the
maximum measured response. Karppinen [2] presented comparisons of strip theory predictions
with model test data for a wide-beam fishing vessel. Heave RAO's for both head and beam seas
were predicted to within 15% even at the relatively high Froude number of 0.38, with a lesser
degree of correlation for pitch RAO's. Roll predictions at zero speed were within 10%.

55
 HISTORY OF DEVELOPMENT
A desire for improved lateral motion predictions was a primary motivation for DREA
to develop its own strip theory ship motions program. Schmitke [3] introduced new methods
for including roll damping contributions from ship appendages. Most notably, the importance
of dynamic lift on foil elements had not been incorporated in previous strip theory programs,
such as Salvesen, Tuck and Faltinsen [4].
The original SHIPMO program predicted ship motions for six degrees of freedom in
regular and irregular waves, and included the roll damping prediction methods developed by
Schmitke. Predictions of slamming pressures and forces have been a feature of SHIPMO since
its original development.
Subsequent revisions to SHIPMO have included improved viscous roll damping
predictions, discussed briefly by Cumming, Haddara and Graham [5], and a hydrostatics
capability for predicting ship draft and trim given ship displacement and LCG. Recent
SHIPMO developments include prediction of the incidence of crew tipping and. sliding events
[6], and the incorporation of JONSWAP, Ochi six-parameter and directional ten-parameter
wave spectra [7].
FEATURES OF SHIPMO
This section presents an overview of the capabilities and brief descriptions of the
standard constituent algorithms in SHIPMO. The basic structure of SHIPMO is shown in Figure
1. The main computational blocks in the program are:
hull form calculations - basically hydrostatics;
added mass and damping calculations - done for each station over
a suf ficiently wide frequency range and stored in large arrays;
response calculations - solutions of the equations of motion, using
data from the added mass and damping arrays and nonlinear roll
terms;

seakeeping calculations - motions, probabilities of events,

slamming pressures, etc., in irregular seas.

The basic computation of pitch and heave response are adapted fromdynamics Frank and
of
Salvesen [8]. The mathematical model developed by Schmitke [3] for the lateral
an unstabilized ship, forms the basis of the sway, roll and yaw motion prediction capability.
The linear equations of motion are formulated with reference to a stability axis system fixed
at the centre of gravity of the vessel. The sectional added mass and wave-making
The
damping can
close-fit method
be computed either by the close-fit or the con formal mapping method. Lewis form
requires complete hull geometry input, while the conformal mapping method uses a
which requires the sectional area, beam and draft. Generally, the Lewis form option is used
to obtain rapid, first estimates of motions for normal sections; however, for excessively bulbous
amplitudes and
sections the MIT bulb form is used. The equations of motion are solved for the
phases in regular waves and the results extended to irregular seas by which
application of the
quantitative
superposition principle. This yields the root mean square motions from
measures of seakeepine are derived.

56
 Relative motions are calculated at any station using the resultant motions of the point
and the incoming wave profile at the station. Experiments have shown that wave deformation
by the hull is significant and in fact, the assumption of an unde formed wave is valid only at
the forward perpendicular. This phenomenon of wave deformation by the oscillating hull,
referred to as dynamic swell-up is accounted for in SHIPMO to correct the relative motion
response using experimental data. Another built-in feature is the calculation of the wave
profile resulting from forward motion. This semi-empirical correction is applied to the
calculation of the probability of deck wetness.
Two semi-empirical seaway spectra are available in SHIPMO: one is the Gospodnetic-
Miles quadratic regression spectrum [9], derived from data obtained at Station India in the
North Atlantic; the second is the Bretschneider two-parameter spectrum, of which the
parameters are the significant wave height and the modal wave period. The option to use a
measured spectrum is also provided with the wave frequency and the spectral ordinate as the
input parameters. Calculations of responses to short-crested seas can be performed by
specifying the spreading angle which is applied in the form of a cosine-squared function.

Slamming algorithms incorporate the Ochi-Motter [10] and the Stavovy-Chuang [11]
methods. Of much greater importance than the probability of slamming, is the statistical
estimate of slamming pressure. SHIPMO calculates P. defined as the most probable peak
impact pressure experienced during time duration t using the probabilities of keel emergency
and relative velocity. Probability of deck wetness (green water over the deck) at station x is
calculated accounting for the freeboard at station x. Because of the wave profile and dynamic
swell-up, probability of deck wetness is generally greatest somewhat aft of the forward
perpendicular.

Human tolerance to low frequency motions typical of conventional displacement ships

is assessed in terms of a vibration ride quality index (VRQI). The VRQI is used to quantify
human tolerance to vertical ship motions. In this procedure, the physiological ef fect of a
vehicle's acceleration time history is assessed by the low frequency- and visceral models that
are appropriate to conventional displacement ship motions.

SHIPMO incorporates semi-empirical estimates of roll damping; furthermore, fairly

extensive comparisons of predicted and measured roll response in oblique waves have shown
good agreement at all relative wave headings, suggesting that the program yields predictions
of suf ficient accuracy for preliminary design purposes. An iterative procedure determines
nonlinear roll damping terms, which are dependent on roll amplitude.
The model of ship roll, sway and yaw motions in oblique long-crested seas, includes the
capability to simulate the ef fects of rudder, fin and passive tank roll stabilizers. The complete
model is obtained by adding the methods for simulating the ef fects of rudder, fin and tank roll
stabilizers to the unstabilized ship model. Rudders and fins are treated as hydrofoils and the
effects of fin-hull interference, fin-bilge keel interference and fin-fin interference in multiple
fin configuration are accounted for. Simulation of the ef fects of passive tank stabilizers is an
additional feature of SHIPMO.
4. APPLICATIONS

For the last ten years, the original version of SHIPMO has been extensively used by the
marine industry in Canada. A large number of correlations of its predictions with
experimental data have been carried out by NRC(IMD) and DREA [12] using experimental data
for a wide variety of monohull forms. More recently, applications of SHIPMO (original
version) to the assessment of seakeeping performance qualities of unconventional hull forms
such as icebreakers, ocean-going barges and stationary drillships have also been associated with

57
correlations against model test data. A number of examples from both the above sources have
been selected for presentation here. In some cases, independent calculations of the motions
using other standard seakeeping codes such as SMP and SCORES were also available for
comparison. RAO's derived from regular wave tests as well as motions measured directly from
irregular wave tests are used for comparison.
The vessels selected and the response parameters correlated against measurements are
listed below:
MV ARCTIC (an icebreaking OBO carrier L/B = 9.0, BIT = 3.39,
CB = 0.67), head seas pitch and heave RAO's;
A fast surface ship represented by the NRC hull form series (L/B
= 8.52, BIT = 3.28, CB = 0.48), head seas pitch and heave RAO's:
An oceanographic research vessel (L/B = 5.6, BIT = 3.0, CB =
0.50), heave and roll motions in beam and quartering irregular
seas;

A fisheries patrol vessel (L/B = 4.43, B/T = 3.11, CB = 0.525),

beam seas roll RAO's.
Comparisons have also been made between pitch and heave RAU amplitudes predicted
by SHIPMO and SMP81 [13]. Limited comparisons of SHIPMO predictions in irregular seas
have also been made with documented predictions from program SCORES [14].

S. COMPARISON OF PREDICTIONS WITH MEASUREMENTS

Amplitudes and phases of heave and pitch RAO's of MV ARCTIC in head seas have
been correlated with regular wave model test data from [15], as shown in Figure 2. Data for
two Froude numbers are depicted over the entire non-dimensional frequency range of interest.
Both pitch and heave amplitudes show close correlation over the low frequency range. The
phase angles are closely correlated over the intermediate frequency range. At the high end the
model phase angle data show some scatter, but their influence on the computed relative
motions, accelerations, etc. is less than that at the lower frequencies as the absolute magnitude
of motions are small. Figures 3 and 4 show comparisons of predicted heave and pitch RAO's
with regular wave test data for a model of the NRC hull form series [16]. The continuous
curves representing SHIPMO and SMP predictions compare very closely. The model test data
for the pitch response closely follow the trends of the numerical predictions, while heave
response show deviations are the intermediate frequencies. Unstabilized roll response RAO's
shown in
for a patrol boat [17] in beam seas at zero speed and at 12 knots forward speed are
Figure 5. The roll response amplitude is predicted well at all frequencies except near
resonance, where the amplitude is slightly- underpredicted. Examples of further investigation
of the degree of correlation of roll response is depicted in Figure 6, in which angle
motions in
is slightly
unidirectional irregular seas are used [18]. Signi ficant single amplitude roll
underpredicted in both quartering and following seas. Results of predictions by SCORES are
also shown for comparison. Significant single amplitude heave motions at the same two
headings are shown in Figure 7, where, although motions are slightly- overpredicted, closer
comparison with measurements is seen.

58
 CONCLUSIONS

Linear frequency domain codes of ship motions in waves incorporate a number of

scientific disciplines. These include dynamics of rigid bodies, theory of ocean waves and their
interaction with floating bodies, and the theory of stochastic processes. The validities of the
various theories are, therefore, manifested in the performance predictions for the particular
vessel being designed or evaluated. As the hydrodynamic model of vessel response to waves
forms the core of all seakeepine prediction codes, it is logical to expect closer correlation of
the RAO's from experiments in regular waves. This is demonstrated in the comparisons
illustrated above. In many instances, tests in irregular oblique waves do not yield data suitable
for correlation of the predicted response owing to a number of reasons. These include
insuf ficient number of encounters, unavailability of phase data, degree of short crestedness
in the case of data collected under natural conditions, etc. On the other hand, often in
performance assessment tasks, as cost considerations preclude regular wave tests, it becomes
necessary to accept data from limited irregular wave tests for evaluation and correlation. The
data sets chosen from irregular wave tests permitted direct correlation of the model test data
with numerical predictions.
ACKNOWLEDGEMENT

The authors would like to thank Dr. Indra Datta of IMD and Dr. Kevin McTaggart of
DREA for their assistance and encouragement in the preparation of the manuscript of this
paper.

REFERENCES

Murdey, D.C., "Seakeeping of the NRC Hull Form Series: A Comparison Between
Experiment and Theory", Proc. of the 19th ATTC, Vol I, July 1980.
Karppinen, T.O., "Comparison of Theoretical Seakeeping Predictions with Model Test
Results", Proc. of the 20th General Meeting of the ATTC, Vol II, p 1047, August 1983.

[31 Schmitke, R.T., "Ship Sway, Roll and Yaw Motions in Oblique Seas", Transactions,
Society of Naval Architects and Marine Engineers 86, 26-46 (1978).
Salvesen, N., Tuck, E.O. and Faltinsen. O., "Ship Motions and Sea Loads", Transactions,
Society of Naval Architects and Marine Engineers 78, 250-287 (1970).
Cumming, D., Haddara, M.R., and Graham, R., "Experimental Investigation of Roll
Damping Characteristics of a Destroyer", in STAB '90, Fourth Conference on Stability
of Ships and Marine Vehicles (Naples, 1990).
Graham, R., "Motion-Induced Interruptions as Ship Operability Criteria", Naval
Engineers Journal 102 (2), 65-71 (1990).
Graham, R., and Juszko, B.A., "Parameterization of Directional Spectra and Its
Influence on Ship Motion Predictions", submitted to Journal of Ship Research.

Frank W., and Salvesen, N., "The Frank Close-Fit Ship Motions Computer Program",
NSRDC Report 3289, June 1970.

Gospodnetic, D. and Miles, M., "Some Aspects of the Average Shape of Wave Spectra at
Station "India" (59 N, 19 W), International Symp. on the Dynamics of Marine Vehicles
and Structures in Waves, London, 1974.
59
Ochi, M.K. and Motter, L.E., "Prediction of Slamming Characteristics and Hull
Responses for Ship Design", Transactions, Society. of Naval Architects and Marine
Engineers, 81, 144-176 (1973).

Stavovy, A. and Chuang S.L., "Analytical Determination of Slamming Pressures for

High-Speed Vehicles in Waves", J. of Ship Research, Vol 20, Dec. 1976.

Graham, R., "Ship Motion Program SHIPMO: Model Test Correlations and Suggestions
for Use", DREA Tech. Memo 84/Z, December 1984.
Meyers. W.G., et al, "User's Manual for the Standard Ship Motions Program SMP",
DTNSRDC/spd-0936-01, Sep. 1981.

Raf, f, A.I., "Program SCORES, Ship Structural Response in Waves", SSC Rep. No.230,
1972.

Datta, I., "MV ARCTIC and Two Alternative Bow Designs - Results of Seakeeping and
Slamming Experiments with Model Hulls 361A, 361B and 361C", IMD Report LTR-SH-
385 prepared for TDC, October 1984.
Rogers, F., "A Preliminary Validation of the Seakeeping Prediction Program SCMA3",
IMD Report LM-1991-24, Limited Circulation, December 1991.
Pink, R., "Model Tests to Evaluate a Flume Stabilization System for a Hydrographic
Survey Vessel", Report No. X /0/2961 submitted to Department of Fisheries and Oceans
by BHC Ltd., November 1981.
Soderberg, P., "Model Experiments in Beam and Quartering Seas", SSPA Report 4325-1
to IMD, February 1986.

60
 Read Sea Directions, Sea States, Speed Range, E3asic
Particulars, Offsets, etc. Compute Hydrostatics & Parameters
Independent of Speed, Roll Amplitude & Frequency. Read in
Other Inputs, Seakeeping Positions, Appendage Data.

Compute (or read from disk) Hull Sectional Coefficients

Over Specified Encounter Frequency Range.

Begin Speed Loop

Compute Terms Dependent on Speed, But Independent of

Frequency & Roll Amplitude.

Begin Sea Direction Loop

Begin Loop Over All Sea States

If Sea State Specified, Estimate Initial RMS Roll Amplitude

(Begin Wave Frequency Loop For Regular Response)

Compute Foil Terms Dependent on Frequency

Compute Hull Coefficients and Exciting Forces

Compute Hull Circulatory Effects

Compile All Exciting Forces and Construct Hydrodynamic Matrix

for Ship Motions Independent of Roll Amplitude

Save All Terms Independent of Roll Amplitude

Compute Bilge Keel, Skeg, Viscous Roll Damping, Eddy-Making

& Tank Terms Dependent on Roll Amplitude

Complete Hydrodynamic Matrix and Solve for Regular Response

Complete Motion, Slamming & Wetness Calculations in

Unidirectional Seas.

Compute Motions, Slamming & Wetness in Short-crested Seas

Figure 1 SHIPMO Overview

61
 Figure 2(b)
Figure 2(a) MV Arctic Pitch and Heave RAO's
MV Arctic Pitch and Heave RAO's 100 /11111'11'1111 11111.'11111111111.1 300
2.5 1.2 SHIPMO - Fr.. 0.19
_ SHIPMO - Fr.. 0.19
- S-11P103 - Fr..0.111 SHIPMO - Fr. . 0.17 SHIPMO - Fr.. 0.17
SHIPFC - Fr. -0.17 Model Tests -Fr.. 0.17 Mt.' Tests- Fr.. 0.17
Pitch Mo. T..- Ff. - C.17 50 Model Tes.- Fr.. 0.19 Model Test - Fr.. 0.19
lbci Tat.- 200
2
0.8
dl
Heave --
0.6 100
- - - SHIPP.- Fr. - 0.19 cil
SHIPIA0- - 617 co
Mad, Tot. - Ff. -517
0-
Wm*, Twb - MI
o ,
Heave E:

Pitch -100
0.5 - -150
2,, -0.2
-200 1111111 1/11.11111 II/ -200
- -0.4
3.5
1 1

4 5 0 0.5 1 1.5 2 2.5 3 4

0 1 2 3
Non-Dim. Frequency,co AL/g) Non-Dim. Frequencym ,l(Ug)

Figure 3(a) Figure 3(b)

NRC Hullform Series NRC Hu Ilf orm Series
1.2
12
SHIPM0 - Pitch Fr.. 0 .2
SHIPMO - Heave Fr.. 02 SMP - Fr.. 0.2
- - - SMP - Heave Fr. -02 o MOtlei Tests - Pitch Fr.. 0.2
o Model Tests - Heave Fr.. 0.2

0.8 g- 0.8
E

co 0.6 0.6
E
E E 0.4
5 0.4 o
o
0.2 0.2

o
2.2 3.4 4.5 o 1.1 2.2 3.4 4.5
o 1.1
Non-Dim. Frequency,co .1(LJg) Non-Dim. Frequency,co

Figure 4(a) Figure 4(b)

NRC Hu liform Series NRC Hullform Series
1.2
1.2

ci_
E 0.8
o
co

0.6

E 0.4
0.4
o
zo
SHIPM - Heave Fr.. 0.3 0.2
0.2 SMP - Heave Fr.. 0.3
o Model Tests - Heave Fr.. 0.3

0
I
1.1 2.2 3.4 4.5
o 1.1 2.2 3.4 4.5
Non-Dim. Frequency,o.) -4(1../g)
Non-Dim. Frequency,co -V(Ug)

62
 Figure 5(a) Figure 5(b)
Hydrographic Survey Vessel Tully Hydrographic Survey Vessel Tully
6 4
o Model Tests Beam Seas o Model Tests Beam Seas
SHIPMO Speed 0 knots 3.5 SHIPMO Speed 12 knots
5
3
o 4 "Ti
c
cr 2.5
3 2
o o
Z 1.5
2
1

1
0.5

o o
o 0.5 1 1.5 2 2.5 3 3.5 o 0.5 1.5 2 2.5 3 3.5
X/L X/L

Figure 6(a) Figure 6(b)

Oceanographic Research Vessel Oceanographic Research Vessel
20 12
Heading 300 Hea'ding 900
+ SH IPM 0 Roll SHIPMO Roll
Model Tests Roll 10
16 Model Tests Roll
SCORES Roll 10 Kt
SCORES Rol!. 10 Kt
o 8
-14
12
o
cc
6
.3 8
4 *17+.
"
4 -
------
- 2

o o
2 3 4 5 6 2 3 4 5 6
Wave Height, m Wave Height, m

Figure 7(a) Figure 7(b)

Oceanographic Research Vessel Oceanographic Research Vessel
2 2.5

2
1.5
E
E
c2;
co
1.5
1
E'
co
e 1
'E e + ,
Heading 300
"
0.5 Cii - Heading 90°
SHIPM 0 Heave .C
4e 0.5 + SHiPM0 Heave
Model Tests Heave
Model Tests Heave

O
1111 //Ill SCORES Heave, 10 Kt

o
SCORES Heave, 10 Kt

1 2 3 4 5 6 1 2 3 4 5 6
Wave Height, m Wave Height, m

63
 Initial Seakeeping Tests on an IACC Yacht Hull Model:
Two Ballast Conditions
S.B. Cohen, R.F. Beck (University of Michigan, USA)

waterline (OSWL) and the new sailing

ABSTRACT waterline (NSWL), both having the
same radius of gyration and VCG. Test
The Partnership for America's Cup were made of upright resistance and
Technology provided a 1:5 scale model model motions in smooth water and in
of an International America's Cup regular waves. The dynamometer used
Class yacht model and keel, which was the Heave-Pitch-Surge (H-P-S)
was tested at two drafts, both having dynamometer which allows the model to
the same radius of gyration and VCG. be free to surge, pitch, and heave
Test were made of upright resistance while restraining all other degrees
and model motions in smooth water and of freedom. Preliminary tests were
in regular waves. The model was free made to determine whether to test the
to surge, pitch, and heave while model in the free to surge mode
restraining all other degrees of (which is more realistic) or in the
freedom. Initial tests were made to fixed in surge mode (which allows the
compare the model in the free to determination of phase angles).
surge mode (which is more realistic)
and in the fixed in surge mode (which The model test conditions were
allows the determination of phase designated OSWL-Fixed for the light
angles). draft loading and fixed in surge,
OSWL-Light (or sometimes just denoted
NOMENCLATURE as OSWL) for the light draft loading
in the free to surge condition, and
3 . heave amplitude (feet) NSWL-Heavy for the heavy draft
CS =, pitch angle (degrees) loading in the free to surge
0 . wave amplitude (feet) condition. The same keel was used for
. density (lbs * sec2/ft4) all tests. Both NSWL-Heavy and OSWL-
Light were tested at two scale
A. . wavelength (feet) speeds. All tests were done in the
Alle* = heave coefficient = WC() 0° heel, 0° yaw condition. The test
zeta5* = pitch coefficient method used for the free to surge
= Lwi/ 2n o cases was an over-speed and under-
RUn* = total resistance coefficient speed method developed especially for
= RT Pg Lwl Co2 this project, and was found to give
Rail* = added resistance coefficient less noisy results than a constant
speed, fixed in surge test.
= RAW / Pg 1--wl
= acceleration of gravity
Turbulence stimulators were put
on the model in accordance with the
INTRODUCTION
recommendations of the ITTC. All
studs were cylinders 3 mm diameter
The Partnership for America's and 2.5 mm high. Studs for the hull
Cup Technology (PACT) provided a 1:5 were placed 25 mm apart (on center)
scale model of an International
with the primary girth at the inter-
America's Cup Class (IACC) yacht section of the sailing waterline
model and keel, which was tested at (OSWL) and profile with secondary
two drafts, the original sailing girths 1 foot forward and 1 foot aft.
65
Studs were also glued on the keel 25 measured to get the "as tested"
mm on center at the 1/4 chord. Model values. The as tested values are
particulars are given in Tables 1 and listed in Table 2.
2 below.
CALM WATER RESISTANCE

The model calm water resistance

Table 1 was measured at only one speed
OSWL and NSWL Model Particulars as corresponding to the highest speed,
Tested full scale. The results for OSWL-
Light and NSWL-Heavy had variations
of 0.10 and 0.20 pounds respectively.
OSWL NSWL To get improved accuracy at the
Weight (% OSWL) 1.0000 1.2557 nominal speed, the free to surge data
VCG above keel / LWL 0.1454 0.1620 with its small speed variations can
0.2426 0.2345 not be simply averaged. These data
Radius of gyration / LWL
did not have enough range, so that a
Moment of Inertia at CG (% was used in which the
1.0000 1.1832 curve fit
OSWL) .
linear fit had a slope equal to the
slope of the resistance curve at 9
knots. The values calculated for Rt
Table 2 calm at the nominal speed did not
OSWL and NSWL Model Particulars as vary significantly when the slope of
Designed the fit was raised or lowered by 5

percent.

OSWL NSWL TEST EQUIPMENT

Lwl (% 1.000 1.070
The wave probes are a
OSWL) capacitance type utilizing a single
L/J3 4.214 4.066 wire sealed in a 0.059 inch (1.5 mm)
WI' 6.213 5.631 O.D. glass tube. Since static
Weight (% 0.9981 1.2528 calibration was the only practical
OSWL) method for in situ wave probe checks,
Radius of 0.2395 0.2395 curves were developed to allow
gyration / L correction for forward speed and
dynamic effects. The corrections were
determined by dynamic testing of the
wave probes at forward speed. The
The VCG and the pitch gyradius probes were oscillated vertically
of the model were found using our using our VMM for a range of
frequencies, amplitudes and forward
standard swing table. The VCG is The correction function was
found simply by applying a known speeds.
found to be dependent on oscillation
torque and measuring the angle of and forward
rest. The gyradius is found from the frequency, amplitude,
speed. In general, the correction was
VCG and the period of free swing. The
less than 10% of the wave amplitude.
known torque was supplied by moving
calibrated weights known distances. precision of the
An attempt was made to keep the VCG
The
capacitance wave probes used in these
as low as possible but the internal tests was determined to be 0.5% for
shape of the hull curved upward so calibrations. Absolute
rapidly that very little longitudinal static
possible without accuracy of the dynamic wave probe
movement was 0.5%; however,
considerable vertical rise. After the calibration was
the dynamic forward speed calibrations of
towing tests were complete, the wave probes gave repeatability
models were reweighed, and the VCG
levels of ±3%. To develop a corrected
and pitch gyradius were again
66
wave height given the measured wave each wavelength a wait time was
height, measured wave period, and
forward speed, calculated from wavemaker start up so
a least squares
polynomial surface fit of the dynamic that the carriage began accelerating
test results was used. The correction at the same point in each wave train.
term was calculated and then added to
the measured wave height for each Data was scanned on all
run.
channels for 10.24 seconds at 100 Hz
The dynamometer used for these (1024 points/channel at a period of
tests was the Heave-Pitch-Surge 0.01 seconds). Scanning was always
(H-P-S) dynamometer. The H-P-S begun at the mid tank access door 160
dynamometer is designed to tow a feet from the wavemaker.
model with a preset constant force
while allowing the model to be free Values of the uncertainties in
in surge, heave, and pitch. A each measured value were determined,
modification was made to allow the H- usually based upon the physical
P-S dynamometer to be fixed in surge limitations of the test equipment as
if desired. This allowed some data found in the device documentation or
to be taken with the model fixed in during the calibration of the device.
surge for comparison with the free to In the case of the wave height
surge data. measurements, the inaccuracy was
determined by taking the standard
Force block specifications are deviation of the measured values at a
linearity and hysteresis of ± 0.15% given wavelength, and then doubling
full scale. For our 50 pound load it (95% probability). In the case of
the calm water resistance
cell this is ± 0.075 pounds. Daily uncertainty, the values were
calibrations indicate that the load determined by the difference between
cell performance was significantly the straight line curve fit results
better, on the order of ± 0.023 of the model tests performed at each
pounds. The RVDT used to measure ballast condition, and the calm water
pitch was carefully checked and the resistance using a forced slope least
squares fit.
absolute accuracy was found to be ±
0.28% of measured value. RESULTS OF THE SEAKEEPING TESTS

The digital position encoders The experimental results

were checked in the same manner as for OSWL and NSWL models are
the RVDT and found to have an presented in figures 1 to 4. For
each model results are given for
absolute accuracy of ± 0.04% of heave amplitude, pitch amplitude,
measured value. total resistance and added
resistance. All the results are
MODEL TESTS nondimensional amplitudes defined for
regular waves and are plotted versus
Three wave probes were used wavelength divided by waterline
length ( X/ LwO.
during model testing, one centerline
forward of the model, one on each
The phase angles of the
beam. The wavemaker was set for each responses are not presented since in
run using calibrated gain settings the majority of cases, the model is
for each desired period. Period free to surge and the phase angles
settings were displayed on a 5 digit can not be precisely defined. The
counter and were set to ± 0.0005 ambiguity results because the midship
of the model moves back and forth
seconds of the desired period. For relative to the wave crest.

67
 operators. Larger error bands around
Three sets of points are shown / Lwi = 1.5 are the result of the
in each figure; one for each of the variation of the incident wave
three data reduction techniques used. amplitude. The actual wave amplitude
The x and + symbols result from varied about 0.3 inches on a mean
statistical averages of the maxima of amplitude of 2.1 inches.
the desired time record. The
analysis is started by first Pitch amplitudes corresponding
determining the peaks and troughs of to the heave amplitudes are presented
the given time record. To compute in figure 4. Again the error bands
the "mean value", the absolute values are large at the longer wavelengths
of the numerical average of the peaks due to the uncertainty in the
and the numerical average of the incident wave amplitude. There are
troughs are added to find the mean no clear trends for the differences
height which is then divided by two. between the results for the three
The same process is used to cases. The non-dimensional pitch
determined the "significant value" amplitude plotted against X/Lwi for
except that the largest 1/3 values of the light and heavy cases indicates
the peaks and troughs are used. The that the heavy results are slightly
"Fourier value" is the amplitude of higher. However, if the x-axis of the
the first nth Fourier coefficient in heavy results are shifted in
a Fourier series representation of proportion to the difference in
the time history, where n is the lengths, the two sets of data
number of complete cycles in the time collapse. This is somewhat to be
history. The zero crossings of the expected since the radius of gyration
time history are found in order to for each of the test cases was
ensure an integral number of cycles identical. While NSWL-Heavy has more
are used in the Fourier series mass and moment of inertia, it also
analysis. For a pure sinusoidal time has more waterplane area and
history the three values will be submerged volume. Apparently, these
identical. are offsetting effects. It would be
interesting to see if theory also
The differences between the predicted very small differences in
three points for a given frequency is the motion amplitudes.
a measure of the consistency of the
maxima and the linearity of the The total resistances of the
system. As can be seen from the models in regular waves for the three
figures, the mean value and the cases are plotted in figure 1. The
Fourier coefficient agree very well. nondimensional coefficients become
The significant value is usually smaller as the wavelength increases
higher since it is an average over because the wave amplitude increases
only the largest maxima. It should (recall that the tests were conducted
be noted that the significant value with a constant wave slope). The
may not be reliable in the longer error bands are largest at the short
wavelengths because only 11 or 12 wavelengths since the wave amplitudes
cycles are encountered. Thus, the had the greatest variation in that
significant value is formed by region. The significant values (the
averaging over only three or four x and + symbols) are now below the
maxima. mean and Fourier values because the
average drag is divided by the square
The heave results are shown in of the wave amplitude.
figure 3. All three sets of results
fall within the error bounds of each At each wavelength the added
other so that to the given level of drag is computed by subtracting the
accuracy it appears there is no calm water resistance, Ro, from the
difference in the heave non- total resistance in waves found in
dimensional response amplitude figure 1. The added resistance

68
 coefficients are plotted in figure 2 used in all tests. It is not clear
for the the cases of OSWL and NSWL whether a conventional heavy heave
Model. staff arrangement would have the
required accuracy for total
When comparing the non- resistance.
dimensional added resistance graphs,
the curves for the heaver model are Better wave measurements would
consistently lower. Compared at the reduce the error levels. The
same wavelength, the dimensional resistance and motions show less
added resistance of the two models is variation from run to run than the
approximately the same. wave amplitudes. In addition, Rtm*
CONCLUSIONS and Radd* use Co2 in the non-
dimensionalization so the effect of
Highly accurate total resis- any wave probe error is increased. A
tance tests in waves are possible. non-intrusive or more accurate wave
Using the H-P-S dynamometer and the probe would reduce dynamic effects.
overspeed / underspeed the
method, However, part of the variation is
total resistance in waves tests is as probably due to the fact that the
accurate as the calm water resis- wave probe is a point measurement
tance. while the model effectively
integrates the wave amplitude over
Accurate calm water resistance the entire model length. Point
values are imperative to obtain measurements inherently have a
precise added drag coefficients. greater variance.
Small inaccuracies in the calm water
resistance are a large percentage of
the added drag.

By making calm water resistance

runs at only the test speed it is
impossible to properly curve fit the
calm water data to arrive at a
sufficiently accurate calm water
resistance value. The overspeed-
underspeed test method gives addition
information in the slope of the
resistance curve.

Phase angles are a problem with

free to surge testing. Because of the
shifting model position relative to
the fixed wave probe, the phase angle
is not defined in the traditional
sense. An "equivalent linear phase
angle" analysis method has been
developed and is been tested with the
next set of tests. The free to surge
method has much less noise, and may
eventually be superior if the phase
calculations can be done.

The data indicate that testing

of these models fixed-in-surge versus
free-in-surge makes no difference in
the pitch, heave, or total resistance
results. The H-P-S dynamometer was

69
 Figure 1
OSWL and NSWL
Rtm* vs. lambda/L

111111111111111111111111111111111111
sig. Rtm*@light
1111111111111111111111111111111111
A mean Rtm*@light
1111111111111111111MININIMIN 0 Fourier Rtm*@light
E
cc sig. Rtm*@heavy
1111111111111111111111111Mill mean Rtm*@heavy

Fourier Rtm*heavy

1111111111111111.MIPI
111111111111111111111=1111111
1 0 1 .5 20
lambda/L

Figure 2
OSWL and NSWL
Radd* vs. lambda/L

IP.FAMMININIIIIII1111111111111
sig. Radd*1@light
111111121011111111E111111111111111
mean Radd*1© light
3111111111011MIIMMI11111111111 A

1111111211M114.41111111111111111111111 O FourierRadd*1 @lighty

111111111,11111.10111111111111111 sig. Radd*1@ heavy

11111111111111111111111111111111 mean Radd*1@ heavy
11111111111111111110. 1111111111111111111
FourierRadd*1@ heavy
N11111111111111111111111=0.
111111111111111111111111111101111
111111111111111111111111.11111111111111

lo 1.5
lambda/L
2.0

70
 Figure 3
OSWL and NSWL
zeta3* vs. lambda/L

11111111111111110
sig. zeta3*@ light

A mean zeta3*@ light

1111111M1111111111111111111111111
Ca

N
linfilliMIIIMEN1111111111 0 Fourier zeta3*@light

sig. zeta3*@ heavy

ifill111111111111111111111111111
mean zeta3*@ heavy

Fourier zeta3*@ heavy

111111111111111M1111111111111111
11111111111111111111111111111111111
111111111111111111111111111111
1.0 15 20
lambda/L

Figure 4
OSWL and NSWL
zeta5* vs. lambda/L

11111111111111111111211"911111111111
11111111101111111iiiimILIM sig. zeta5*@light
111111TOMMEN111111111111111111111 A mean zeta5*@light
INIS111111111111111111111111111111111111111111111
Fourier zeta5-@ light
Ki1111111111111111111111111111111111111111111
N
sig. zeta5*@ heavy
11111111111111111111111111111111111111
111111111111111111111111111111111111111111111111 mean zeta5"@ heavy
11111111111111111111111111111111111111111 Fourier zeta5"@ heavy
111111111111111111111111111111111111
111111111111111111111111111111111111111

10 1.5 20
lambda/L

71
 Run Length and Statistical Error Estimation
for Seakeeping Tests and Trials
R.D. Pierce (Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT Estimate the statistical errors and

run lengths associated with the mean value,
A seaway and the resulting responses of standard deviation, variance, autospectra
models or full-scale craft are random and RAO's of a random process.
processes. The accuracy of the final
results depends heavily on the length of the
NOTAT ION
test runs. Therefore, in the planning,
execution, and evaluation phases of a
Symbol Definition
seakeeping test or trial, run length and its
effect on the accuracy of the data must be
AR Spectrum area ratio
accounted for.
AR- Low frequency spectrum area ratio
Two basic problem areas exist:
Statistical bandwidth, rad/sec
Run length estimation prior to a
BOPT Optimum statistical bandwidth,
test or trial so that the mission can be
rad/sec
carried out efficiently, and data gathered
bPSD Autospectrum Bias error
to a desired accuracy.
Error estimation for data taken in Standard deviation bias error
the past to determine what accuracy level DF Fixed-point half-power bandwidth,
was reached. rad/sec
The run length directly affects the DFe Encountered half-power bandwidth,
level of accuracy or confidence in the rad/sec
processed results. This accuracy level is, DFE Effective half-power bandwidth,
in turn, a trade-off between the task rad/sec
priority and the statistical variability of Fm Fixed-point spectrum peak
the input medium (the seaway in this case). frequency, rad/sec
A higher priority mission will call for a FEm Effective peak frequency, rad/sec
higher data accuracy and longer run length me Fixed-point spectrum peak
while a changing sea condition will put frequency, rad/sec
constraints on this accuracy. In effect, FFT Fast Fourier Transform
items 1 and 2 must be considered in all Gravitational acceleration
missions to aid in pre-mission planning and 32.17 ft/sec2
to better understand the reliability of Data segment length for FFT, sec
data gathered from a previous mission. oPT Optimum data segment length, sec
This paper presents and describes meth- LaPTENC Optimum craft motion data segment
ods to: length, sec
Determine the autospectrum effec- Fast Fourier transform size
tive peak frequencies and half-power band- N Number of samples
widths needed to estimate run length and NDF Number of degrees of freedom
statistical error for given data samples. RAO Response Amplitude Operator
Estimate random process mean val- SPSD Autospectrum statistical error
ues, standard deviations, variance, SR Sample rate, samples per second
autospectra and Response Amplitude Opera- time, seconds
tors (RAO's). Run length, sec

73
 Encountered or craft motion run becomes large these estimates will tend to
Te
length, sec scatter about one particular value - the
Minimum run length (fixed), sec average. The scatter of these estimates
TMIN
Minimum run length (craft motion represents random error. If the average is
TMINENc different from the true value, then the
or encountered), sec
Vc Craft velocity, knots estimate is biased. Only bias and random
Value of nth data sample errors inherent to the estimation method
xn
a Fixed-to-encounter frequency are considered. Therefore, for the purposes
parameter of this discussion, craft or wave data are
Craft heading, deg assumed 100percent accurate. Bias error is
EPSD Autospectrum random error defined as the difference between the
RAO random error average estimate and its actual value,
ERAO
divided by the actual value. Random error
EA. Mean value random error is defined as the standard deviation of the
E- Standard deviation random error scatter divided by the actual value. (An
(Tx
exception to this is the mean value random
y(o)) Coherency function at frequency (i)
error where the standard deviation of the
Mean value estimate scatter is divided by the standard devia-
Standard deviation tion of the random process.) The distribu-
ax tion (probability density function) of the
Frequency, radians per second
scatter is necessary to define random error
adequately. Statistical error is the
SPECTRUM MODEL AND PARAMETRIC RELATION-
combination of bias and random errors; it
SHIPS
is defined as the square root of the sum of
A spectrum model is used to develop the the squares of the bias and random errors.
Statistical error may also be referred to
run length and statistical error equations .
A summary of the model application to pre- as mean square error.
Figure 1 presents an example of how these
trial run length estimation and post-trial
statistical error estimation is presented. errors could be viewed in practice. Three
assumptions are made in this example: (1)
Fixed-point wave to encountered wave trans-
formations, as well as multipeaked spectra, the actual parameter value is known, (2) the
average value is obtained from a larger
are related to this model.
number of trials than the 20 trials shown,
Statistical Error
inthis figure, and (3) the scatter of these
estimates is Gaussian distributed. The
A finite run length cannot determine
percent occurrence frequency is shown for
exactly the descriptive properties or
parameters of a random process (i.e., mean a Gaussian distribution; 68 percent of the
variance, estimates, for example, will fall within
values, standard deviation,
autospectrum, RAO); these parameters can plus or minus one standard deviation
interval (a confidence interval) about the
only be estimated. Deviation of the esti-
average value. More sophisticated methods
mate from the true parameter value is the
statistical error which is influenced by are required to describe the confidence
interval when the average and the standard
measurement error in the transducers,
deviation of the scatter are also estimates
signal conditioning, and recorder, and by
(See Bendat and Piersol [1].)
nonstationary conditions . Nonstationarity,
a condition which exists when the true
parameters vary with time, can make these Spectr m Model
The derivations of all statistical error
parameter estimates meaningless if the
equations and tables given herein are based
variation is significant.
Statistical error will generally con-
on an assumed shape or model for the
autospectra of the random processes. This
sist of two parts, bias error and random
shape is a parabola - a second-order curve.
error. If, for example, the same parameter
is estimated for identical trials, then as
The parameters specifying this curve are
of the sample peak frequency and half-power bandwidth.
the number of members, N,

74
 Parameter A Percent Occurrence Frequency Shown
Estimates for a Gaussian Distribution
+3as
+Zas
O
O
Actual
oo o
+6 s 'i
t i
0 o Average 68% 95% 99.7%
Average f _20_0_ojD (_2_Q -m T_
Bias O o -2as /
-3Gs
i

Trials
Fig. 1. Definition of Statistical Error

These parameters were chosen because they

may be estimated easily from an autospectrum Peak Frequency
plot. Figure 2, for example, shows how 1.169
these parameters can be estimated from a rad/sec
Standard
wave autospectrum. Similarly, multipeaked
deviation
spectra are fitted by a set of these
is 0.5475ft.
parameters (Figure 3). This model is
applied to two problems: pre-mission run
length estimation and postmission statis-
tical error estimation. For post-mission
statistical error estimation the data have
been collected, checked for stationarity,
and reduced so that autospectrum plots are Half-Power
available for the wave and motion measure- Bandwidth
ments. To evaluate the autospectrum plots 0.553
an estimate of the peak frequencies and rad/
half-power bandwidths must be made. The sec
statistical error is determined by applying
these parameters (peak frequencies and
half-power bandwidths) with the mission run
length to the statistical error equations
or tables. Because autospectrum estimates
require a trade-off between random and bias
errors, these data could be reprocessed if
an optimum trade-off is desired. Pre-
mission run length estimation is more
complicated, less exact, and requires more
assumptions than post-mission estimation. 0.0 0.64 1.28 1.92 2.56 3.2
These estimates represent an "educated
Frequency (Radians/second)
guess." First, to estimate run length it
is necessary to obtain the peak frequencies Fig. 2. Estimation of Peak Frequency and
and half-power bandwidths or the fixed- Half-Power Bandwidth for a Single Peak
point wave spectrum expected to exist Spectrum (wave autospectrum).
during a mission. A real-time spectrum
analyzer would be invaluable for this speed and heading), these peak frequencies
purpose. Then, depending on the craft and half-power bandwidths must be converted
conditions planned for this mission (craft into wave encounter peak frequencies and

75
 for
(ß is 180 deg. for headseas, + 90 deg.
beam seas, + 45 deg. for quartering seas,
and zero deg. for following seas).
The spreading of the half-power band-
rad/sec width is approximated by the Jacobian,
3Fm DF
Fme D Fe

DFe = DF 11+2 a Fml (3)

The amount of spreading is assumed to be

constant over the peak half-power band-
width. These fixed-point wave to encoun-
-0.754-"1
rad/sec
0.314 rad/sec tered wave transformations are defined as
different factors. Values for the factor
1
from Equation (1), are the mean
11+ a Fml '

value run time multiplication factors.

0'314
rad/sec 1
i

i Values for the factor 1 + 2 a Fmi fr°m

Equation (3), are the Standard Deviation
0.0 0.64 1.28 1.92 2.56 3.2 and Autospectrum Run Time Multiplication
Frequency (Radians/second) Factors.
Fig. 3. Multipeaked Spectrum (wave
Example 1 - Estimating the Wave Encounter
autospectrum)
Peak Frequency and Half-Power Bandwidth
half-power bandwidths (those "seen" by the from Single-Peaked Autospectrum.
Figure 2 presents an example of a typical
craft) . Here the assumption is made that the
single-peaked wave autospectrum. Fromthe
wave encounter spectrum represents the
plot, estimates are made of the frequency
worst case craft motion spectrum (narrowest
half-power bandwidth and lowest peak fre- of the peak power and the bandwidth
at half-power. The
quency). Run lengths for the craft motion (frequency range)
data are calculated using these encounter
autospectrum is in the fixed-point fre-
quency domain and must be converted to
wave spectrum parameters in the statistical
error equations. The run length estimates encounter frequency. From Figure 2 the peak
power freq-uency Fm is 1.169 rad/sec and the
will vary depending on the choice of halfpower bandwidth DF is 0.553 rad/sec.
tolerable statistical error.
Run length estimation requires the Equations (1), (2), and (3) are used to
transformation of fixed-point wave spec- convert to encounter frequency. For head
seas and a craft speed of 60 knots, the wave
trum peak frequencies, Fm, and half-power
bandwidths, DF, into wave encounter peak encounter peak frequency is
and half-power bandwidths Fme = 5.47 rad/sec
frequencies F
encounter peak frequency is and the wave encounter bandwidth is
DFe- The wave
DFe = 4.63 rad/sec
approximated by the Doppler shift equation
This concludes Example 1.
derived from the linear theory of gravity
waves.
Multipeaked spectra can occur when
Fme = + a Fml (1)
estimating pre-mission run lengths or post-
Where a is related to craft conditions by: mission statistical error. For such spec-
a=-( 1.6888 Vc cos(S) )/ g (2) tra, estimates are necessary for the peak
frequency, the half-power bandwidth, and
where: Vc is craft speed in knots, g is
the spectrum area ratio for each peak. This
gravitational constant (32.174 ft/sec), p
area ratio is the area under each peak of
is craft heading into the waves in degrees

76
 the spectrum (approximately equal to the (1) are 0.505 for the first peak (0.754 rad/
half-power bandwidth times the spectrum sec) and 0.357 for the second peak (1.382
peak value) divided by the area under the rad/sec). The encounter peak frequencies
whole spectrum. Therefore, an effective are found by dividing the factors into the
peak frequency F be estimated by: peak frequencies giving the first encounter
1
peak frequency of 1.493 rad/sec, and the
FEm = second peak of 3.87 rad/sec. For the
AR1+AR2+ encounter half-power bandwidths, equation
1 FL FL (3) is used. The conversion factors are
An effective half-power bandwidth DFE can 0.338 for the first peak, and 0.218 for the
be estimated by: second peak. The encounter half-power
1
bandwidths are found by dividing this
DFE - 2
,2 factor into the half-power bandwidths:
AR1 0.929 rad/sec for the first bandwidth and
DE). DF2 1.442 rad/sec for the second bandwidth.
Fml, Fm2,... and DF1, DF2,... are the peak Next, these encounter peaks and bandwidths
frequencies and half-power bandwidths for are applied to Equations (4) and (5), as
each peak, and AR", Ap2, ... are the spectrum shown by Examples 1 and 2. The effective
area ratios. peak frequency is 2.27 rad/sec and the
effective half-power bandwidth is 2.338
Example 2 - Calculating the Effective rad/sec. The area ratios are not converted;
Fixed-Point Peak Frequency and Half-Power they remain at one-third for the first peak
Bandwidth from a Double-Peaked Autospectrum. and two-thirds for the second peak. This
Figure 3 presents an example of a double- concludes Example 3.
peaked autospectrum. First, the individual
peak frequencies, half-power bandwidths, PARAMETER STATISTICAL ERROR AND RUN LENGTH
and spectrum area ratios are estimated: The ESTIMATION
first peak is at 0.754 rad/sec with a 0.314
rad/sec bandwidth and a one-third area The previous defined statistical error
ratio. The second peak is at 1.382 rad/sec and detailed methods for calculating param-
with a 0.314 rad/sec bandwidth and a two- eters (effective peak frequencies and half-
third area ratio. Next, the effective peak power bandwidths) necessary for calculat-
frequency using Equation (4) is found: ing these statistical errors. In this
F Em = 1.03 rad/sec section, the methods for estimating the run
The effective half-power bandwidth is found length and resulting statistical errors for
using Equation (5): mean values, standard deviations, vari-
DFE = 0.565 rad/sec ance, autospectra, and RAO' s are presented.
This concludes Example 2. Although the derivations of the statis-
tical error and run length equations are not
Example 3 - Calculating the Effective included in this report, some general
Encounter Peak Frequency and Half-Power comments about the use of these equations
Bandwidth from a Double-Peaked Fixed-Point are necessary. The approximations made in
wave Autospect rum. these derivations generally bias or weight
Using the wave autospectrum given in these error and run length equations in a
Figure 3, assume that the craft is traveling conservative manner (slightly larger error
at 35 knots in these seas at a 135 deg (bow) and slightly longer run length). Estimated
angle. The peak frequencies, half-power parameters are used to replace actual
bandwidths, and area ratios for this parameters in these equations. Therefore,
spectrum are given in Example 2. The an estimated error and run length for the
encounter peak frequencies and half-power parameters will be obtained.
bandwidths are calculated using equations
(1), (2), and (3). At 35 knots, the factors Mean Value of a Random Process
for peak frequency conversion from equation Mean value estimates, /2, are made using

77
 random error from Equation (8) is:
the following method:
N EA.= 0.02
fix =1 xn (6) or 2 percent. For this run length, the area
N
n=1 ratio AR, iS approximately zero (10n/T equals
where xn is the value of the nth member, and 0.355 rad/sec). The apparent area below
N is the total number of members. These this frequency is due to mechanical offset
estimates are unbiased, so statistical in the plotter hardware. Since the mean
error is the same as random error. value random error is normalized by the
random processes standard deviation, the
Mean Value Statistical error Estimates . random error can be expressed in physical
The scatter in the mean value estimate is units by multiplying by the standard
approximately Gaussian distributed even if deviation. For Figure 2, this standard
the random process x is not Gaussian (See deviation is 0.5475 ft, so the estimated
Bendat and Piersol (1).) The estimate's mean value random error is 0.011 ft. This
concludes Example 4.
random error, efix, is given by:

AR, 1-AR, Mean Value Run Length Estimates

E- 2 2 (7) Equation (8) can be rearranged to give run
10 T Fm
length in seconds.
where T is run length in seconds, Fm is the 2
effective peak frequency in rad/sec from T =
Fm (9)
the autospectrum of the random process, and E-Y X

AR, is the spectrum low frequency area Given the peak frequency of the expected
ratio. The spectrum low frequency area fixed-point wave spectrum andthe amount of
ratio is estimated by dividing the spectrum tolerable random error (the statistical
area from zero to 10n/T rad/sec by the error estimate is unbiased), a mean value
spectrum's total area. For a 39-sec run run length estimate can be obtained using
length, this ratio is estimated at 0.02 for Equation (9). This run length estimate
the spectrum in Figure 2 (area from zero to would apply only to mean value estimates
from the fixed-point wave data. Craft
0.804 rad/sec divided by the total area).
obtained by
Generally, longer runs will be made; motion run length estimates are
translating this peak frequency to a craft
therefore, AR, is zero, and the mean value
or wave encounter peak frequency. Applying
random error is given by:
Equation (1) to Equation (9) gives:
2
Ei; = (8) 2
J TFm Te= E- F 11+aFml (10)
Given an autospectrum estimate charac- m
YX

terized by the effective peak frequency and where Te is the craft motion run length
half-power bandwidth, mean value random estimate. (Only positive wave encounter
error is estimated from Equation (8) If frequencies are seen by the craft.) From
this spectrum has multiple peaks, Equation equation (9)
(4) must be used to estimate the effective 1
peak freq-uency (Example 1). This effective Te =T
peak frequency is then applied to Equa-
,11+a Fml
tion (8). The craft motion run length estimate is the
fixed-point run length multiplied by the
Example 4 - Calculating the Mean Value mean value run time multiplication factor.
Random Error.
If an 88.4 sec run length is used for mean Example 5-Mean Value Run Length Estimates
value estimates from the wave data corre- for Wave and Craft Motion Data.
sponding to the spectrum in Figure 2 (1.169 If the craft test plan calls for a 10-
rad/sec peak frequency) , the mean value knot task in following seas and the wave

78
 spectrum given in Figure 2 (1.169 rad/sec In practice, this bias error may be ignored
peak frequency) is expected to exist during because it will probably be small compared
this mission, then the wave data run length to the random error of the estimate. For
is estimated using equation and the (9)
example, either eAx or e'd. must be 14
craft motion run length is estimated using
equation (11). For 1 percent mean value percent before either value produces a 1
random error, the run length estimated for percent bias error. If this bias error is
the wave data is 176.8 sec. The mean value sufficiently small, the statistical error
run time multiplication factor is 2.489, so is the same as random error.
the craft run length estimated for this task
is 443 sec. This concludes Example 5. Example 6 - Calculating the Standard Devia-
tion Random Error
Standard Deviation of a R-ndom Process If a 1657-sec run length is used for
Standard deviation estimates are made standard deviation estimates from the wave
using the following method: spectrum given in Figure 2 (1.169 rad/sec
peak frequency and 0.553 rad/sec half-power
_1 v
N
fix )2
(12)
bandwidth), the standard deviation random
error from Equation (13) is:
N
n=1 e-
ax = 0.04
(A 1/(N-1) factor will unbias this estimate
or 4percent. Since the standard deviation
from mean value random error if the xn random error is a percentage of the random
samples are statistically independent -- a processes standard deviation, 0.5475 ft
rare condition for wave and motion data.) from Figure 2. The estimated random error
Standard Deviation Statistical Error of this estimate is 0.022 ft. This
Estimates. The random error of the standard
concludes Example 6.
deviation estimate is given by:
3g" Standard Deviation Run Length Esti-
ea = (13) mates. Run length is given by:
1/5-a DF T
where DF is the half-power bandwidth in rad/ T=
sec from the autospectrum of the random DF e? (15)
ax
process x, which is assumed to be Gaussian.
T is the run length in seconds . The scatter Given the peak frequency and half-power
of this estimate is approximately Gaussian, bandwidth of the expected fixed-point wave
spectrum as well as the amount of tolerable
provided e-ax is less than 0.13. random error (same as statistical error
Given an autospectrum the standard since the estimate is assumed to be
deviation random error can be estimated unbiased), a standard deviation run length
directly from Equation (13). If this estimate can be obtained using Equation
spectrum has multiple peaks, Equation (5) (15). This run length estimate would apply
must be used to estimate the effective half- only to standard deviation estimates from
power bandwidth (Example 1). This band- the fixed-point wave data. Craft motion run
width can then be applied to Equation (13). length estimates are obtained by translating
In addition to the random error, a bias this half-power bandwidth to a craft or wave
error source exists which will reduce the encounter half-power bandwidth. Applying
estimated value of the standard deviation Equation (3) to Equation (15) gives:
from the true value. This bias error b-dx
represents the combined effects of the Te -
random error of the estimated value and the 2
E- DFIl + 2 aFml (16)
ax
random error of the mean value. The bias
error is given by:
2
where Te is the craft motion run length
b - = + E-a (14) estimate. This run length estimate is
a. 2 )x

related to the fixed-point wave data run

79
length estimate by: is approximately chi-square distributed
with the number of degrees of freedom NDF
Te-T (17) given by:
11+21aFn,l
1
The factor that relates these two times is NDF = 2
2 eA.
the standard deviation and autospectrum run -x
time multiplication factor. (NDF is 200 for e = 0.05) . These variance
estimates are biased downward by the square
Example 7 - Standard Deviation Run Length of the mean value estimate random error -
Estimates for Wave and Craft Motion Data.
a 10 percent mean value random error yields
If the craft test plan calls for a 40 knot
(45 deg) and the wave a 1 percent bias error.
task in bow seas
spectrum given in Figure 2 (1.169 rad/sec
Autospectrum of a Random Process
peak frequency and 0.553 rad/sec halfpower
The autospectrum statistical error and
bandwidth) is expected to exist during this based on the
run length estimates are
mission, equations (15) and (17) canbe used autospectrum estimation method which uses
to estimate the wave and craft motion data
overlapped Fast Fourier Transform (FFT)
run lengths.
processing of windowed data segments.
For 5-percent standard deviation random
Nuttall [2] and Welch [3] describe the
error, the run length estimated for the wave details of this method. A 50-percent
data is 964 sec (from equation (15)). The segment overlap and a full cosine data
autospectrum and standard deviation run
window are the reduction parameters speci-
time multiplication factor is 0.229, so
fied for this method. In this method, data
using equation (17) the craft run length segments of length L seconds (L equals FFT
estimated for this task is 221 sec. The mean
size divided by sample rate) are transformed
value random error for these conditions can
by the FFT algorithm. At each freq-uency the
now be estimated using equation (8) for the absolute value squared of the transform
wave data, value is calculated, and these values from
= 0.002 or 0.2 percent each transform are averaged to form each
and, equation (8) with equation (1) for the autospectrum estimate. The data segments
motion data, are windowed or weighted by a full cosine
or 0.3 percent
= 0.003
function (
1-I-CO5 -t-r for 0 < t < L,
This concludes Example 7.
zero otherwise). These data segments are
If a multipeaked fixed-point wave spec- also overlapped by 50 percent - the input
trum is expected during a mission, the data (time history) are shifted by L/2
effective half-power bandwidth (Example 2) seconds before a new segment is taken.
is used for estimating the fixed-point wave This method of autospectrum (or cross-
spectrum run length. The autospectrum and spectrum) estimation is used because it
standard deviation run time multiplication provides a good trade-off between cost,
factor cannot be used to estimate the craft accuracy, and flexibility. It can accommo-
motion run lengths because the factor is date arbitrarily long run lengths. The full
based on peak frequency estimates. In- cosine data window is near optimum for
stead, the effective encounter half-power producing minimum bias error (Nuttall [4] ) .

bandwidth from Example 3 is applied to The spectral window or filter associated

Equation (15) to obtain craft motion run with the cosine data window has a 1/0)6
asymptotic decay -a necessary criteria for
length.
spectral moment estimation, or double
Variance Estimates differentiation or double integration in
Variance estimates (standard deviation the freq-uency domain . Cross-spectra maybe
squared) are occasionally used. The vari- obtained without a separate processing
ance random error is double the error for phase. Computer memory is conserved and the
standard deviation estimates. The scatter FFT algorithm is cost efficient. The only

80
significant disadvantage of this method is actual spectrum peak are reduced from the
that the random error is 1.043 times greater actual value by this bias error. Bias error,
than the minimum possible random error. (If however, is only specified at the spectrum
a 10 percent random error is obtained using peak - the region of maximum bias - and is
autocorrelation techniques to estimate why the parabolic model (peak frequency and
autospectra, a 10.43 percent random error half-power bandwidth) should fit the nar-
is obtained using the FFT method, provided rowest local detail desired in the
the data reduction bandwidth is the same in autospectrurn estimate. The derivation of
both cases.) bias error does not assume a Gaussian random
process. The random error equation is
Autospectrum Statistical Error Estimates assumed to be valid from any point on the
Given an autospectrum estimate using spectrum. The derivation of this random
this FFT method, the statistical error can error equation does make the assumption
be estimated. If the spectrum is multipeaked, that bias error is zero or very small and
this error is estimated at each peak. Note that the random process is Gaussian. The
that statistical error is primarily a scatter about the actual spectrum (random
convenient summary of the estimate's qual- error distribution) is approximately chi-
ity. Bias error and random error should be square distributed with the number of
stated separately, because the meaning or degrees of freedom NDF given by:
nature of the error is more easily under-
stood. Autospectrum estimates from this
2
FFT method are biased by the data reduction NDF = ,2 (23)
bandwidth where the bandwidth measure used cP SD
in this section is the statistical band-
width associated with the full cosine data Example 8 - Calculating Autospectrum Sta-
window. This bias error bpsD at the tistical Error.
autospectrum peak is: The roll angle spectrum given in Figure
4 is used in this example. This spectrum
bpsD =0.154(4 peaks at 0.565 rad/sec with a 0.107 rad/sec
DF half-power bandwidth. A data length of 2100
where DF is the half-power bandwidth in rad/ sec was collected. The above FFT processing
sec associated with this spectrum peak, and method was used to produce this autospectrum
B is the statistical bandwidth in rad/sec: estimate; the statistical bandwidth was
2.079 0.075 rad/sec. The bias error is:
B= 27r
bPSD = 0.076
where L, the length in seconds of the or 7.6 percent . The random error is:
individual data segment processed by this EPSD = 0.208
FFT method, is defined by: or 20.8percent. The statistical error is:

SR
L= (2 0 )
SPSD = 0.221
or 22.1 percent. The number of degrees of
where M is FFT size, and SR is sample rate freedom is:
in samples per second. NDF = 46
The autospectrum estimate random error This concludes Example 8.
EPSD S
Eun Length Estimates . The
1
EPSD- autosprectrum run length estimates are made
0.92 by finding an optimum statistical bandwidth
B T (21)
11 27r 1302T that minimizes run length given a fixed
where T is the run length in seconds. The statistical error SpsD This optimum band-
.

statistical error SpsD :

width is:

lb-P SD
SP SD = (22)
BOP T = 1.704 ITTp7) DF (24)
Autospectrum estimates at or near the

81
 gives:
0.7973
TMINENC = 5/2 DF
rad/ SPSD 11+2aFml
sec
where TMINENC is the craft motion run length
estimate. The run length for the fixed
point wave ;EN is obtained from Equation
(25). TIAIN is related to TMINENC by the
standard deviation and autospectrum run
time multiplication factor. For multipeaked
0.107 rad/sec
fixed-point wave data, these methods must
be applied to all peaks, and the longest run
length is chosen.
If run length is fixed (the data are "in
hand"), the optimum statistical bandwidth
is also the bandwidth that will minimize
statistical error. Bias and random error
still have the same relationship to statis-
tical error. This error is determined
0.63 1.26 1.88 2.51 using:
0.0

Fig. 4.
Frequency (Radians/second)
Roll Angle Spectrum
and
SPSD
DF
42r 0.7973)2/5
T

At this bandwidth, the minimum run length

)1/5
11.458 DF4
MIN i s- -
BOPT= 2z (30)

0.7973
TMIN = DF
S5/2 (25) Example 9- Calculating the Optimum Statis-
PSD 2n tical Bandwidth for Minimum Statistical
Error Given a Fixed Run Length.
For this optimum condition, bias error and The roll spectrum given in Figure 4 was
random error are related to statistical not processed for minimum statistical
error by: error. Using equation (30) with a 2100-sec
bpsD SpsD (26) run length and a 0.107 rad/sec half-power
and bandwidth (from Example 8), the optimum
statistical bandwidth is:
SpsD (27)
EPSDV BOPT = 0.0852 rad/sec
The minimum statistical error (Equation 29)
When autospectrum run length estimates is:
are made for craft motion data, the peak S PSD = 0.218
frequency and half-power bandwidth of the or 21.8 percent. The data were therefore
expected fixed-point wave spectrum are used processed close to the optimum condition,
as well as the amount of tolerable statis- since the error obtained in Example 8 was
tical error. These values are then applied 22.1percent. From equations (26) and (27),
to Equation (25). This run length estimate the bias and random errors are:
would apply only to autospectrum estimates
from the fixed-point wave data. Craft
bPSD = 0.098 or 9.8 percent
motion run length estimates are obtained by and
translating this half-power bandwidth to a
EPSD = 0.195 or 19.5 percent.
craft or wave encounter halfpower bandwidth .
Applying equation (3) to equation (25)
This concludes Example 9.

82
 Fast Fourier Transform (FFT) Data Seg- motion data are actually obtained, then
ment Length. During data reduction, the from equation (29), the minimum possible
optimum data segment length LopT that will statistical error is 15 percent. The
be a segment of the full data run is given optimum segment length for the wave data is
from Equation (19) by: 63.0 sec and for the motion data 10.96 sec.
2.079 At the 3198 and 557 sec run lengths, the
LOPT = 27r
OPT
(31) standard deviation random errors are es-
Where is given by equation (24).
BopT
timated using Eq-uation (13) and the results
from Examples 2 and 3:
Overlapping of segments is used until all Wave data (effective half-power
data points are processed. bandwidth from Example 2)
Equation (31) is for fixed point spec-
tra. For wave encounter or craft motion e& = 0.027 or 2.7 percent
data the segment length is given by: Motion data (effective encounter
half-power bandwidth from Example 3)
27r 2.079 e&x = 0.032 or 3.2 percent
LOPT ENC (32)
1.704 DF I 1 + 2 a F, This concludes Example 10.
If the actual run length is different from Response Atop_litude Operator of a Random
the specified or estimated run length, the Process
minimum statistical error value in equation Response Amplitude Operator (RAO) esti-
(25) is used to find the necessary segment mates are made by dividing the response
length. spectra by the input spectrum - motion
spectra divided by the wave encounter
Example 10 - Minimum Autospectrum Run spectrum. To obtain this encountered wave
Length Estimate for Wave and Craft Motion spectrum, data from an over-the-bow wave
Data. height sensor must be used. In either case,
If the craft test plan calls for a 60- the methods for estimating statistical
knot task in head seas and the double-peaked error or run length are based on the
wave spectrum given in Figure 3 is expected procedures already established for
to exist during this mission (the half- autospectrum estimates. If fixed-point
power bandwidth for both peaks is 0.314 rad/ wave data are used, the run lengths for the
sec; the first peak is at 0.754 rad/sec, and wave and craft motion data must be chosen
the second peak is at 1.382 rad/sec), to have the same amount of autospectrum
equations (25) and (28) are usedto estimate random error. (This condition is met if the
the wave and craft motion data run lengths. run lengths are specified using the methods
For 12-percent autospectrum statistical already presented for autospectra.) Since
error, the run length estimated for the wave the fixed-point wave data are statistically
data is 3198 sec for either peak. The independent of the craft motion data, the
autospectrum and standard deviation run RAO random error is 1.414 times the
time multiplication factor is 0.174 for the autospectrum random error. This random
0.754 rad/sec peak and this factor is 0.103 error is F-distributed with the same
for the second peak. Since the first peak degrees of freedom as the autospectrum chi-
has the larger factor, the craft run length square distribution. The bias error is
estimated for this task is 557 sec. The assumed to be 1.414 times that of the
optimum segment length (or a length close autospectrum estimates, so the statistical
to it) must be used during data reduction. error is also 1.414 times the autospectrum
For the wave data, this length is found statistical error. When evaluating the
using equation (31) to be 70.4 sec. Using results from a seakeeping mission, the RAO
equation (32), the segment length for the statistical error is estimated by determin-
motion data is 12.25 sec. (The 70.4-sec ing the autospectrum error and applying the
length is multiplied by the 0.174 factor.) 1.414 factor. When estimating run lengths,
If 1830 sec of wave data and 318 sec of the autospectrurn methods are used; however;
the tolerable level of statistical error
83
would be based on the amount of error
desired for the RAO estimates. If an over- REFERENCES
the-bow wave height sensor is used, the wave
sensor data are probably correlated with Bendat, J.S. and A.G. Piersol,
the motion data. In the frequency domain, "Random Data: Analysis and Measurement
correlation is measured by the coherency Procedures," Wiley-Interscience, New York
function obtained fromthe cross-spectrum. (1971).
At a particular frequency the RAO random Nuttall, A.H., "Spectral Estimation
error decreases as the coherency increases. by Means of Overlapped Fast Fourier Trans-
This relationship is given by: form Processing of Windowed Data," Naval
Underwater Systems Center Report 4169 (Oct
(33)
ERAO= EPSD 2 ( 1 - 7 (W) ) 1971).
where ERA° is RAO random error, Epsp is the Welch, P.D., "The Use of Fast Fourier
autospectrum random error, and y(w) is the Transform for the Estimation of Power
coherency function at a particular fre- Spectra: A Method Based on Time Averaging
quency w. When evaluating the results from over Short, Modified Periodograms," IEEE
a seakeepingraission, this relationship can Trans. liudio Electroacoustics, Vol. AU-15
be used to estimate the BAO random error (Jun 1967) pp. 70-73.
provided coherency estimates are available. Nuttall, A.H., "Minimum Bias Windows
The bias error is assumed to be 1.414 times for Spectral Estimation by Means of Over-
the autospectrum bias error. These errors lapped Fast Fourier Transform Processing,"
can then be combined (square root of the sum Naval Underwater Systems Center Report
of the squares) to give the statistical 4153, (Apr 1973).
error. When estimating run lengths, zero
coherence is assumed. This results in the
same procedure used for RAO estimates from
fixed-point wave data. Even if experience
is developed for a particular craft,
coherency should not be assumed to be
greater than 0.5, because at this value the
autospectra and the RAO's would have the
same random error. Autospectra are still
required to document the motion experienced
during the mission. Transfer function
estimates (square root of the BAO esti-
mates) maybe made occasionally. The random
error of these estimates is one-half the RAO
random error. No distribution is assumed
for its random error. Its bias error is
assumed to be one-half the BAO bias error.
A special application of these BAO error and
run length methods can exist. In preparing
for a seakeeping mission, the comparison of
two wave height transducers may be neces-
sary (both fixed-point, one fixed and one
over-the-bow, orboth over-the-bow). If the
method of comparison is the evaluation of
the BAO between these measurements, then
the BAO statistical error equations can be
used when planning this test or evaluating
its results. Statistical error could
easily be greater than the measurement
error.
84
 SESSION I-C

OCEAN ENGINEERING
 The Ocean Engineering Committee Report
J.I. Collins (Arctec Offshore Corporation, USA)

It would appear that the current level of Ocean Engineering testing in North America is
lower than usual. Indeed, it may be concluded that the world's market is at a low ebb.
The offshore oil industry, probably the strongest supporter of ocean engineering scale
model testing has suffered from continual declines in research and development budgets
for some time.

Further evidence of the general depression in ocean engineering scale model testing is
provided by the presentation of only one paper in the "Ocean Engineering" session of the
23rd ATTO meeting in1992 and reports from a very limited number of facilities
performing ocean engineering testing.

Reports from three test facilities have been received: Chicago Bridge and Iron Technical
Services Company (CBITSC), Offshore TechnologyResearch Center (OTRC) at Texas A &
M ( a new facility, described in Session IIIB) and Arctec Offshore Corporation (A0C).

CBITSC have performed testing on a radio controlled submarine, shear current loads on a
moored tanker and testing of a damping system for a TLP. The submarine (scale 1:18)
included an on-board data acquisition system and the measurement of lift and drag loads
on all planes and rudders and six degrees of freedom loads on the propeller shaft. CBITSC
have also added a plunger type wavemaker to permit the simulation of higher frequency
wave components.

OTRC has conducted testing on twin flexible risers in waves and some commercial
projects for the offshore oil industry. This is a "new" basin offering multi-directional
seas and a deep pit. The risers were selected to represent TLP risers and were tested to
examine some of the aspects of potential interference phenomena. An innovative testing
technique using distorted scales (horizontal and vertical having different values) was
used.

87
At AOC the general level of ocean engineering testing has been lower than usual. A few
single point mooring concepts and a spread moored spar buoy have been tested. A Joint
Industry Project involving the towing of two unequal sized cylinders in which drag and
lift foreces were measured. The cylinder study included towing tests at high Reynolds
numbers with some cases incorporating forced oscillations as the cylinders were towed.

At CBITSC and AOC initiatives have been taken to convert to 80386 and 80486 data
acquisition systems. The lower cost of such microprocessor based equipment and
associated peripherals and the general ease and flexibility of software develoment
present a strong case for such conversions. In addition, clients generally request copies
of test data on PC floppies for their own subsequent analysis, data bases and correlation
with numerical models.

OTRC and AOC have added non-contact motion sensing capability based on optical tracking.
Both facilities are also investigating the use of ultra sonic sonar and infrared systems
for non-contact motion sensing.
 Comparison of Wave Reflection Equations
with Experimental Wave Basin Data
D.C. Cotter, S.K. Chalcrabarti (CB&I Technical Services, Inc., USA)

ABSTRACT

Numerous methods for calculating than 10 percent and preferably less than
wave reflection have been presented in 5 percent [Jamiessen and Mansard (1)]
the literature. These methods utilize over the range of wave heights and peri-
the recorded wave heights from two or
three probes and, in some cases, the
ods that the basin is capable of pro-
phase ducing. Since the wave absorber takes up
relationship between the wave
probes. The valuable space from the wave basin, the
present paper discusses
three of the calculation methods and the design of an efficient wave absorber is
always a challenge. If there is only
difficulties encountered in applying the
methods to "real" wave tank conditions. limited space available for a wave ab-
Results sorber, efficient wave absorption may be
are presented for the wave a problem.
reflection from the beach of a two dimen-
sional wave tank, and recommendations are
The most commonly used wave absorb-
presented concerning the most appropriate
method. ers are beaches of constant slope which
extend to the bottom of the basin. They
INTRODUCTION are constructed of concrete, sand, gravel
or stones. The slope of these beaches
It is
must be mild for efficient wave energy
important to simulate open
ocean conditions in a laboratory envi- absorption. Typical slopes are from 1:6
ronment as closely as possible. One of to 1:10 [Ouellet and Datta (2)]. A beach
the factors that contaminate the gener- with variable slope can reduce the total
ated wave at the test site is the reflec- length of the beach. A parabolic slope
tion of the incident waves from the end is often used along with surface rough-
walls of the wave basin. Even if the ness and porous materials. The position
basin is long,
of the parabolic beach should be adjust-
during the course of a able with water depth in order to main-
test run of reasonable duration, the mod-
els placed in the tain its efficiency.
center of the basin
experience reflection from the end wall.
In order to dissipate the wave energy and Another concept for laboratory
minimize the problem of wave reflection, beaches is a progressive wave absorber.
wave absorbers are generally installed at Jamieson and Mansard (1) constructed an
the far end of the wave basin.
upright wave absorber consisting of mul-
The tiple rows of perforated vertical metal
amount of wave reflection is quantified
as a wave reflection coefficient sheets. The porosity of the sheets de-
(Cr) creased towards the rear of the absorber.
which is defined as the ratio of
re- This provided an efficient wave absorber
flected wave height to the incident wave
height. within a relatively short
For a solid vertical wall, the length with
about 5 percent wave reflection.
reflection coefficient is 100 percent,
i.e., the magnitude of the reflected wave
is equal to that of the incident wave. PURPOSE OF THIS STUDY
For an efficient beach, the reflection
In a recent paper, Isaacson (3) dis-
coefficient should be consistently less
cussed three methods of computing wave
89
reflection coefficients from the measure- linear wave theory. This is the area
ment of wave profiles. These methods that is investigated here.
utilize measurements at several fixed
probes (two or three) in line with the COMPUTATION OF REFLECTION COEFFICIENT
direction of wave propagation. The waves
are considered to be regular and sinu- There are several methods available
soidal so that their parameters can be to compute the reflection coefficient.
computed by the linear theory. As stated earlier, the reflection coeffi-
cient is defined as the ratio of the
In computing the reflection coeffi- amplitude of reflected wave to the corre-
cient from the measured data, one is sponding amplitude of the incident wave.
confronted with three unknowns: the Note that for a regular wave, the fre-
incident wave height, the reflection co- quency of the reflected wave is the same
efficient and the phase of the reflected as that of the incident wave. The
wave train. The number of known indepen- reflected wave creates a standing wave
dent variables depend on the number of system in the tank having a node and an
fixed probes used. In particular, for n antinode. Therefore, the reflection
probes, 2n - 1 equations may be written. coefficient may be computed from the wave
Thus, a minimum of two fixed probes are heights measured by a single traversing
required for the three unknowns. In this wave probe. If the wave probe is moved
case, the known quantities are the two along the direction of the wave propaga-
wave heights at the two probes and the tion, then the probe will read a maximum
phase difference between the probes. If wave height at the antinode and a minimum
three probes are used, then the three height at the node. Then the wave re-
heights at the three probes and two flection coefficient is obtained from
phases may be utilized to set up a least
square technique. A third method simply Hmax Hmin (1)
uses the wave heights measured at three cr
locations to determine the three un- Hmax Hmin
knowns. Isaacson (3) formulated these
An alternate approach employs two
three methods and u;ed numerical simula-
fixed wave probes instead of a traveling
tion to generate wave profiles. In order wave probe. In this case, the two probes
to determine the accuracy of the methods, are located at distances of L/4 and L/2
errors were introduced in each variable from the reflecting structure, where L is
including the wave heights, phase angles the wave length. This assumes that the
and nondimensional distances. The con- node and antinode as well as the wave
clusions reached in Isaacson's study are in many in-
length are known. However,
as one would suspect: the method of
stances, such as for a perforated or
three heights without regard to the phase sloped structure (beaches), the phase of
angles is the least accurate, while the the reflected wave is not known apriori.
method of three probes which use all five
measurements is the most accurate. It is In general, there are three unknown
clear that the least square method will parameters which describe the reflection
improve with a larger number (than three) of a regular wave: the incident wave
of measurements. In all these calcula- height, the reflected wave height and the
tions, however, sinusoidal wave forms phase difference between the two. Assum-
have been assumed. ing ai (= H/2) and ar to be the ampli-
tudes of the incident and reflected waves
It is, therefore, interesting to
(Cr = ar/ai) and e to be the phase dif-
study the accuracy of measurement with ference, the free surface elevation,
actual wave tank data. In this case, the n, for a wave probe at a location given
profiles are generally not sinusoidal. by xn is
Moreover, the wave lengths may be differ-
ent from the computed values based on the

90
 nn = ai cos(kxn - wt)
ai exp(ikxn) + ar exp[i(kxn -

+ ar cos(-kxn - wt + E) (2) - An exp[i(01 + .5n)] (8)

n = 1,2,.... n= 1,2
in which k and w are the wave number and The incident and reflected wave ampli-
wave frequency given by the linear wave tudes are derived from this set of equa-
theory for a given wave period, T, and a tions.
water depth, d. Assuming the location of
the first probe to be xi and writing the 1 2 2
distance between the nth probe and first ai - [Ai + A2 - 2A1A2
21sin A21
probe to be 8n, we have
cos(A2 62)]1/2
xn - xi + bn or (3)
1 2 2
kxn ar - [Ai + A2 + 2A1A2
kxi + An (4) 2Isin A21

where Au - k8n. Then, cos(A2 - 82)]1/2

the profile for
the nth probe is given by
The phase difference between the two is
tin = ai cos(kxi + An - wt) + written in terms of A = 2kx1 - f,
ar cos(kx1 + An + wt - E) (5)
2 2 2
Ai - ai - ar
In complex notation, cos A -
2aiar
r7n (ai exp(ikxn)
+ ar exp[i(kxn - E)]) exp(-iwt) From Eq. 10, the method fails for A2
0, r, 2r...etc. which corresponds
(6) to a
(ai exp[i(kx1 + An)] probe spacing of multiples of half wave
lengths, Therefore, A2 should be out-
+ ar exp[i(kx1 + An - e)) exp(-iwt) side the range of 0.4L < A2 < 0.6L
The recorded elevation at the nth probe [Goda and Suzuki (4)].
may be written in terms of an amplitude,
An, and a phase, 8n, The derivation of the incident and
relative to the reflected wave is based on the waves
first wave probe record (i.e., Si - 0)
is given as being represented by the sine function
(linear theory). Any nonlinearity in the
wave will produce error in the measured
n(m) An cos(wt - 01 - 6,) amplitudes and phase for the two probe
(7) arrangement. This error may be reduced
where 01 is the phase angle correspond- if additional probes are used so that a
ing to the first wave probe. An equiva- least square estimation is possible.
lent expression in the complex domain may
be written as in Eq. 6. Method #2 - Three Fixed Probes
If a three probe arrangement
Method #1 - Two Fixed Probes [Mansard and Funke (5)] is used, then
Using the expressions five measured quantities, namely, three
for the as-
sumed and actual elevations in Eqs. 5 and wave amplitudes (An, n = 1,2,3) and two
7 and using the complex equivalent form, phase angles (82 and 83) may be used
the following set of equations may be to derive the three unknowns.
written [Goda and Suzuki (4), Isaacson
(3)] For convenience, let us write Eqs. 6
and 7 as follows:

91
 tin [bi exp(iAn) 3
S3 - E An exp(i(8, + LT)]
+ br exp(-i)] exp(-iwt) (12) n=1

(m) 3
nn = Bn exp(-iwt) S4 - nEl An exp(i(8n - An)]
n = 1,2,3 (13)
S5 = S1S2 - 9
where
The method will fail when S5 - O. For
bi = ai exp(ikx1) equal probe spacing A, this occurs
br ar exp[-i(kx1 - when A = 0, r, 2r.... For unequal
probe spacing such that p = A2/A3, the
En - An exp[i(01 + 6n)]
condition S5 = 0 corresponds to

In order to minimize the error in the sin2 A3 + sin2(pA3)

estimate in the least square sense, write
the error term is written as + sin2(A3 - pA3) - 0 (27)

3 which gives
E2 = E [bi exp(iAn)
n=1
sin A3 = sin(p3) = 0 (28)
br exp(-i) - Bn]2 (17)
This occurs where A3 = 0, r, 2r,... and
The quantity E2 is minimized in the usual A2 = 0, r, 2r,...
least-squares sense which gives rise to
the following two equations in the Method #3 - Three Fixed Probes
unknown complex quantities bi and br. In a three probe arrangement, the
3
number of measurements available is more
E exp(iAn) [bi exp(iAn) than the number of unknowns. Since only
n=1 three unknowns are involved, the three
br exp(-iAn) - Bn] = 0 (18) wave heights at the three probes are
sufficient [Isaacson (3)]. This method
3 also avoids measuring the phase angles.
E exp(-1.An) [bi exp(iAn)
n=1
In terms of earlier notation using
br exp(-in) - Bn] = (19) Eqs. 8 and 11, the measured wave ampli-
tudes are related to the unknowns as
Upon solving the set of simultaneous
equations, the following expressions are 2 2 2
derived An = ai + ar + 2aiar (cosA + 24) (29)

S2S3 - 3S4 where A = 2x1 - E and Ai = 0 as before.

(20) Solving for A
S5
cos A = fi (30)
SiS4 - 3S3
ar = (21)
S5 fi cos(2An) - fn
sin A - (31)
sin(2)
and
where
3
Si = E exp(i2An) (22)
n=1 2 2 2
An - ai - ar
fn (32)
3
S2 = E exp(-i2) (23) 2aiar
n=1

92
 Eliminating A from Eqs. 33 and 34 and
Moreover, in certain areas, the solution
after some algebraic manipulation,
becomes imaginary. For example, when k
2 2
is near unity and the height measurement
a- + ar A is not accurate, the method may fail.

2aiar =r The least-squares method which uses

the maximum number of measured variables
where in the computation is expected to be the
most accurate. The recommended spacing
2
for this method is p = 0.45 or 0.65
2 2 .
Ai sin[2(A3 - A2)] - A2 sin(2A3) + A3 s [Isaacson (2)].
A in(2A2)
sin[2(A3 - A2)] + sin(2A2) - sin(2A
3) TEST SETUP
The wave absorbing beach at CM is
graded stone in a rack covered with a
plastic mesh. The rack is fixed at an
2
r- 1
[A.2
1
- + [A2 -
2 angle of 15° to the still water surface
and extends 10 ft. below the surface.
3
2sin(2A3)
The depth of the beach is 12 in. The
Overall length of the beach is 30 ft.
11/2 thus occupying a significant portion of
- 2[A2 - A] [A2 - A]
1 3 cos(2A3) the wave tank of 250 ft. length. A
schematic of the basin system is pres-
ented in Fig. 1.
The solution for ai and ar in terms of The
A and is r setup consists of three wave
probes placed in line along the longitu-
dinal axis of the tank. The probes were
capacitance type and were located 130 ft.
ai = [JA + r + JA r) (37) from the wavemaker. The wavemaker is a
pneumatic type and was used to generate
regular waves in the period range of 1.0
sec. to 4.0 sec. The water depth
ar = JA +r- JA r ) (38)
at the
test section was maintained at 5.54 ft.
However, because of the false floor
Once ai and ar are known, the incident sections placed in the tank, the depths
wave height, the reflection coefficient
at the generator and the beach end were
and the phase angle deeper. At the beach end, the floor
for the reflected
wave may be computed. section sloped down to 10 ft. Wave sets
were run with the wave probes at equal
spacing and at 60%/40% spacing. Data
This method fails when the denomina-
tor in Eq. 35 or 36 becomes zero. Then, recording began after the waves from the
in terms of A3 wavemaker reached the test area and
continued for a period long enough to
sin(203) ensure that 120 seconds of reflected wave
-
sin(2A3) data was obtained. Only the portion of
+ the run which includes the reflected wave
sin(2A3 - 463) = 0 (39) was analyzed.
which gives A3 = nn or nff/g or nff/(1 - A)
TEST RESULTS
for an integer n [Isaacson (3)]. For ex-
ample, when g 0.4 or 0.6, the method In the process of determining the
fails when A3 = 0, W, 5W/3, 21í, 571/2....
wave reflection coefficient, the investí-

93
gator is faced with three choices: (1) comparisons, the results from the 9.00
method used; (2) choice of wave number; ft. separation will be used.
and (3) spacing between probes. Indi-
rectly, the second choice requires the A comparison of the three methods is
accurate measurement of the distance presented for equal spacing (Fig. 4) and
between probes. This measurement becomes for unequal spacing (Fig. 5). Overall,
increasingly more sensitive as the wave the poorest results occurred at
length decreases.
Spacing Method Period A2 A3 A3
The wave length, L, can be computed (sec.) -4
from Linear Theory or calculated from the
wave period, T, the distance between Equal 2 2.25 3.22 6.45 2r
probes, d, and the time elapsed for a 3 2.25 3.22 6.45 2r
crest at Probe 1 to reach Probe 2, t, Unequal 3 1.75 5.57 9.28 3r
where 3 2.25 3.30 5.49 5r/3
3 3.75 1.75 2.92 r

L d (40) Results for unequal and equal spac-

-t
ing are compared in Fig. 6 for Method #2
and in Fig. 7 for Method #3. It should
Figure 2 presents the effect of the be noted that the radical in Eq. 40 and
choice of the wave number on the reflec- 41 became negative for Method #3 (unequal
tion coefficient for Method #1 (two
spacing) for wave periods of 2.75 and
probes) for a wave period of 1.75 sec- 3.75 seconds.
onds. In this case, the determination of
the experimental wave length and subse- CONCLUDING REMARKS
quent use in the formulations generates a
reflection coefficient of 5% rather than computational methods have
Three
the 20% value using the theoretical wave been compared for their stability and
length derived from Linear Theory. In accuracy with laboratory generated regu-
subsequent analyses, the results based on lar waves. In the process, the reflec-
experimental wave length will be pres- tion coefficient for an artificial beach
ented. in the CBI wave tank has been presented.

During the test runs, the probes The spacing of the probes has great
were either equally spaced (x2 - 11.25 influence on the accuracy of the results
ft. and x3 - 22.50 ft.) or at 60% spacing as claimed by Isaacson (3). The methods,
(x2 - 13.50 ft. and x3 - 22.50 ft.).
indeed, fail at certain spacings. How-
Using the two sets of wave runs, results ever, an equally important parameter is
for Method #1 can be presented for five found to be the wave length. The reflec-
different spacings (see Fig. 3). The tion coefficient was found to be quite
three most notable error zones are at sensitive to the wave length (i.e.,
nondimensional wave number). The compu-
Spacing Period tation of wave length by (linear) theory
(ft.) (sec.) A2 will not provide sufficient accuracy in
the reflection coefficient.
13.5 2.25 3.28
22.5 4.00 3.03 #2 with unequal spacing
Method
22.5 1.75 9.06 proved to be more reliable than Method
#3. It was not necessarily more reliable
Overall, the best results from
than Method #1 if the proper spacing is
Method #1 were determined from the short-
er wave probe separations. In subsequent

94
chosen. It is expected, however, that
the reliability of Method #2 will in-
crease with a greater number of probes.

REFERENCES

Jamieson, W.W. and Mansard, E.P.D.,

"An Efficient Upright Wave Absorb-
er", Proceedings on Coastal Hydrody-
namics, ASCE, Newark, Delaware, June
1987, pp. 124-139.

Ouellet, Y. and Datta, I., "A Survey

of Wave Absorbers", Journal of
Hydraulic Research, IAHR, Vol. 24,
No. 4, 1986, pp. 265-280.

Isaacson, M., "Measurement of Regu-

lar Wave Reflection", Journal of
Waterway, Port, Coastal and Ocean
Engineering, ASCE, Vol. 117, No. 6,
1991, pp. 553-569.

Goda, Y. and Suzuki, Y., "Estimation

of Incident and Reflected Waves in
Random Wave Experiments", Proceed-
ings of Fifteenth Coastal Engineer-
ing Conference, Vol. 1, 1976, pp.
828-845.

Mansard, E.P.D. and Funke, E.R.,

"The Measurement of Incident and
Reflected Spectra Using a Least
Squares Method", Proceedings of
Seventh International Conference on
Coastal Engineering, Sidney, Austra-
lia, Vol. 1, 1980, pp. 154-172.

FIGURE I
CM WAVE TANK LAYOUT WITH MOVABLE FLOORS
26
SECTION A-A INSTRUMENT
BRIDGE 15°

31.5 18

PNEUMATIC ADJUSTABLE CONCRETE

WAVE MAKER STEEL PANEL WITH CURRENT
PANEL FLOOR 10 x 10 WINDOW
0 TO 18 FT 0 TO GENERATORS
9°4_

8 4, DEEP PIT

1
33
,OMME
AMINE

BEACH
250
PLAN VIEW

95 NOTEZ ALL DIMENSIONS ARE IN FEET

 FIGURE 2

METHOD #1
PERIOD = 1.75 secs.
SPACING (4 9.0 ff.

0.50

0.40

EXP VALUE
CD 0.30
()
L.1-1

La..1
0.20
a ./ LINEAR THEORY

0.10 \ .Ncrs
/7
0.00
3.50 3.75 4.00 4.25
3.00 3.25
NON-DIMENSIONAL WAVE LENGTH

FIGURE 3

RC vs PROBE SPACING
METHOD #1

1.00

0.75

10.50

0.25

0.00
2.00 2.50 3.00 3.50 4.00
1.00 1.50
WAVE PERIOD (sec)

o 13.5 ft 0 22.5 fit

9.0 ft 11.25 ff

96
 FIGURE 4

COMPARISON OF ANALYTICAL METHODS

EQUAL SPACING

1.00

0.75

0.50

0.25

0.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00
WAVE PERIOD (sec)

METH #1 o METH #2 METH #3

FIGURE 5

COMPARISON OF ANALYTICAL METHODS

UNEQUAL SPACING

1.00

0.75

0.50

0.25

0.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00
WAVE PERIOD (sec)

METH #1 o METH #2 METH #3

97
 FIGURE 6

COMPARISON OF METHODS FOR

METHOD #2

1.00

0.75

0.50

0.25

0.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00
WAVE PERIOD (sec)

EQUAL UNEQUAL

FIGURE 7

COMPARISON OF SPACING FOR

METHOD #3

1.00

0.75

C-)
J
Lil
LIJ
0.50
LJ cp
c_.)

0.25

0.00
1.50 2.00 2.50 3.00 3.50 4.00
WAVE PERIOD (sec)

° EQUAL UNEQUAL

98
 SESSION II-A

RESISTANCE AND FLOW

 The Resistance and Flow Committee Report
G. Karafiath (Carderock Division, Naval Surface Warfare Center, USA)
predicting ship oil loss from model test
results is in its infancy.
Nevertheless, several oil loss trends
A new development with regard to the were established and the visualization
conduct of traditional resistance and of the oil, water, and trapped air flow
propulsion testing at DTMB is an is of great benefit to understanding the
increase in the size of the models used oil loss phenomena. The tests were
in the propulsion test. This increase conducted with a light machine oil dyed
in propulsion test model size is brought red and the post test clean up of the
about by the desire to have the same Circulating Water Channel added
hull and propulsor model size (possibly significantly to the testing cost.
the same models) as the cavitation model
size used in the Navy's new Large SHIP HYDRODYNAMICS LABORATORY
Cavitation Channel. Cavitation test con- UNIVERSITY OF MICHIGAN
siderations tend to result in propulsor Prepared by Stuart Cohen
and hull models that are Larger than the
minimum size models needed for the The Ship Hydrodynamics Laboratory has
propulsion test. Thus, for a 36 ft long conducted two series of experiments
destroyer model we will have a hull relevant to the Resistance and Flow
Reynolds number of 5 x 107 and a Committee. The first is a non-
propeller blade chord Reynolds number of traditional use of the towing tank for
1.5 x 106 during the model propulsion determination of flow patterns and
tests. With such large models, the pressure fields around two Ford heavy
scale effect on the hull, propulsor, and truck radiators. Full scale tests of
appendages will be minimized. the cab sections, including chassis,
engine and radiator, firewall and hood
There is an increasing use of laser were towed at speeds of 1 to 4 fps
equipment to support hydrodynamic test- corresponding to 10 to 40 mph. Sections
ing at DTMB. A laser-based wave height of the hood were transparent acrylic for
measurement method that tracks the flow visualization. Twenty-one pitot
bright spot of a laser beam piercing the tubes measured axial velocity in way of
free surface is under development. the radiator. Components were placed
Laser Doppler Velocimetry (LDV) wake and aligned by divers. Future work is
surveys, both the traditional type wake expected to show that fan design has a
survey and the measurement of wake in larger effect than component placement.
front of an operating propeller have
been conducted. Laser light sheets have The second series of tests were added
been used to visualize the flow and vor- resistance in waves of International
ticity behind bodies with and without America's Cup Class models. The IACC
operating propellers. Currently there is models were tested free-to-surge and
a developmental effort to measure such fixed-in-surge. For these particular
velocities. In addition, low powered models, the added resistance (and
lasers are being used in some mundane motion) and accuracy of measurements
alignment applications such as the were the same with either test method.
alignment of pitot tubes prior to At the speeds tested, these IACC models
calibration. have non-dimensional resistance
coefficients linear with respect to wave
One new type of testing that has never slope, and therefore, the previous
been previously performed at DTMB is the conclusion regarding method of testing
testing of an oil tanker's cargo oil cannot be assumed to be true in all
tank arrangement to determine oil spill cases. Tests for added resistance,
outflow due to a grounding accident and changes in draft, moment of inertia,
due the subsequent effects of current, heel and yaw angles have been conducted
tide and sea state. The testing and will be reported in the future.
methodology and the methodology of

101
 Experiments and Predictions
of the Resistance Characteristics
of a Wave Cancellation Multihull Ship Concept
M.B. Wilson, C.C. Hsu, D.S. Jenkins
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT Ln Length for hull element n

Ls b Setback distance (see Fig. 1)
This paper presents results of both analytical predictions Lsp Outer hull spacing (see Fig. 1)
and towing trasin validation experiments for the trasic Rf
resistance characteristics of a new small waterplane area, tri- Frictional resistance, estimated from ITTC
hull ship concept termed the Wave Cancellation Multihull extrapolator
(WCM). This is a variant of the O'Neill Hullforrn (01IF)- Rr Residuary resistance
type arrangement but which exploits extensive wave Rw Wavemaking resistance
cancellation effects between the various hull elements: the Rwx, Wavemaking resistance due to the outer hulls
center body, the center strut, and the two outer hulls. Rw,cb Wavemaking resistance due to the center body
Significant reductions in total ship wavemaking resistance Rw,cs Wavemaking resistance due to the center strut
have been predicted in a high speed range normally associated Rw4bo Interaction wavemaking resistance due to the center
only with fine form monohull ships. A tapered-strut-type body and outer hulls
center body shape has been developed as an alternative to the Rw,cbs Interacdon wavemaking resistance due to the cen=
original OHF centerbody-and-strut configuration. Towing body and center strut
basin tests confirm the predictions of the thin ship theory
Rw,cso Interaction wavemaking resistance due to the center
wavemaking resistance variations. The resulting complete-
ship wavemalcing resistance can be reduced to such an extent Stillt and outer hulls
S Wetted surface area (total)
that the total resistance of a WCM ship geometry is Tr,
comparable to that of a monohull at high ship speeds, despite Draft for hull element n
the substantially larger wetted surface area. Summary V Ship speed
comparison plots are presented for EHP/Displacement versus
speed for several 4300 LT ship forms, including a candidate
WCM arrangement. These show, for example, that at 30
14
knots it is possible to achieve a 28% reduction of effective
horsepower with a realistic WCM configuration compared L
with the SWATH VII performance (best performing SWATH
to date). It is also shown that the candidate WCM form has
Lib
about the same effective horsepower as a slim combatant X
monohull form at the speed in the neighborhood of 30-32
knots, despite 88% more wetted surface area. L2
TOP VIEW
NOMENCLATURE

Cf ITTC turbulent frictional drag extrapolator line

CR Resiclunry resistance coefficient 111111M11111=111111
or Total measured resistance coefficient DWL iflINIIMIMii
cw Wavemaking resistance coefficient
II T II II
4ID
cw,c Wavemaking resistance coefficient due to center hull
cw,0 Wavemaking resistance coefficient due to outer hulls
cw,i Interaction wavemaking resistance coefficient
fn Hull shape function for hull element n
Figure 1 - Schematic of O'Neill Hullforrn
g Gravitational acceleration constant
Represented by Model 5355-2

103
Ko Wave number = g/V2 7 n/2 2 2 3
Displacement, long tons R = 167r,pic-0 (P + Q )sec Ode
o
Fluid density
EYn Source strength distribution for hull element n
wheze

ABBREVIATIONS

DTMB/CDNSWC David Taylor Model Basin fr

Carderock Division, Naval Surface a(x,zic°ssin NI( 0(x COS() + y sin e)sec2
Q1 =
Warfare Center
OHF ONeill Hullform x exp( oz sec 2 e)cixdz
SWATH Small Waterplane Area Twin Hull
WCM Wave Cancellation Multihull
with g= acceleration of gravity, xo =g/V2, V=ship speed, and
p =fluid density. The formula is based on Cartesian
coordinates (x,y,z) moving with the ship. The origino has
INTRODUCTION been taken in the undisturbed water surface with oz vertical
upward and ox parallel to the direction of motion. For a thin
The main advantage of a small 1.vaterplane area multiform
ship hull form symmetrical about the xz plane, the source
ship is improved seakeeping. This has been the basis of strength is distributed over the center plane of the hull and is
interest in the twin-hulled SWATH arrangement. However, related to the hull shape function by the relationship
even the best SWATH ship configurations require relatively
large propuLsion power (compared with a monohull) because V af
of the large wetted surface area and thus higher frictional Cr(x,z)= (3)
resistance. As a result, there has been a search to try to find
relief from the large resistance penalty that accrues due to where f (x,z) is the submerged hull shape function. For
increased wetted surface area of multiple hulls compared with multihull ships if the disnirbanccs caused by each hull
a monohull witb the same displacement. The key idea is that element are small, then by superposition, the functions P ana
multihull configurations offer a wide array of possibilities for Q can be expressed in the forrn
reducing the wavemaking resistance component of drag by
exploiting arrangements of the hull elements that promote
extensive wave cancellation.
r
The ONeill Hullform (OHF) is a promising example of a n(xn ,z){ rsin}[x 0(xn cos() + yn sin 0)sec29]
trimaran arrangement of hulls. See Figure 1. It features a
Qj=JJ
lqrge submerged center body plus surface-piercing strut outer X exp(ic ozsec20)dxndz (4
hulls connected to the upper hull (1). Some advantage in
terms of total resistance has already been shown for the
Hullform at speeds above 21 knots in a comparison Coordinates xn and yn depend on the arrangement of the n
with a typical 4000-LT SWATH configuration according to th hull element. The summation is carrie,d out over all the
the test results of Wood (2, 3). The present work was aimed hull elements. Wavemaking resistance Rw given in Equation
at achieving more improvement by exploiting the (1) then contains components produced by the wavemaking of
wavemaking interference characteristics of multiple hull each hull element (like-product terms) ai2 and 1%2 and
ships. It is demonstrated that distinctive total resistance components produced by the wavemaking interactions of
reductions are indeed possible at high speeds with different hull elements (cross-product terms) 2fPiPj and
arrangements of the small waterplane area cancellation 2EQiQi. The like-product terms are positive, but the cross-
multihull (WCM) ships as compared with the best- product terms can be either positive or negative, depending on
perfonning SWATH. the hull arrangement There is the possibility of a reduction
in total wavemaking resistance because of the wavemaking
WAVEMAKING RESISTANCE THEORY interactions of groups of ship hulls (corresponding to the
negative cross-product terms).
Drag force experienced by a floating body moving in calm
water consists mainly of the components of fluid friction and Smdies reported in this paper started with consideration of
wavemaking resistance. The latter can be estimated by the vravemaking resistan= of the O'Neill Huliforrn (OHLF)
potential flow analysis. The velocity potential of a moving tri-hull ship (1). As depicted in Figure 1, the 011F has a
ship can be built up using suitable singuLarity distributions. large submerged c,enter body with a connecting surface-
Following Lunde (4), the thin ship approximation for the piercing center strut and a pair of identical surface-piercing
wavemaking resistance due to a source distribution an can be strut-like out hulls placed synunetrically with respect to the
written as
104
 center hull. The center strut and the outer hulls of die-- whose principal dimensions are given in Table 1. The center
example discussed here have parabolic nose and tail shapes. body-plus-center strut provides 80 percent of the total ship
The center body is a body of revolution with elliptical nose displacement, so each outer hull provides 10 percent of the
and tail shape. total. Results of calculations showing the effect of varying
the distance between outer hull centerlines Lsp and the
Here, the total wavemaking resistance has three main setback distance Lsb on the wavemaking resistance of this
components due to: center body and center strut, the outer basic 011F configuration are given in Reference (6). These
hulls, and the interactions between the center hull elements results show that the outer hull spacing Lsp has relatively
and the outer hulls
little influence while the setback distance Lsb has profound
R w = Rw,c Rw,0 Rw,i effects on the total wavemaking resistance of the 011F
(5) arrangement. For subsequent discussion the outer hull
where spacing has been held constant at 106 ft (32.3 m) for the
candidate designs.
Rw,c = components due to center hull elements
Rw,c = Rw,cb Rw,cs Rw,cbs
with Rw,cb = center body contribution TABLE 1- Particulars for Tri-Hull Ship Geometry Used
Rw,cs = center SIMI contribution for Calculations
Rw,cbs = center body-center strut interactions OHF Wave Cancellation
Rwoz, = component due to outer hulls Multihull (W.4)

= components due to center hull-outer hull interactions hem Model 5355-2 WCM-2 Modd 5485

Rw,i = Rw,cbo Rw,cso

with Cosa Body Length (ft/m) 354.99/10820
Rw,c1x, = center body-outer hulls contribution Crater Soot Length ((1/m) 280.05/8536
Rw.cso = center strut-outer hulls contribution Carer Body Mm Beam (film)
Caster Sam Max Beam ((/m)
21.45,6.54
9.84/3.00
Caner Hull LeogUr (/m) 390.0/118.87 3904,11887
Corer Hall 61in Beam ((t/m) 8.65/2_64
Systematic analyses of the wave resistance characteristics Caw Hull Max Beam ((/m) 19.63/5.80
9.95/3.03
21.58.6-58
have been carried out to determine the influence of the main Omer Hull Max Draft
Our Hull Lo;th
(Mm) 31.189.69 30.49.14 2922)891
((.'ur) 190/357.91 190.0157.91 ! 9)0157 3:
geometric parameters of the O'Neill Hullform tri-hull Outer Hall Max Beam ((t/m) 66fV201 660201 6.60201
Omer Hall 61sm Draft
arrangement. The results are presented in dimensionless Omer Hull Spscmg
((uto)
(( tim)
13.88/4.23
106.03231
1388/423
1060/32.31
11881422
10603231
coefficient forms Tool Waled Sad.= 0 fin 2) 3740513475 41885/3891 40823/3792
Total Dispiscomem (LT/ionne) 43004369 430614369 430014369

R
C
W
=- 2--s Cw +C W, 0 +C
,C (6)
Figure 2 shows the various components of the
where S is total wetted surface of the all the submerged hull wavemaking resistance coefficient Cw,c for the 011F center
elements. It has been convenient to use these familiar hull. Results include the contributions due to the submerged
conventional resistance coefficients for displaying the drag center body and the surface-piercing center strut, plus their
component bre,akdown for a given arrangement. However, wavemalcing interaction. While the strut-alone component is
because all multihull configurations have larger wetted
surface area than, say, a monohull of the same displacement, 0.0050
final comparisons of resistance performance for several ship
0.0045
types are never made on the basis of resistance coefficients
involving surface area. 0.0040 Center body
Center strut
0.0035
------ Centel' bOdy-Center strut interaction
A modification of the small waterplane area twin hull L. 0.0030 Total center hull wavemaking resistance
(SWATH) ship resistance computer code by Chapman and
0.0025
Wernli (5) was used for the present computations. C..)

0.0020

WCM CONCEPT DEVELOPMENT 0.0015

0.0010
O'Neill Hullform
0.0005
Understanding the resistance performance of the °Neill
Hullforrn was the starting point for this work. The potential o ;--,e.",-;-1'.1,4P-;"
14 18 18 20 22 24 28 20 30 32 34 38 38 40
flow wavemaking resistance theory has made it possible to Speed, V (mots)
analyze the detailed character of the components of wave drag.
Note that this cannot be accomplished by experimental Figure 2- Calculated Wavemaking Resistance
means. Calculations have been carried out for a 4300 LT
version of the °Neill Hullform (built as Model 5355-2) Coefficient Characteristics of Center Hull for
°Neill Hullform Model 5355-2
105
 0.0025
quite small, the center body and interaction components are --- Center body-outer hulls
substantially larger, and they have pronounced humps and 0.0029 Center strut-outer hulls
hollows. In contrast, the coefficient of wavemaking 0.0015
Total interaction
resistance component of the outer hulls varies more
0.0010
gradually, as indicated in Figure 3.
0.0005
Wavemaking interaction resistance component of the OHF 0
Model 5355-2 arises from both the center body-outer hull and 0.0005
the center strut-outer hull interactions. The calculated
variations of the coefficient Cw,i for three different values of -0.0010

setback distance are displayed in Figure 4. 0.0015

-0.0020 Setback =99.1 ft

0.0050
-0.0025
14 16 18 20
.1.1.1.1.1,1,1.1.1
22 24 26 26 30 32 34 38 33 80
0.0045
Speed, V (knots)
0.0040 Outer hull spacing = 106.0 ft
0.0025
0.0035 0.0020
--- C_enter body-outer hulls
C_enter strut-outer hulls
0.0030 0.0015
Total interaction

C.) 0.0025 0.13010

0.0020 0.0005
0.0015 o

0.0010 -0.0005
0.0005 0.0010
o .1.1.1.1,1
20- 22
16
1.1.1.1
24 26 28 30 32 34
1.1.
36 3a 40 -0.0015
14 16
Speal. V (knots) -0.0020 Setback= 155.9 ft
Figure 3- Calculated Wavemaking Resistance -0.0025
14 16
.1,1.1.1.1,1.1.1.1.1.1.
18 20 22 24 26 28 30 32 34 38 33 40
Coefficient of Outer Hulls for O'Neill Speed, V (knots)
Hullforrn Model 5355-2 0.0025
Center body-outer hulls
0.0020 Ocnter strut-outer hulls
0.0015
Total interacnon
It is noted that certain interaction wavemaking resistance
components oscillate through negative values in some ranges 0.0010
of speed. Total wave resistance of course, is a positive 0.0005
definite quantity, as seen from the formula of Equation (1). S.
o
Negative interaction wavemaking resistance involves cross L) = 106 ft
product terms from different hull elements, and it has a -0.0005
-

physical explanation. Since wavemaking resistance is a -0.0010

pressure drag, it is the result of integration over the entire
0.0015
wetted hull of the local pressure force vector component in
the longitudinal direction. Interaction wavemaking arises -0.0020 4 (c) Setback= 210 ft
from only the wave pattern due to a hull element interfering -00025 1.1.1.1.1.1.1,1./.1.1.1. 24 26 28 30 32 34 36 38 40
with the waves of a neighboring hull element. If the hulls 14 IS 13 20 22
Speed, V (knots)
are arranged such that the trough of the neighboring hull
wave system occurs in the vicinity of the crest of the wave Figure 4 - Components of Interaction Wavemaking
system generated by the local hull, it is possible to envision Resistance Coefficient for O'Neill Hullform
a situation where the net wave system, due to the Model 5355-2
interference, features a small net wave crest or even a trough
near the bow and a Larger wave amplitude near the stem of the
local hull. Thus if just the interfering wave systems are Depending on the amount of setback and ship speed, the
considered, there is the possibility that a wave-induced thrust influenc.e of the setback can vary considerably. For instance,
or negative pressure drag component is produced that involves for the smaller setback shown in Figure 4a, the wavemaking
only the effects of those wave systems. Any improvement of interaction effect is largely detrimentaL With the larger
the general 011F-type of in-hull arrangement should be aimed setbacks, there are more substantial ranges of negative Cw,i,
at taking advantage of the special wave interference situations and these shift to higher speeds. See Figure 4c.
for which negative interaction wavemaking resistance could
play a role. Another important feature is that major contributions to

106
 the wavemaking resistance of the O'Neill Hullform, as shown 0.0050
in Figure 2, arise from the centerbody and from the center
0.0045
body-center strut interaction. Large resistance improvements
could be possible if proper alterations were made to the 0.0040 Center hull
geometry of the center hull. Outer hull
c 0.0035
Outer hull spacing = 106.0 ft
0.0030
Wave Cancellation Multi hull o
With the calculated wavemaking resistance results of the Q 0.0025
3-
011EF as motivation, alternative ini-hull ship forms have been 0.0020
designed with special attention to exploiting features that lead
0.0015
to reduced interaction wavemalcing resistance. The new hull
forms are referred to as Wave Cancellation Multihull (WCM) 0.0010

ships. Principal dimensions of an example of the 4300 LT 0.0005 ........

(4369 tonne) prototypes studie-d are summarized in Table 1.
0.0000
..................
Figure 5 is a sketch of this tri-hull ship concept_ The first 14 16 18 20 22 24 26 28
I
30
,
32
I
34 36 38 40
generation of the WCM geometries are constrained to have Speed, V (lcnots)
the same waterplane are-a, total displacement, and identical Figure 6- Calculated Wavemaking Resistance
outer hulls as the OHF ship represented by Model 5355-2. A Coefficient Components of Center and
key difference is that the °HT center body-center strut Outer Hulls for Configuration WCM-2
combination has been replaced by a tapered, strut-like center
hull with a trapezoidal cross section. Much of the 0.0025
displacement volume of the WCM center hull has been 0.0020
150.00 ft setback
shifted to deeper submergence and the strut-like upper part ------ 200.00 ft setback Outer hull spacing = 106.0 ft
0.0015 250.00 ft setback
that pierces the surface has a smaller slenderness ratio
(beam/length) compared with the OHF. The interference 0.0010
wavemaking effect between the center body and the center 0.0005
../.
strut of the OHF has been essentially eliminated. Total 0.0000 -
-5.-
_
wetted surface area of the WCM-2 is about 12 percent larger U
than that of the OHF represented by Model 5355-2. -0.0005

0.0010

0.0015

0.0020

// ///////// /./ ///// 0.0025

14 16 18 20 22 24 26 28 30 32 34 36 38 40
Speed, V (knots)
Figure 7 - Calculated Interaction Wavemaking
Resistance Coefficient Components at
Three Different Setback Distances for
TOP VIEW Configuration WCM-2
0.0050
150.00 ft setback
SLW
0.0045
------ 2013.00 ft setback Outer hull spacing = 106.0 ft

DVIL iEUìi
Ismossmmit
II T2
0.0040

0.0035
250.00 ft setback

0.0030
82
U 0.0025
FRONT VIEW
0.0020

Figure 5 - Schematic of Wave Cancellation 0.0015

Multihull (WCM) Ship sc
0.0010
...........
0.0005 . -
Wavemaking resistance performance of this tri-hull ship 0.0000
form has been extensively studied. Figure 6 shows wave 14 18 18 20
1

22 24 26 28 30 32 34 36 38 40
resistance coefficients versus speed for the outer hulls Cw,0 Speed, V (knots)
and the center hull Cw,c Results are displayed in Figures 7 Figure 8 - Effect of Outer Hull Setback on
and 8 for the coefficients versus speed of the interaction Calculated Total Wavemalcing
component Cw,i and the total wavemalcing resistance Cw, Resistance Coefficient for Configuration
WCM-2

107
with contours for three different setback distances. It is clear TABLE 2- Physic,a1 Model and Ship Geometry for Wave
from Figure 7 that the larger setbacks can result in more Cancellation Multihull Experiments with
extensive negative wavemaking interaction contribution Model 5485
involving the center and outer hulls. The high speed Ship Modr.1
resistance advantage of larger setback distance carries over
into distinctive net reductions of total Cw, as indicate,d in f.ster-t
15.46 (4.71 m)
Leo& 390 ft (118.91 m) f-t
Figure 8. There is a surprisingly wide range of high speeds Draft 29.22 ft (8.91 m) 1.158 ft (0.353m)
10 ft (305m) 0.396 f-t (0.121m)
where the destructive wave interference works to produce the Beam at DWL
19.13 ft (5.83 m) 0.758 ft (0.231m)
Mader= Beam
desired wave cancellation effect. Wened surface arca 27,195 ft2 (2526m2) 42.722 ft2 (3.97m2)
Displa= 3,412 LT (3467 t) 462.6 Ibm (209.8 kg)

EXPERIMENTAL RESULTS FOR WCM OuncrHull fF_ach).

ft (57.9 m) 7.53 ft (13m)
Length 190
Draft 13.88 ft (4_23 m) 0.55 ft (0168m)
Resistance experiments have been conducted on a model
WCM in order to provide direct comparisons with the
predictions of the multihull thin ship the,ory.
Dt
Beam
Wetted surface area

Total Shtp
6.6 ft (2.01 m)
5308
417
ft2 (493m2)
LT (424 t)
0.262 ft am m)
8.338 ft2 (0.775m2)
56.6 lbm (25.7 kg)

Oust hull sqraraticc 106 ft (313m) 4.2 ft (128 m)

Wetted surface arca 37,811 ft2 (3513m2) 59.4 ft2 (532m2)
bkdel Displascarciat. Volume: 148.600 ft3 (4208 m3) 9.253 ft3 (0-262 m3)
lbm (261.1 kg)
A new wood model was built for the center hull of the Mass: 4,246 LT (4314 t) 575.7

WCM Model 5485. The existing outer hull models

employed for the earlier OHF model experiments by Wood 106 ft (32.3 m) and setback distances of 150, 180, 200, and
(2, 3) were used here as well. Principal dimensions of the 250 ft (45.7, 54.9, 61, and 76.2 m) corresponding to ratios of
idealized WCM hull form are given in Table 1. For the 0.385, 0.462, 0.513, and 0.641 with respect to the center
physical model of the center hull, the two lower corners of
the trapezoidal idealized section shape were rounded with a hull length.
bilge radius equal to about 1/4 the local maximum beam, Comparison plots of the measured residuary resistance
applied to individual se,ctions along the length of the body. coefficients and the calculated wavemaking resistance
The final painted finish on the model is an oil base enamel coefficients corresponding to the teste,d configurations are
applied over an acrylic primer-surfacer. Turbulence presented in Figures 9a through 9d. There is generally good
stimulation was provided by 0.024 in. (0.61 mm) diameter
trip.vires fitted flush with the hull surface at a distance of 5
ageement between the calculated and measured results at
speeds above 22 lcnots. At lower speeds, the agreement is
percent of the length aft from the stem profile on each of the
three hull elements. There were no extra appendages such as less consistent for the setback distances of 150 and 250 ft
rudders or control fins fitted to the bare hulls. Table 2 (45.7 and 76.2 m), but still rather good for the cases of 180
provides the principal dimensions of the physical model and 200 ft (54.9 and 61 m) setback. For the case of 250 ft
assembled for these tests, with the design scale ratio of (76.2 m) setback there is a noticeable region of discrepancy
between the measured CR and the calculated Cw in the form
A,=25.23 to be used for data extrapolation.
of a hump centered about the speed 20.5 knots. The
computed results have been re-nm and the test results checked
Procedure
Model resistance experiments were carried out in the several times, so the explanation of the mismatch is not
DTMB/CDNSWC CarriagenDeep Water Towing Basin with clear. This does not detract from the good agreement at
the model fixed in sinkage and trim al the design waterline. speed.s higher than 24 knots.
Drag force was measured with a single 100 lb (445 N)
capacity DTMB block gauge. Residuary resistance
Overall, the validation experiments for the WCM
resistance performance demonstrate that both the thin ship
coefficients, based on the total wetted surface area, were
obtained from measured total drag by subtracting the sum of theoretical development and the resistance trends derived from
estimated viscous drag plus the parasitic form drag of the it are credible and accurate enough to exploit for concept
design purposes, especially in the range of high speeds.
turbulence stimulator wires. In coefficient form
(7) DISCUSSION
CR = C T [Cf C Dtw C11/S

This work has shown that in the high speed range above
where Cf = turbulent friction drag coefficient taken as the 25 knots, the wavemaldng resistance of the basic OHF tri-
1957 ITTC extrapolation line; and CDtw = tripwire drag hull ship concept can be significantly reduced by a
coefficient based on the wire diameter d times the total combination of placing the outer hulls with large setback
girthwise wire length 1 below the waterline of all three hulls. distance and reshaping the center hull. The new feature of the
For the very slender hulls of this WCM arrangement, the WCM concept involves shaping the center hull in order to
residuary resistance coefficient is essentially all due to reduce the contributions to wavemaking resistance due to
wavemaking drag C. The experiments include,d WCM wavemaking interactions between the center strut and the
configurations with the scaled outer hull centerline spacing of center body, as well as the interactions between the center

108
 hull and outer hulls. This concept is the subject of a patent 0.0040
application by C.C. Hsu. 0.0036

0.0032
For realistic applications of the WCM concept, the
geometry of the center hull and the amount of outer hull 0.0028 Theory for Cw
setback will be limited by ship structural considerations, 0.0024
Experiment for CR
practical arrangements of machinery, and ship motions 0.0020
criteria that are beyond the scope of the present discussion.
0.0016

0.0012 AA
0.0040
0.0008 A
0.0036 AAA
0.0004
0.0032 9 (d) Setback = 250 ft
0.0000 t
0.0028 Theory for Cw 14 16 18 20 22 24 26 28 30 32 34 36

0.0024
o Experiment for CR Speed, V (lcnots)
Figure 9- Comparison of Calculated and
L.) 0.0020
Experimental Results for Total Cw for
0.0016 WCM Ship Model 5485
0.0012

0.0008
To illustrate how the WCM performance measures up, two
important comparisons are included here. Figure 10 shows
0.0004

0.0000
14 16 18 20 22 24 26
Speed, V (knots)
I

28
II
9 (a) Setback = 150 ft

30 32 34 38
the effective horsepower-per-displacement ratio versus ship
speed for three multihull ships: the WCM-2 with 210 ft (64
m) outer hull setback; the OHF Model 5355-2 with 155.9 ft
(47.5 m) outer hull setback (Wood (3)); and the SWATH VII
twin-hull ship (results from Reference (7)). All three
0.0040
performance curves are for ships with total displacement of
0.0036 4300 LT (4369 tonne) and correlation allowance CA =
0.0032 0.0005. Above 23 knots, the WCM-2 ships shows clear
superior performance over the other two. At the ship speed
0.0028 Theory for C. of 30 knots, the WCM-2 configuration requires 22 percent
0 Experiment for CR
0.0024 lower horsepower than the OHF ship, and 28 percent lower
0.0020 horsepower than the SWATH VII.
0.0016
12
0.0012
WCM-2, 210 ft setttack
0.0008 ex, 10 - - - OHF 5355-2, 155.91 ft setback
SWATF1 VII

11111
0.0004
9 (b) Setback = 180 ft
-
0.0000
14 16 18 20 22 24 26 28 30 32 34 36
Speed, V (knots)

0.0040

0.0038

0.0032 2

0.0028

0.0024 14 15 16 17 118
11/111111
19 20 21 22 23 24 25 26 27 28 29 30
C.)
0.0020 Speed, V (blots)
0.0016
Figure 10- Comparison of EHP/A for Multihull Ships
0.0012
WCM-2, OILF 5355-2, and SWATH VII
0.0008
Figure 11 shows the effective horsepower-per-
0.0004
displacement performance of the WCM-2 with 210 ft (64 m)
0.0000 outer hull setback compared with the performance of two
14 16 18 20 22 24 26 28 30 32 34 surface combatant monohull ship forms, designated as
Speed, V (knots)
Monohull 1 and 2. Selected particulars for these two ships

109
are given in Table 3. As before, the total resistance results TABLE 3- Selected Particulars for Scaled
have been scaled for ships with the same total displacement Monohull Geometry
of 4300 LT (4369 tonne) and CA = 0.0005. In this Item Monohull 1 Monohull 2
comparison, the larger value of displacement-length ratio for
Monohull 2 (A/(.0 1L)3 = 84) produces a steep incre,ase in Length, L (ft/m) 434.41/132_41 371.31/113.18
total resistance beyond a speed of 24 knots compared with he Max Beam, Bx (ft/m) 45.06/13_73 47.01/14.33
variation for Monohull 1 (A/(.01L)3= 52.45). Both Max Draft, Tx (ft/m) 15.98/4.93 16A9/5.03
monohull ship forms require noticeably less effective Wetted Surface, S (ft2 /m2) 22274/2069 20272/1883
horsepower than the WCM ship at speeds below 25 knots. Displacement, A (LT/tonne) 4300/4369 4300/4369
However, in the high speed range, the performance
comparison favors the WCM ship. The cross-over point for L/Bx 9.64 7.9
Bxrrx 2.82 2.85
equal horsepower requirement occurs at a speed somewhat
CB 0.481 0.523
over 25 knots for Monohull 2 and at about 32 knots for A/(.01L)3
-
(LT(ft)3 ) 52.45 84.0
Monohull 1.

12
WCM-2, 210 ft setback
Monohnit 1
monohnt is because of the higher wetted surface area of the tri-
Monohull 2 hull ship. The variations of the residuary resistance again
show the dramatic influence of wave cancellation for the
WCM-2 configuration, which lends to total drag advantage in
the range of high speeds.

While the WCM ship configuration appears to have

4
favorable total resistance characteristics at high speeds, even
in comparison to a monohull, the large resistance hump at
the moderate speed of around 21 knots presents a feature

o
14 15 16
1/1iiIIIIIt
17 18 19 20 21 22 21 24
Speed. V (knots)
25 26 27 28 29 30
which will markedly affect any propulsion system design.
This situation would have to be handled similarly to the
design problem associated with the drag hump of a hydrofoil
craft near take-off speed. Clearly, ship cruising speeds in the
range 18-22 knots should be avoided.
Figure 11 - Comparison of EliflA for Multihull Ship
WCM-2 and Monohull Ships 1 and 2
200,000
WCM-2. 210.0 ft setback
180.000
OHF 5355-2, 155.91 ft setback
The favorable performance of the WCM-2 configuration at 160,000 SWATH VII
high speeds can be attributed to the variation of residuary
140.000
resistance and the remarkable effect of wave cancellation due
to destructive wave interference. This effect dominates the 120,000
situation at high speed despite the much larger wetted surface 100,000
area of the tri-hull arrangement. To show what is going on,
80.000
the detailed variations of residuary resistance versus speed are
60,000
given in Figure 12 for the multihull ships WCM-2, OHF
Model 5355-2, and SWATH VII --- all at 4300 LT (4369 40.000
tonne) displacement. Frictional resistance magnitudes are 20.000
roughly the sarne for all three multihulls. For example, at
o
30 knots, the frictional resistance values Rf for the WCM-2 14 16 16 20 22 24 28 26
and the SWATH VII are 10.7 percent and 4 percent larger, Speed, V (krats)
respectively, than the Rf for the OHF. The residuary Figure 12 - Comparison of Residuary Resistance for
resistance curves show the advantage of the WCM-2 at high Multihull Ships WCM-2, OHF Model 5355-2,
speeds above 24 knots, as well as the exaggerated hump of Rr and SWATH VII
for the WCM-2 at about 20.5 knots.
Thus far, only the resistance performance changes achieved
Dimensional comparisons of the drag components Rr and by using the new tapered center hull design with the
R f versus speed are shown in Figures 13 and 14 for the displacement-sharing of 80-10-10% have been discussed.
WCM-2 and the two monohull combatant ship forms Further total resistance improvements may be possible by
Monohull 1 and 2. As expected, the frictional resistance of alterations to the outer hull shapes, by other center hull shape
the WCM-2 ship is much larger than for either of the changes, or by combinations of more pairs of small wing

110
 200.000 200,000
180.000
WCM-2. 210.0 ft setback WCM-2, 210.0 ft setback
180,000
Monohull 1 Monohull 1
160.000 Monohull 2 160,000 Monohull 2
140.000 140,000

120,000 120.000
f-
cn
100.000 100.000

80.000 80,000

80.000 80,000
40.000
E
40.000
20.000 20.000 r's
O
O
14 16 18 20 22 24 25 30 14 16 la 20 22 24 26 zs 30
Speed. V (knots)
Speed. V (imam)
Figure 13 - Comparison of Residuary Resistance for Figure 14 - Comparison of Frictional Resistance for
Multihull Ship WCM-2 and Monohull Ships Multihull Ship WCM-2 and Monohull
1 and 2 Ships 1 and 2

hulls. Although the trends from the present results are comparable to that of a monohull ship in the speed range
strictly valid for ships in the range of displacement around around 32 knots (based on EHP/LT). This means that the
4000-4500 LT, it is believed that similar designs could lead seakeeping benefits of a small waterplane area ship can be
to large resistance improvements for ships of different achieved without the resistance penalty that afflicts the
displacements. Dimensions and arrangements of hull SWATH ship concept at high speeds.
elements are dependent on the operating speed and overall The attributes of reduced power requirements, good
displacement of the ship. seakeeping, and the wide open deck area available with the
WCM arrangement could lead to promising applications for
CONCLUDING REMARKS oceanographic and survey ships, ferries, cruise ships, and
surface combatants.
The proposed new variant of the tri-hull
Hullform has been developed by systematic attempts to ACICNOWLEDGMENTS
exploit Wave cancellation and wavemaking-reduced hull
arrangements to achieve substantial wave interference This work was supported by the CDNSWC Independent
involving the center body and outer hulls, and the center body
and the center strut. The important alterations incorporated in
Exploratory Development Program and the Surface Ship
Technology Block Program. The authors wish to express
the WCM include locating the outer hulls well aft (with large their appreciation to Dr. Bruce Douglas, CDNSWC Code
setback) and the replacement of the center body-plus-center 0112 Director of Research, for his support and advice at
strut of the original OHF with a more slender, tapered center several stages of this project. The participation of Mr. G.
hull form.
Calculation procedures used for exploring the favorable
Robert Lamb in the early stage of the work on 01IF
resistance improvement is gratefully acknowledged. The
geometries have been validated by comparisons of calculated
wavemaking resistance with towing basin measurements computer implementation of the thin ship theory analysis of
the 011F was originally carried out by Dr. Yoon H. Kim,
made with the WCM Model 5485. Generally good agreement CDNSWC Code 1522. The authors are grateful to Ms.
in the high speed range observed in these comparisons gives
confidence in the wavemaking resistance trends developed for
Melanie Joy, Code 1522, for typing the manuscripi
the WCM arrangement.
REFERENCES
Resistance improvement of multihull ships depends very
much on the hull element arrangements and to a somewhat
lesser degree on the hull shapes. This means that the best W.C., "A New Small Waterplane Area Ship
arrangement for a particular case will vary with ship Concept," Paper No. AIAA-86-2382, 8th ALAA
displacement and speed of operation. Advanced Marine Systems Conference, San Diego,
September 1986.
The key result in all t.his is that in the range of high
cruise speed, the projected reductions in wavemaking
resistance for the tri-hull WCM configuration are sufficiently Wood, LE., "Effect of Longitudinal Location of
large that there are marked improvements in total effective a Pair of Outer Hulls on Resistance for the 4300
power compared with that for a SWATH ship of the sarne ton O'Neill Hullform Concept (01IF) Ship
displacement, despite the increase of wetted surface area. In Represented by DTNSRDC Model 5355-1,"
fact, the total resistance performance of the tri-hull WCM is DTNSRDC/SPD-1147-01, August 1985.

111
Wood, J.E., "Effect of Increased Outer Hull Setback on
Resistance for the ONeill Hullform Represented by
DTRC Models 5355-1, 2." DTRC/SHD-1147-02,
August 1988.

Lunde, J.K., "On the Linearized Theory of Wave

Resistance for Displacement Ships in Steady and
Accelerated Motions," Transactions SNAME, Vol. 59,
1951, pp 25-76.

Chapman, R.B. and R.L. Wernli, "Operating

Instructions for 'DRAG' Computer Prograrn," Naval
Undersea Center Technical Note 1385, 1974.

Wilson, M.B. and C.C. Hsu, "Wave Cancellation

Multihull Ship Concept," paper for Intersociety High
Performance Marine Vehicle Conference (HPMV '92)
Arlington, 24-27 June 1992.

Kirkman, K.L., B.J. Young, J.W. Kloetzli, and P.

Majumdar, "Model Tests and Engineering Studies of
SWATH VII Small Waterplane Area Tvvin-Hull Ship,"
Hydronautics, Inc. Technical Report 7694-1, 1976.

112
 Computation of the Hydrodynamic Forces and Moments
Developed on a Body of Revolution at Combined
Angles of Attack and Angles of Drift
and Yawing Angular Velocities
Y.S. Hong (Carderock Division, Naval Surface Warfare Center, USA)

INTRODUCTION Young (2).

The lateral forces and its moments
The accurate prediction of the are computed under the assumption that
hydrodynamic forces and moments devel- there is separation in the two-dimen-
oped on the hull and appendages of a sional cross flow. The boundary layer
submerged vehicle is required for de- equation is solved to the separation
termining its stability, control, and point and the frictional drag is inte-
maneuvering characteristics. Various grated to compute frictional force.
analytical methods have been developed Furthermore, it is assumed that a con-
to make these predictions, including stant pressure is acting on the two-
those of computational fluid dynmaics, dimensional section beyond the separ-
but none of them have been totally ation point.
successful. This is because it is
particularly difficult to accurately INVISCID FORCE COMPUTATION
determine both the distribution of the
velocity over the surface of the hull The coordinate system, oxyz moves
and the location of the separation at a speed of Un, which is the mean
lines. The location of the lines of speed of the bo6y along the positive
separation has an important effect on ox-axis. The ox-axis is parallel to
the magnitude of the hydrodynamic for- the body axis and the oz-axis is
ces and moments developed on the hull. directed vertically downwards. The
The present method is an exten- origin, o is located at the center of
sion of the technique discussed in gravity of the body. When the body
Reference 1. It is assumed that the is undergoing a rotational motion, a
total hydrodynamic force and moment can positive drift angle is defined as
be divided into inviscid and viscous shown in Figure 1.
parts. The inviscid part of the forces The total velocity potential can
and moments is computed using three- be expressed as
dimensional potential theory. The
body surface is discretized with many
13.(x,y,z) = 00(x,y,z) + 0(x,y,z) (1)
surface elements and the unknown str-
engths of the source and sink at each where 0 is disturbance velocity poten-
surface element are assumed to be tial due to the body. The gradient of
constant. The velocities at surface 0o is given as
elements are computed and saved for
later computation of the viscous for- Vcpo = (Rw-y, wx, 0), (2)
ces and moments. The viscous part of
the forces and moments is computed The disturbance velocity potential can
with the application of the boundary be expressed in the integral form as
layer theory for laminar and turbu-
lent flows. The axial force is com- 1 ee
0(x,y,z) = -- jj G(P,Q)a(Q)dS(q) (3)
puted with the method of develped by 117-1- s

113
where P(x,y,z) is the field point, u U
Q(x0,y01z0) the source point, S the Ko = 2m I -- (1 -)YdY (9)
body surface, a the unknown strength 0 U1
of sources and sinks distributed
on the body surface, and G is the where u is the velocity in the wake
Green function which is given as parallel to the body axis and U1 the
incoming velocity parallel to the
body axis. The velocity U1 is can be
1
G(P,Q) = (4) expressed as

Rwcosip, (10)
where U1 =

r2 = (x-xo)2 + (y-yo) + (z-zo)2 (5) In Equation (10) Kt is the momentum

area at the tail and Ut the velocity
The unknown strength of sources and at the edge of the boundary layer at
sinks, a is determined by the body tail. In order to compute Kt ,the
laminar boundary equation of a body
boundary condtion.
The forces acting on the body is of revolution is solved to the tran-
expressed as sition point and then, the turbulent
boundary layer is computed up to the
e fi pñdS (6) tail. The detailed procedure of the
P
1 2
S
numerical solution of the boundary
layer equation is given in Reference
where n is the unit normal vector at 2.

the body surface and p, the pressure The method to compute normal for-
around the body, is ce and its moment is slightly differ-
ent from that to compute axial force.
The sectional crossflow drag is integ-
P [1V012 - IV(1) 12] (7) rated along the axis of the body.
P = 2 o
The two-dimensional flow is given by

where p is the density of the fluid. U =2sin()

The detailed procedure of the num-
erical solution of Equations (3) and where U2 is the two-dimensional free
(6) are given in Reference 1. stream velocity. The velocity U2 is
computed by velocity potential.
VISCOUS FORCE COMPUTATION Furthermore, it is assumed that the
flow separates and the separation po-
For the computation of the axial int must be known to perform the com-
force which is parallel to the ox-axis putation. The crossflow drag is ex-
the method developed by Young (Ref. 2) pressed as
is used.
The total drag coefficient is ex- (12)
D2D = DF + DP
pressed as
where DF is the frictional drag and D
2K U is the form drag. The frictional
o .2
drag can be computed by numerical sol-
CA = S = §t (t)3 (8)
1 ution of two-dimensional boundary
layer equation.
where S is the surface area of the The pressure in the separated
body. The momentum area of the wake region is assumed constant and the
far downstream is defined as pressure coefficient is expressed as

114
 0
sp
= 160 degrees, at the bow
C
P
= 1 - (ll)2, when 0 <0 (17)
sp
U2 = 105 + 25r' + CO, at the stern
(13)
U C = 0, when r' = 0.42
= 1 - ( sp)2, when 0 > 0 0.5, when r' = 0.66
sp
U2
where U is the velocity at the point Equation (17) is the separation angle
of sepaPStion and 0 is the angle at when the drift angle is zero. When
the point of separaigon as shown in the drift angle is not zero, the fol-
Figure 1. Then the form drag is com- lowing expression is used
puted as follows:
(0sp)0 = 0sp+(126.6-8r'-0 sp )IP/20 (18)
n
D = pU2 f C (14)
P P rcos02 de2
0 In Equation (15), 0 is deter-
mined by the results of 9Inviscid force
The viscous forces and moments are computation and the subsequent viscous
the integral of Equation (12) and given forces show some discrepancies compa-
as follows: red with the test results. As correc-
tion factor to the forces and moments
X2 = SCA the following equation is used

Y2
=f D2Dcos(0c + y )d1
c
Yc = 25r' + 101 (19)

Z2
=f D2Dsin(Oc + y )d1
c
(15) NUMERICAL RESULTS
K2 = 0
To validate the present method,
M2 = -f xD2Dsin(Oc + yc)d1 the computed forces and moments on a
body of revolution with an L/D=11 are
N2
=f xD2Dcos(0c + y )d1
c
compared with the experimental results
which will be published in the future.
where 0c is the angular location of the The Reynolds number for the experi-
direction of U2 in the crossflow section ments was 14 million, and this value
and y is correction to shown in was used in performing the computa-
0c
Figuré 2. tions. Comparisons have been made for
four turning rates (r'), five pitch
angles (0), and five drift angles (110.
TOTAL FORCES AND MOMENTS The overall agreement between the pre-
dictions and the experimental data was
The total forces and moments are very good. In this paper, a compari-
the addition of Equations (6) and (15). son is made between the computations
and the data for two conditions,
F = g1 + g2 (16) namely, (1) r'=0.74 and 0=-12; and (2)
r'= 0.42 and 0=-4. Figures 3 and 4
are comparisons between the computa-
SEPARATION ANGLES AND CORRECTION tions and the experimental data for
FACTOR the first and second conditions,
respectively. As can be seen, there
The two-dimensional separation is good agreement with the experimen-
angles are empirically determined based tal results for the entire range of
on the test data of forces and moments angles of drift.
and they are optimal for the present
study if expressed as

115
CONCLUSIONS
Inviscid forces and moments
The present method is based on (=X1,Y1,Z1,K1,M1,N1)
the potential theory and boundary layer Viscous forces and moments
method. The results computed for the
(=X2'Y2'Z2'K2'M2'N2)
body of revolution with the present
method agree very well with those of (Fx'Fy,Fz) = (X',Y',Z')pL2U2/2
o
test. To be able to derive the hydro-
dynamic coefficients for the stability Gravitational acceleration
and maneuvering, the computation of Length of body
forces and moments of fully appended
body of revolution is under progress. (M ,M ,M ) = (K',W,N1)pL3U2/2
x y z o

ACKNOWLEDGMENT n Unit normal vector

Pressure
The author acknowledges the sup- Sectional radius of body
port of Dr. J. Feldman and Mr. D. r' Turning rate (=L/R)
Goldstein. This work was sponsored by Radius for the rotational
the Defense Advanced Research Projects motion
Agency (DARPA). Reynolds number
Re = UoL/v
S Body surface area
REFERENCES Uo = Rw Velocity of body at C.G.
U Velocity at two-dimensional
sp
Hong, Y.S., "Computation of the Hy- separation
drodynamic Forces and Moments on a Body Ut Velocity at the body tail
of Revolution With and Without Appen- Axial free stream velocity
U1
dages," David Taylor Model Basin, Car- Crossflow free stream velo-
U2
derock Division Headquarters, Naval city
Surface Warfare Center Report No. Fluid density
DTRC/SHD-1003-06, August 1991 Viscosity of the fluid
Young, A.D., "The Calculation of the Momentum area
Total and Skin Friction Drags of Bodies Momentum area at body tail
of Revolution at Zero Incidence," Tech- Strength of source and sink
nical Report of the Aeronautical Rese- Angular velocity of body
arch Committee R&M 1874, 1939 Pitch angle
Angle of two-dimensional
separation
NOMENCLATURE Drift angle
Total, undisturbed, and
Total axial drag disturbance potentials
CA
C Pressure coefficient
DP Beam or Diameter
Frictional Drag
DF
Form drag
DP
Crossflow drag
D20
Total forces and moments
(=F ,F ,F ,M ,My ,M z )
x y z x

116
 Uo = Rw

x, Fx

001 = R

Turning rate (r') = L/R

01
z

Figure 1 - Coordinate system

Figure 2 - Crossflow drag

117
 LEGEND
o = COMPUTATION
o - EXPERIMENT

D
cri

rb

o
pffj
r; o

0.0 10.0 20.0 0.0 20.0 0.0 20.0

L! (deg.) p (deg.)
CN

CIS

C\1 U o 6

rb
cf" c

C\1

20.0 10.0 20.0

0.0 10.0 20.0 0.0 10.0

p (deg.) P (deg.) p (deg.)

WD=11 for
Figure 3 - Forces and moments of a body of revolution with
turning rate of 0.74 and pitch angle of -12 degrees

118
 LEGEND
o = COMPUTATION
o - EXPERIMENT

Cr)

d-e CC; o
o
o
t,c)

-
o o
o o

0.0 10.0 20.0 0.0 20.0 0.0 10.0 20.0

p (deg.) p (deg.) p (deg.)
.-

rb 8
o C,C1 C

Q Q

0.0 10.0 26.0 10.0 10.0 26.0 10.0 20.0

p (deg.) p (deg.) p (deg.)

Figure 4 - Forces and moments of a body of revolution

with L/D=11 for
turning rate of 0.42 and pitch angle of -4 degrees

119
 Resistance Tests with UBC Series Fishing Vessels
S.M. Calisal, J. Mikkelsen (University of British Columbia, Canada)
D. McGreer (Kvaerner Masa Marine Inc., Canada)

Abstract
Introduction
Resistance model test results for
a systematic series of low L/B, heavy- Model series have proven to be
displacement vessels are presented. The very useful tools for predicting the
UBC Series is based on westcoast seiners resistance of ships. As part of a
and trawlers. These vessels have low research program to reduce fuel
L/B and L/V113 values that are outside consumption of fishing vessels a
systematic model series has been
the range of existing model series data. developed at the University of British
A parent hull form was developed that Columbia based on a
has 14% less resistance and yet has the typical British
Columbia purse seiner.
same displacement as a typical fishing
vessel. A series of 13 models was
generated by systematically varying L/B, During the development of the UBC
BIT and CID. FISH [1] fuel consumption prediction
Results from resistance
program it was found that the resistance
tests in calm water are presented. From of many westcoast fishing vessels cannot
these model tests, an algorithm has been be predicted by existing model series
developed to predict the resistance of data (see Table 1). Westcoast vessels
similar small vessels. The typically have lower length to beam
implementation of bulbous bows to the ratios and lower length to volume ratios
parent hull shows reduction in the than previously reported model series:
overall resistance by approximately 17%. BSRA Trawler Series [2,3], Webb Trawler
Also, the introduction of stern fairing Series [4,5,6],
reduces the resistance by about 25% at and the ITU Series [7].
the design speed. Series CI, L/B BIT L/V1/3
UBC .53 2.6 - 4 2 - 3 3 -
-.61 4.47
Nomenclature BSRA .5 - 4.3-5.8 2 - 3 4.35 -
.63 5.1
beam ITU .35 3.3 - 5 2 - 3.2 3.4 -
Cb block coefficient -.56 6.1
Cm midship area coefficient Webb .42 3.2 - 2.3 3.85 -
Cp -.53 5.75 5.22
prismatic coefficient
Cr resistance coefficient Table 1. Comparison of
Fn Fishing vessel
Froude number Parameter Ranges
gravitational constant
wave number The trend for new fishing vessels
in British Columbia, is to lower length
waterline length to beam ratios. The advantages of a
LCB longitudinal center of buoyancy larger beam are: increased fish hold
total resistance volume, increased stability for hauling
draft at midship excluding keel the net, larger deck area and the
V displaced volume ability to carry a longer net drum. The
we encounter frequency disadvantage of larger beam is increased
W, resistance. The series data could be
wetted surface area used to calculate the tradeoffs of
increasing beam with increasing
resistance. Also the series data will
121
be useful in calculating the resistance 25.4 mm (1 in) centers 10% of LBP aft of
of many existing small vessels. the forward perpendicular. The pins
were 3.18 mm (.125 in) in diameter and
A systematic series of models was extended 2.54 mm (.1 in) from the hull
generated from a parent hull form. The surface.
series parent was developed by testing
stern and bow The towing tank procedure was
different chine,
configurations to find a hull with the first calibrated using a Series 60 block
lowest resistance. This hull was 60 hull model. Tests of both calm water
selected as the series parent and was resistance and response in regular waves
scaled to generate a systematic series were conducted in order to be certain
of 13 models. The series was created by the tank would accurately predict the
varying the length to beam ratio, beam results for the UBC Series models. The
to draft ratio and the block results of the calm water tests are
coefficient. shown on Figure 1. The data is plotted
against Series 60 data supplied by the
David Taylor Marine Basin. The results
A resistance prediction algorithm show good correlation between both tanks
has been developed by a regression
with minimal separation in the data.
analysis of the model test results. The
algorithm is suitable for implementation
on computer and is presently used in the
UBC FISH program.

Model Test Procedure

eux,
The resistance experiments were
conducted at the BC Research Ocean LIJ
Engineering Centre located adjacent to
the UBC campus. The towing tank is
67.06 m (220 ft) long, 3.66 m (12 ft)
wide and 2.44 m (8 ft) deep. The models
were constructed of wood and were
between 1.52 m (5 ft) and 2.13 m (7
SPEED (knots)
ft) long. They were built at a scale of
13.75:1.
Figure 1. Resistance Tests of Series 60
The models were tested at about 10 Hull
speeds in the Froude number range of .2
to .45 which correspond to full scale Tests of the response of the model
speeds of between 6 and 12 knots. Two in regular seas is shown on Figures 2
drafts were tested, a lightship draft and 3. The collected data is compared
and a loaded draft. The lightship draft with data collected at the Delft
is defined as the draft the vessel would Shipbuilding Laboratory on a Series 60
have when departing for the fishing model [8]. The collected heave and pitch
rounds. The fish holds are assumed to
be empty and a full supply of fuel and 18 Delft ShipOufichng

stores is assumed. At the loaded draft, 16

Labarotry

it is assumed that the fish holds would E 14

a C.Research
be full and a half supply of fuel and Dell Data-10%
4./ 12
stores would be on board.
1 Dent Data .10%

The modelswere towed from a ? 08

E
bracket located at midship. The bracket < 06
was fastened as low as possible so that 04
trim moments caused by towing would be E

minimized. The models were free to 02

heave and trim. Resistance was measured
by a strain gauge force dynamometer and 30 4 55 6
25
the sinkage and trim were measured by
high resolution potentiometers. To
60
induce a turbulent boundary layer on the Figure 2. Heave Response of Series
Model at Froude Number = 0.30
models, brass studs were inserted at

122
 Purse seining is a stationary
1.6 -
fishing method in which a large net is
e 14-
Reference Data
deployed in a circle surrounding a
G
E 12-.
school of fish. The net is then drawn
8 C Research
".
closed on the bottom and the net reeled
0 in over the stern.
Deffl Data, 1 V. These vessels
08
typically have low LIS ratios for large
Dettt Data-10%
hold capacity, maneuverability and
06- stability.
04-
:.7!

0.2 -
A number of model tests were
conducted to ensure that the parent hull
0 would have good resistance
25 3 35 4 45 55 65
IL
5 6
characteristics. The first experiment
e was to determine if modifying the hull
to have a double chine or round bilge
Figure 3. Pitch Response of Series 60 would reduce the resistance. Two models
Model at Froude Number = 0.30 were designed so that the area of each
station and waterline breadth at each
response data is mostly contained within station remained constant. This ensured
a 10% band width of the Delft data. that all the sectional area curve,
Discrepancies between the two sets of displacement, form coefficients, and
data are probably due to discrepancies center of buoyancy remained unchanged.
in the tested Series 60 model. The model The wetted surface also did not change
tested at B.C. Research was slightly significantly. The wetted surface of
altered from the standard Series 60 the double chine hull was 1.2% greater
Model. The results for both calm water than the single chine hull and the round
resistance and regular sea tests of bilge hull was .2% greater than the
standardized hulls shows that the tank single chine hull. The body plans of
will provide reliable resistance data the three hulls are shown in Figure 4. A
for the UBC Series fishing vessel significant reduction in resistance was
models. achieved with the double chine and round
bilge designs (see Figure 5). At a
Parent Hull Development typical cruising speed of 10.5 knots the
The first step in the development resistance was reduced 8% for the double
of the UBC Series was to develop the chine hull and 10% for the round bilge
parent hull form. hull. Since the double chine hull had
The starting point performance
for the development of a parent for the that nearly matched the
series was a fishing vessel hull that round bilge design, the double chine
had been designed by Gerry Stensgaard at hull was selected for further refinement
BC Research. It was a 21.34 m (80 ft) because it probably would cost less to
purse seiner typical construct.
of steel and
aluminum vessels currently being
constructed on the west coast of North Additional experiments where then
America for purse seining of salmon and conducted to improve the double chine
herring.

Figure 4. Body Plans of the Single Chine, Double Chine and Round Bilge Models

123
 A new bow shape was developed 700 -
hull.
that reduced the half entrance angle
from 42 degrees to 30 degrees. The 600 Double Chine
sectional areas of each station were
again held constant but the design New Bow & Stem
waterline half- breadths were reduced in 500
the bow region to reduce the entrance 0 Series Parent
angle. To maintain the same sectional
bow was 400
area curve, the profile of the
made deeper and the stem was made EHP
vertical below the design waterline. 300
The stern profile was also modified.
The run angle was decreased by making
the buttocks slightly concave. 200
700
100
600
Single Chine

7n1 Double Chine o

500 5 6 7 8 9 10 11 12 13
Round Bilge SPEED (knots)
400 Figure 6. Resistance Optimization of
Parent Hull
EHP
300 depth to generate models in a systematic
series of 7 models. The L/B ratio was
increased by 30% and decreased by 15%.
200
The B/T ratio was both increased and
decreased by 20%. The models were
100 generated by holding the beam constant
and scaling the length to change the L/B
ratio and scaling the depth to change
the BIT ratio. This could easily be done
6 7 8 9 10 11 12 13 by computer.
SPEED (knots)
Figure 5. Ef fect of Chine Conf iguration The geometric properties of the
on Total Resistance models are given in Table 2. The Parent
Hull is Model 1. Scaling the length and
The resistance decreased slightly depth of the model did not alter Cb, Cm,
when compared to the previously tested C or LCB/L. In order to determine the
double chine hull. measurements of the effect of changing the block coefficient
wave profiles indicated that the bow scaling function was developed that
a
wave had been reduced by the new bow when applied to the parent hull offsets
shape but, since the reduction in
decreased the block coefficient by 14%
resistance was less than expected, it at the loaded draft and 16% at the
was concluded that the new bow may have lightship draft.
had favorable resistance characteristics L.J7",
A new model Wdel L ws Lli3 BIT Cb CP
but the new stern did not.
(sA m.)
was constructed that had the new bow and 1
(m.)
1552 1.125 3.06 249 0.615 0.700 3.36
the original double chine model stern. 2 1.319 0.956 2.60 2.49 0.615 0.700
0.700
3.01
4.00
at 10.5 2.49 0.615
This configuration achieved 3 2.017 1.462 3.98
0.700 3.57
0.948 3.06 2.99 0.615
knots a reduction of 7% over the double 4 1.552
1.99 0.615 0.700 3.12
5 1.552 1.124 3.06
chine hull or 14% when compared to the 6 2.017 1232 3.98 2.99 0.615 0.700 4.25
This 0.531 0.653 3.53
single chine hull (see figure 6). 7 1.552 1.016 3.06 2.49
0.531 0.653 420
3.98 2.49
hull was then selected to be the parent 8 2.017 1.321
106 2.99 0.531 0.653 175
hull form for the UBC fishing vessel
9
lo
1.6.2
tssz
0.620
liso 106 199 0.531 0.663 IV
series. 11 2.017 1.461 I% 1.49 0.615 0.700 171

12 2.017 1208 I% 199 0.531 0.653 4.46

1.496 I% 1.99 0.631 0.653 190
13 2.017
Tests of Series Models
design was Table 2. Geometric Properties of UBC
Once the parent hull
established it was scaled in length and Series

124
 Scaling the hull to reduce the
block coefficient also reduced the i= 1 2 3 4
midship section coefficient and the
prismatic coefficient. The scaling did
not change the LCB or LOE significantly. d1,0 .074654 .076958 -.000162 -.018642
The new block coefficient hull was then
scaled in the same way as the original
block coefficient hull to generate 6 di,1 .001879 7.7 10-5 7.42 10-5 1.05 10-5
additional models with different L/B and
BIT ratios.
di,2 -.000701 1.8 10-5 .000907 .000119
Development of Resistance Algorithm

Two resistance algorithms have

di,3 -.05158 -.005247 -.001137 -.001944
been
developed based on regression analysis
of the model test data. The first
di,4 .009871 .000572 -.000661 .001169
algorithm is based on the equation
developed by Oortmerssen [9] for the
analysis of small vessels and the second
is an equation developed at the Table 3. Coefficients for Resistance
Institute of Marine Dynamics for Algorithm for Cn=0.615
semiplaning hulls. The Oortmerssen
equation was found to have the best fit An example plot of the accuracy of
for the loaded draft resistance data and the resistance algorithms is shown on
the IMD equation was best for the Figure 7. For the Parent Hull model,
lightship draft data. residual resistance coefficient, Cr,
versus Froude number is plotted. The
The Oortmerssen equation is based on calculated resistance values are
Havelock's equation for wave resistance reasonably close to the model data for
of a two-dimensional pressure all of the models. For the Cn=0.615
disturbance. The vessel is replaced by
a pressure-point at the bow and a
equation, 86 regression data points were
negative pressure-point at the stern. used resulting in an average error in
The equation is as follows: prediction of the total resistance of
-m -m -m 4.6%. There are two limitations of the
this resistance algorithm. The algorithm
Cr = C1e9Fn2 + C2eFn 2 + C3eFn2 sinFn2 + smooths out the hump in the resistance
-m curve at the Froude number of around .35
and the algorithm is not accurate above
C4eFn2cosFn2 a Froude number of .425.
where: 35

L B
Ci = d + ui,lg + ui,2( T) 2i3O + ui,3T 30

2
25
+ di,4(T)
and 20
Cr
m = 0.14347 C -2.1976
15

The coefficients d1 10
determined from regression analysis of
the model test data. Two sets of 5
coefficients were computed, one for each
block coefficient. This was found to o
give a much better fit to the model test
0.2 0.25 0.3 0.35 0.4 0.45
data. An example set of coefficients for
one block coefficient, Cb=0.615, are Fn
given in Table 3. Oortmerssen's Figure 7. Total Resistance versus Froude
coefficients for the equation for m were Number for Parent Hull Model
used in the analysis.

125
 700
The Effect of Bulbous Bows on the UBC
Series 600
Further testing of the UBC Series
parent has shown that it is responsive 500
to the addition of a bulbous bow.
Several experiments were conducted with
side-bulbs [10]. Side bulbs protrude to 400
the side rather than forward as with a EHP
conventional bulbous bow (see Figure 8).
300
Four bulb shapes were developed for the
UBC Series parent using the OPTIHULL
computer program developed at UBC. 200

The mathematical methods used to

develop the side bulb designs are 100
described in reference [10]. Briefly the
method optimizes an objective function o
6 7 8 9 10 11 12 13
SPEED (knots)
Figure 9. UBC Series with Side Bulb

The results for Bulb 2 are given

in figure 9. At speeds above 9 knots
the resistance in substantially reduced.
At a typical cruising speed of 10.5
knots the resistance was reduced 16.6%.
At speed below 9 knots the resistance
increased somewhat. Experimentally
observed wave breaking aft of the side
bulb seems to have caused the increase
in resistance at low speeds.

The Effect of Stern Fairing on the

UBC Series
During resistance tests of the UBC
Series, it was noted that the eddy
making took place behind the immersed
transom. If some of the pressure could
be recovered through the use of a faired
stern, it would reduce the overall
resistance of the vessel.

Figure 8. Lines Plan for Side Bulb 2 Results of stern fairing tests
are shown on Figure 10. The results show
which is the sum of the frictional a large decrease in hull resistance over
resistance and the wave resistance. The the entire speed range. At the design
frictional resistance is calculated by speed of 10.5 knots, this reduction is
the ITTC-1957 formula and the wave approx. 25%. Tests performed on the
resistance is calculated by Michell's parent hull with a bulbous bow addition
Integral. The hull surface is show that stern fairing will further
represented by tent functions which reduce the hull resistance. An overall
leads to a resistance eauation that is a hull resistance reduction of 30% at
function of the hull offsets. The design speed will occur with the
quadratic programming method is used to combination of a side bulb and stern
find the offsets that give the minimum fairing. Pressure taps were installed
resistance. Constraints are used in the through out the stern fairing to verify
optimization to limit the offsets to that virtually all the decrease in drag
practical values. is due to a change in stern force.

126
 References
700

600
Parent Hull 1 Calisal, S.M., McGreer, D.,
a-- Stern Fairing Rohling, G.F., "A Fishing Vessel Energy
500 Analysis Program", Marine Technology,
Stern Fairing SNAME, Vol 26, No. 1, Jan 1989, pp. 62-
and Side Butt
73.
ELI

300 2 Pattullo, R.N.M., and Thomson,

G.R., "The BSRA Trawler Series (Part
200 I)", Transactions of the RINA, Vol 107,
100
1965, pp. 215-241.
3 Pattullo, R.N.M., "The BSRA
Trawler Series (Part II)", Transactions
5 6 7 8 9 10 12 13 of the RINA, Vol 110, 1968, pp. 151-185.
SPEED (Knots) 4 Ridgley-Nevitt, C., "The
Resistance of Trawler Hull Forms of .65
Figure 10. Results of Stern Fairing on Prismatic Coefficient", Transactions of
Parent Hull the SNAME, Vol 64, 1956, pp. 433-468.
5 Ridgley-Nevitt, C., "The
Conclusions Development of Parent Hulls for a High
Displacement-Length Series of Trawler
The resistance algorithm presented in Forms", Transactions of the SNAME, Vol
this paper will be useful for estimating 71, 1963, pp. 5-30.
the resistance of low L/B displacement 6 Ridgley-Nevitt, C., "The
vessels. These data have not Resistance of a High Displacement-Length
previously been available for these Ratio Trawler Series", Transactions of
vessels. The systematic series is based the SNAME, 1967, pp. 51-77.
on a westcoast seiner however it is 7 Kafali, K., "A Study on the
applicable to other types of small Fishing Vessels Forms", Turkish Ship
vessels. Research Institute, Publication No. 25,
Technical University of Istanbul, 1980.
Tests to optimize the hull form of 8 Gerritsma, I.J., "Shipmotion in
the UBC series parent have indicated
Longtitudinal Waves", International
that significant reductions in Shipbuilding Progress, Vol. 7, 1960
resistance are possible for these Van
vessels.
9 Oortmerssen, G., "A Power
Using a double chine rather Prediction Method and Its Application to
than single chine reduced the resistance Small Ships", International Shipbuilding
8% and reducing the entrance angle Progress, 1971.
reduce the resistance 7%. Additional 10 Calisal, S.M., and Goren, O.,
reductions in resistances of 16.6% was "Optimal Hull Forms for Fishing
achieved with a bulbous bows effective Vessels", Thirteenth
in multiple speeds STAR Symposium,
and multiple load SNAME, pp. 41-51.
configurations. The addition of stern
fairing will significantly reduce the
pressure drag on the stern which can
reduce the drag coefficient by about
25%.

Acknowledgments
The authors would like to thank
Energy Mines and Resources and
Department of Fisheries and Oceans for
funding the research program. They also
would like to thank Gireesh Sadasivan
for his work optimizing the parent hull
design. Many people assisted in the
research project including especially
Bob McIlwaine of Pacific Fisheries R&D
Ltd., Gerry Stensgaard, and George
Roddan of BC Research Ocean Engineering
Centre, as well as Leon Barnard and Y.C.
Chow of UBC.

127
 Comparison of Wake Flow Measurements between
5-Hole Pitot Probes and Laser Doppler Velocimetry
in the Ship Propeller Plane
M. Pirrone, W. Lindenmuth
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT U0 Free stream velocity (model tow

speed).
Wake flow measurements have Up Axial or stream-wise component of
traditionally been made with pitot flow.
probes. Most commonly used probes are of UAG Ideal axial component of flow
the 5-hole spherical and hemispherical
based upon geometric angle to free
head on cylindrical stem variety. A stream velocity.
limitation is that the measurement can
be intrusive and the probe may not
UT Tangential component of flow
physically fit in the area to be perpendicular to propeller blade
measured. axis.
One modern solution uses UR
multi-component Laser Radial component of flow parallel
Doppler
Velocimetry (LDV) to measure the wake to propeller blade axis.
profile. A question arises as to how the r/Rp Ratio of measured radius to
laser measurements compare with the propeller radius.
pitot probe measurements.
INTRODUCT ION
Experimental studies were conducted
at the David Taylor Model Basin, In the past, the David Taylor Model
Carderock Division, Naval Surface Basin (DTMB), collected wake survey and
Warfare Center (formerly the David wake flow data using 5-hole pitot probes
Taylor Research Center) where pitot with spherical or hemispherical heads.
probes and LDV measurements were More recently, Laser Doppler Velocimetry
simultaneously collected in the
(LDV) technology has matured. Non-
propeller plane of a single screw ship intrusive flow measurement techniques
model. Non-simultaneous wake have become commonplace, but have not
measurements have also been collected on replaced pitot probe measurements. Pitot
twin screw shaft and strut ship models. measurements are significantly less
expensive to make and don't need special
The two flow measurements correlate operator training and handling.
well at the outer propeller radii. Near
the hub or in low-flow regions, the Pitot data were traditionally
pitot probe axial velocities fall off analyzed using Pien's(1) methods for
more rapidly than LDV measurements. flow around five-hole spherical head
probes. Pien's derivation of the
NOMENCLATURE pressure-velocity relationship was based
on the potential flow around a sphere
All nomenclature corresponds to ITTC and assumed no interaction between the
standard definitions. probe pitch and yaw planes. With no
interaction, the flow in each plane can
LDV Laser Doppler Velocimetry. be decoupled from the other plane. This
Two-axis method Pitot tube calibration method is termed the "two-axis" method.
method based on Pien(1). At high angles of inflow, the decoupling
Matrix method Pitot tube calibration assumed by the two-axis analysis is not
method based on Treaster valid. For example, the two-axis method
and Yocum(2). derives longitudinal velocities
exceeding free stream at locations where
the flow should not be accelerated.

129
 Each pitot probe hole is connected
More sophisticated methods of to a separate differential pressure
calibration and analysis have been transducer, connected to a reference
developed for pitot probes. The method pressure source. The pressure data are
published by Treaster and Yocum(2) amplified, conditioned, converted to
accounts for interactions between the digital format, and stored in computer
radial and tangential measurement files. The pitot probe surveys were
planes. This method is termed the analyzed with both the two-axis method
"matrix" method. and the matrix method.
LDV wake surveys were made with a
The first objective of this study is three-component laser system built by
to make a direct comparison of flows TSI, see figure 4. Due to a shadowing
derived by the two analysis methods in effect caused by the propeller shaft,
order to assess whether the matrix only one half of the measurement plane
method is, indeed, more accurate. The could be collected. For the single-screw
second objective is to compare wake data model, the LDV and pitot data were
gathered using pitot probes with data collected simultaneously. The LDV survey
gathered using the LDV laser system. plane was offset one inch ahead of the
Advantages and disadvantages of both tip of the pitot probes, to avoid probe
interference with the LDV measurements.
systems are presented.
The twin-screw LDV data were not
To address the two objectives: (1) collected simultaneously with the pitot
rake surveys, The LDV data were offset
Open water pitot tube calibration longitudinally
verification data were collected in the approximately an inch
tow tank. The results were analyzed and up to 0.3 inches radially.
using both the two-axis method and the
matrix method, allowing a comparison of An open water verification
the two methods. (2) Wake surveys of experiment was also conducted for the
single and twin-screw naval ship models pitot probes, to enable a comparison of
were collected with pitot probes and the matrix method and two-axis method in
LDV. (3) The experimental uncertainties uniform flow. Open water tests were
conducted by attaching the pitot probe
of the pitot and LDV methods were
assessed. (4) The results were compared to a towing strut. Pitch angle was
between varied by tilting the strut, and
to show the differences with an
measurements with the two instrument measuring the angle
inclinometer. The yaw angle of the probe
systems. These comparisons are ongoing,
originally started several years ago. As was set with a protractor. Three pitch
the equipment and procedures are refined angles (0.0, +8.5 and -8.5 degrees) and
with each passing year, we expect the two yaw angles (0.0 and +5.0 degrees)
differences between the two measurement were used. The calibration curves were
calculated previously. The matrix method
systems to decrease.
calibrations were conducted in air.
EXPERIMENTAL SET UP Subsequent water calibrations showed
little difference.
The wake survey experiments were
conducted on Towing Carriage II at the RESULTS AND DISCUSSION
David Taylor Model Basin with both The open water pitot test analyses
single and twin screw naval ship models.
Stern bodyplan lines of the two models are summarized in Table 1. The values
represent the difference between the
are shown in figures 1 and 2.
Five-hole cylindrical pitot probes measured velocity component and the
idealized (geometric) velocity
with a 1/8-inch diameter hemispherical component, normalized by the idealized
head are precisely mounted onto a
"rake", which rotates in the propeller axial velocity. This check test used
plane, see figure 3. The model was towed calibration curves previously obtained
at a steady speed during data with these tubes for use in wake
surveys. The data confirm that the
collection. matrix method produces axial velocities

130
much closer to the ideal velocities than axis method, and have been corrected to
the two-axis method. This is likely due matrix method. The LDV and Pitot matrix
to the fact that the matrix method results are in excellent agreement.
accounts for pitch-yaw interaction, Notice that there is no area of very low
whereas the two-axis method does not. axial flow. This is because the
propeller is not obstructed by any hull
Wake survey data from simultaneous surfaces such as skegs. The
measurements on the single screw model's corresponding LDV data plane lies about
propeller half-plane are shown in an inch ahead of the propeller disk with
Figures 5 and 6, at the propeller tip radius r/Rp = 0.92. Thus, the two sets
radius (r/Rp = 1.0) and near the hub of propeller plane nominal wake data are
(r/Rp = 0.29), respectively. Data were offset both longitudinally and radially.
reduced using the pitot probe matrix
method and LDV techniques. The pitot probe data show features
in the wake that are attributed to the
The r/Rp = 1.0 results exhibit good shaft struts, i.e., at 0=10°. They are
agreement except for the 0-25° region. not evident in the comparatively coarse
In this area the pitot probe measurement (10°) LDV data grid and likely were
falls off more rapidly, and approaches missed.
zero. The LDV method measures a higher
axial velocity UA throughout this range. MEASUREMENT UNCERTAINTY
One important consideration for this
phenomenon is that the model hull has a Measurement uncertainty is expressed
center skeg that surrounds both the top with a 95% confidence limit. Individual
(0°) and the bottom (180°) of the survey uncertainties are composed of a bias
plane. At these angles, the probe is component and a precision component. The
quite close to the skeg. Consequently, total uncertainty is the root sum square
there is a large flow deficit. It is of individual components. Two thousand
felt that the difference between the two pitot samples are averaged for each data
measurements in the 0-25° reduced flow spot, minimizing precision uncertainty.
region is due to the following reasons:
In this low-flow region, the 1 inch The overall pitot measurement
spatial offset between the two uncertainty results from the following
measurement systems can significantly sources: probe geometry and positioning,
affect the results. Additionally, in tubing volume flow rates, pressure gauge
axial flow regimes approaching zero, the sensitivities, calibration procedures,
pitot measurement uncertainty grows and signal conditioning/instrumentation.
quite large. This error is due to
pressure gauge sensitivity, and the Tube flow uncertainty is only
increased time required to achieve significant for axial flows near zero
steady-state pressures in this low-flow percent of free-stream. In these flow
regime. regimes, more time is required to
achieve steady-state values.
For the radius r/Rp = 0.29, the
results deviate more. The nearness of Pressure gauge sensitivity is 0.5%
the hub at this radius influences the of the full-scale value of 5 PSI
pitot probe response, see Figure 6. differential. At low velocity regimes,
Because the LDV measurements are non- this increases to about 1.5%. Geometry
intrusive and are not affected by the and positioning error is within 1/2
presence of the hub, it is reasonable to degree, 0.8%. Instruments contribute 1%
accept the LDV result as the more error, calibration procedures 1%.
credible one. Overall uncertainty is approximately
2.2% for the pitot probes.
Figure 7 depicts a wake survey
measurement in the propeller plane of a There are up to thirteen sources of
twin-screw destroyer model at a radius error in LDV velocity measurements. Only
r/Rp = 0.96. The measured pressures were the most significant ones are discussed.
originally reduced using only the two-

131
 Reflective seed is used to provide two methods can be as high as 3% in the
back-scatter targets for LDV. Seed range of pitch and yaw angles studied.
particle size and density affect their
trajectory relative to the flow field. Table 2 summarizes the relative
The requirements are more exacting in strengths of pitot and LDV measurements.
highly sheared flows, such as in Pitot measurements are cheaper, quicker,
boundary layers. Seed particles present and require less specialized training.
no significant limitation here. For high axial flow regions they are
quite adequate. LDV measurements are
The LDV strut causes a disturbance preferred in low-flow situations, where
that can interact with flow around the high turbulence is expected, and near
ship model. In open water flow, measured surfaces that would cause interaction.
strut disturbance is only about 0.1% of
Uo, but it's effect is as much as 2% of Matrix method pitot probe analysis
U0 in measurements near the hull. and the three component LDV can both
measure propeller plane nominal wake to
about 2.2% uncertainty. A greater
System noise includes the effects
deviation between LDV and matrix method
carriage vibrations and processing results is evident near the hub, where
uncertainties. These are important for
the non-intrusive LDV measurements are
assessing flow turbulence, but their
more credible. Measurement techniques
effect is minimal on mean velocity with both pitot tubes and laser methods
measurements owing to the large (256)
are advancing in accuracy. If only the
number of samples taken. The estimated
mean flow quantities are needed, the
error is 0.1% of U0 for the axial pitot probe survey is more economical.
component of flow and 0.3% for the worst
Detailed flow parameters, such as the
cross flow component. Reynolds stresses and turbulence are
better addressed using the LDV system.
Misalignment of the five laser beams this effort at
As stated earlier,
cause bias errors and calibration factor comparing measurement methods continues.
errors. These errors are typically less Refinements are expected which would
than 2% of Uo, and are evident in open decrease the differences between the two
water calibration runs. Therefore, they measurement systems.
can be accounted for in the data
analysis. ACKNOWLEDGEMENTS
Structural bending of the LDV system The authors wish to thank Han-Lieh
supports cause position errors that are Liu, Toby Ratclif fe, and Gabor Karafiath
accounted for in processing the data. for their contributions to this paper.
Positioning accuracy is 1.3 mm over the Thanks to Lisa Fisher for typesetting.
entire traverse range and is comparable
to movements of the model when it is REFERENCES
being towed.
Pien, P.C., "Five Hole Spherical
Overall LDV measurement uncertainty Pitot Tube", Report DTNSRDC 1229, (May
is estimated to be 2% of free-stream
1958).
velocity.
Treaster, A.L., and Yocum, A. M.,
CONCLUSIONS "The Calibration and Application of
Five-Hole Probes", ISA Transactions Vol.
The open water pitot verification 18, No. 3. (Mar 1979)
data shows that the matrix calibration
method is preferred to the two-axis Fry, D.J., "DTNSRDC Three Component
calibration method because it yields Laser Doppler Velocimetry: Towing Tank
results much closer to the actual flows. System Measurement Errors", Report
Normalized axial velocities are not DTNSRDC/SPD 1163-03 (Oct. 1985).
greater than one as was derived from the
two-axis calibration calibration method.
The difference in values between these

132
 '99M.

-a-- a 2 in

:00 in. ill

if
nl
Ill
i i
II la
ill in in
In Ill
III Il iiiir
III III
i
__u____
,.e number
11 YY YY YY Y
1111,
YY
i i
9 7 5 3 I 24 6 8 TO

10 25 ;1`,..es

0.nm.

Figure 1. Single screw ship stern lines.

Figure 3. Typical pitot probe and rake.

4_ ,7KPROPELLER DIA.
180°

Figure 2. Twin screw ship stern lines Figure 4. LDV strut arrangement.

133
 1.2
1.2

1
1

0.8
0.8
0 0.6
o 0.6

P 0.4
P 0.4

0.2
0.2
tx 48,

O XX
o
vr off o c'
01) c)ir
-0.2
I*49:1
Fir II f
't 04'
It f ilf
41 t;;11:,
I Iii
-0.2 20 40 60 80 100 120 1 40 160 1E
0
0 20 40 60 80 100 120 140 160 180
Angle
Angle Figure 7. Twin Screw r/Rp = .96 Pitot &
Figure 5. Single Screw r/Rp = 1.0
0.92 LDV

Ua Pitot

Ut Pitot

A Ur Pitot
Ua LDV
; ..
fit)
6.4',T.'.1h1.,:f7R, r
erklillWoses,-dr41, Ut LDV
411111t Ur LDV
0 20 40 60 80 100 120 140 160 180
An le
0.2

o
01A...474v...A A 4 ! AAAR AM
Leerzl,Trphylit TY*,

-0.2
0 20 40 60 80 100 120 140 160 180
Angle
Figure 6. Single Screw r/Rp = 0.29

134
 AUA/UAG AUT/UAG AUR/UAG
0° pitch 0° yaw 2 axis 2.1 2.4 0.4
matrix -0.5 1.5 0.8
+8.5 pitch 00 yaw 2 axis 2.4 5.6 -3.8
matrix -0.4 7.4 -0.3
-8.5° pitch 0° yaw 2 axis 3.8 ' 5.2 -5.3
matrix -1.9 4.8 -0.3
00 pitch +50 yaw 2 axis 2.9 0.5 0.7
matrix -0.3 -0.2 1.2
Table 1. Percent velocity difference between measured and actual inflow
velocities normalized by true geometric axial velocity.

PITOT LDV
SET UP TIME Minimal - 1 day. A medium amount of set-up and
alignment time is required -1 week.
OPERATOR TRAINING Little extra training required. Extensive laser training and experience
REQUIRED required.
EQUIPMENT & Small initial investment Large initial investment
OPERATING COSTS Low operating costs. Can Low-medium operating costs.
easily survey complete plane. Difficult to survey complete plane (can
only survey half of plane).
ACCURACY Good accuracy in higher Better cross flow resolution.
velocity flow areas away from Better accuracy in very low velocity
surfaces. flow/high turbulence areas or where
wall interactions are expected.
LIMITATIONS/SPECIAL Cannot collect wake inflow Allows wake data to be collecte,d in
CAPABILITY data in front of an operating front of an operating propeller.
propeller.

Table 2. Summary of strengths of pitot and LDV measurement methods.

135
SESSION II-B

PROPULSION
 PROPULSION

TECHNICAL COMMITTEE MEMBERS

Ki-Han Kim (Chairman) David Taylor Model Basin

Terry Brockett Arete
William Day David Taylor Model Basin
Walter Gearhart Pennsylvania State University
David S. Greeley Atlantic Applied Research Corp.
David HaIly Defense Research Establishment
Atlantic (Canada)
David Keenan Massachusetts Institute of Technology
Frederick Stern University of Iowa
James Uhlman, Jr. Naval Undersea Warfare Center
William Vorus University of Michigan

138
 The Propulsion Committee Report
K-H. Kim (Carderock Division, Naval Surface Warfare Center, USA)

INTRODUCTION was reported by Dai [8] with a design procedure for the
stern and the propulsor as an integrated system. No experi-
Since the 22nd ATTC conference in Newfoundland in mental data are available in the open literature for the
1989, five conferences of direct interest to the Committee propulsor behind a full-stem. Although it will require con-
have been held: the 5th International Conference on tinuous efforts for practical application of CFD tools to the
Numerical Ship Hydrodynamics in Hiroshima, Japan (Sep. realistic propulsor geometry and wake, the recent progress
1989), the 19th International Towing Tank Conference in in this area provides the propulsor community with a
Madrid, Spain (Sep. 1990), the 18th Naval Hydrodynamics renewed optimism.
Symposium in Ann Arbor, USA (1990), the SNAME
Propellers/Shafting '91 Symposium in Virginia Beach, USA In the area of propeller design, a versatile preliminary
(Sep. 1991), and the International Symposium on Propulsors design tool has been developed at MIT [9] based on a lift-
and Cavitation in Hamburg, Germany (June 1992). ing-line vortex lattice method. This code can handle a vari-
ety of propeller classes, including the open single, ducted
This report is intended to summarize the advances and multi-component propulsors, such as the contrarotating
during the three years since the 22nd ATTC conference in propeller and propulsors with stator vanes ahead or behind
1989 in the areas of propulsor design and analysis, numeri- the rotor. Several compound propulsor designs and tests
cal prediction of propulsor flow and propulsor-hull interac- have been reported. DTMB designed and tested a contraro-
tion, and experimental meaurements of the propulsor flow tating [10] and a postswirl propulsor [11] for surface ships
and the performance. Five member organizations con- as a means of energy savings. The designs were developed
tributed to this report. by using the MIT lifting-line code [9] and extended lifting-
surface design codes. Open water and powering experi-
Substantial enhancement has been made to both poten- ments showed good agreements with design predictions.
tial and viscous flow calculations of propeller performance Partial-span stator vanes have been designed to improve the
and propeller-hull interaction. MIT has been continuously root cavitation erosion of a U.S. Coast Guard ship. Experi-
improving the panel method for steady and unsteady pro- ments showed substantial improvement in cavitation in the
peller analyses [1,2,3]*, including cavitation [4]. Although root area [12,13]. MIT developed an analyis method for a
the accuracy of global propeller performance calculations, ducted propeller with preswirl stator vanes [14]. Using this
such as for thrust and torque, has not been improved com- method a duct and a set of stator vanes were designed for an
pared to lifting-surface computations, the pressure distribu- existing rotor. Experimental results showed a very good
tion on the blade surface can now be computed much more agreement between the theoretical predictions and mea-
accurately with panel methods [5] except for the tip area. surements of thrust and torque on each component.
Accurate predictions of tip flow require more re,search. MIT
is currently investigating the tip flow using the panel code. Two papers were presented dealing with non-conven-
The Iowa Institute of Hydraulic Research (IIHR) has be,en tional propellers. A comprehensive literature survey and a
malcing a significant improvement to the prediction of preliminary design and evaluation method of cycloidal
propulsor flow and propulsor-body interactions using the propulsors were reported by Brockett [15]. A free-wake
viscous-flow method [6]. David Taylor Model Basin vortex-lattice lifting surface method was developed for ana-
(DTMB), formerly David Taylor Research Center, also lyzing cyclic-pitch propellers at MIT [16].
reported applications of Navier-Stokes solvers to propulsor-
hull-duct interaction problems. A Navier-Stokes simulation A new acoustic wind tunnel was built at the Atlantic
of the flow around a ducted propulsor, including the pro- Applied Research Corp. (AARC) with a major purpose of
peller-hull-duct interaction [7], showed favorable agreement measuring non-cavitating propulsor noise on submersibles.
with the experimental data. A Navier-Stokes calculation of Preliminary test results are presente,d in this report.
the flow around a ducted propulsor behind a full-stem body
The following contributions from the five member or-
*The references are listed at the end of the report. ganizations are presented in no particular order.

139
 DAVID TAYLOR MODEL BASIN (DTMB) bers. LDV measurements around the cavity and foil as well
as in the vicinity of the cavity surface have been performed
by Ki-Han Kixn and the results have been compared with the results from a
numerical non-linear method [17] in which the effects of the
Major accomplishments by DTMB during the past three tunnel walls were also included. Forces have been inferred
years are the design and experimental validation of several from momentum integration of the velocities and have also
compound propulsors and improvements in the 2-D foil sec- been compared to those from the numerical model.
tion design and in the analysis of the propulsor-hull-duct
interactions by using Navier-Stokes flow solvers. An Ducted Propellers
improvement was also made to the steady panel code devel- Several experiments on ducted propellers have been
oped by AMI for the prediction of propeller flow and the conducted, using a support mechanism which can isolate
blade pressure distribution [1]. propeller and duct forces. Description of these tests, includ-
ing correlation with theory, may be found in [19, 9].
A contrarotating [2] and a postswirl propellers [3] for
surface ship were designed as a means of energy savings
and tested in the towing tank. The MIT lifting-line code [4] THEORY
and extended lifting-surface design codes were used. Open
water and powering experiments showed good agreements Propeller Panel Methods: Steady, Cavitating and Unsteady
with design predictions. Partial-span stator vanes have been Steady panel methods have been developed, both for
designed to improve the root cavitation erosion of a U.S. open propellers [6] and for ducted propellers [14]. Progress
Coast Guard ship. Experiments showed substantial has been made in treating complex geometries, and further
improvement in cavitation in the root area [5,6]. work in this area is anticipated. A potential-based low-order
panel method has been developed for the analysis of sheet
A two-dimensional Navier-Stokes solver was used to cavitation in nonlinear theory. The method has be,en applied
design a thick foil section that has improved turbulent to 2-D and 3-D hydrofoils [17, 16, 4] in steady flow and to
boundary characteristics [7,8]. Comparison of the predic- propellers in unsteady flow [15]. An unsteady propeller
tions with the experimental data showed good agreement panel code has been developed as an extension of the previ-
and the desired flow characteristics of the thick section. A ous work in steady flow [5, 18]. Unsteady forces on ducted
Navier-Stokes simulation of the flow around a ducted propellers have been computed using a hybrid method
propulsor on an axisymmetric body with a reasonable full- comprising a panel representation of the duct and a lifting
ness, including the propeller-hull-duct interaction, showed surface representation of the blades. This research is
favorable agreement with the experimental data [9]. A pro- reported in [12].
cedure was developed for the design of a ducted propulsor
behind a very full stem by using a Navier-Stokes solver Lifting SwfacelThickness Interaction
[10]. It was suggested that the stem and the propulsor be An existing lifting-surface vortex-lattice analysis
designed as an integral system by properly accounting for method for propellers in steady flow has been extended to
the interaction between the hull-propulsor-duct. include the coupling between loading and blade thickness
[13].

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Cyclic Pitch Propulsors

A free-wake vortex-lattice lifting surface method for
by D.P. Keenan, J.E. Kerwin, and S.A. Kinnas analyzing cyclic-pitch propellers has been developed. A
brief description of the method and associated visualization
EXPERIMENTS techniques has been given in [11].

Unsteady Foil Experiment Complex Propulsor Lifting Line Theory

In this experiment, a two-dimensional "propeller-like" A vortex-lattice lifting line theory has been developed
foil section is subjected to a gust disturbance created by an which treats open, ducted, single stage and multistage
upstream pair of oscillating foils. Detailed LDV flow maps propulsors on a consistent basis [1]. An example of a recent
as well as surface pressure measurements are being made to parametric study using this procedure may be found in [8].
provide data for correlation with theory. Reduced frequen-
cies up to 7 have been achieved. Work to date is reported in PropellerlBoundary Layer Interaction
[21, 2].
An interactive boundary layer calculation methods
developed by Drela [3] has been combined with out poten-
2-D Cavity Flows tial based panel code by Hufford [7]. Comparison with
A supercavitating hydrofoil experiment has be,en per- experiments performed by Jessup [10] is encouraging.
formed at the MIT water tunnel [20]. The hydrofoil has
been tested at several angles of attack and cavitation num-

140
 IOWA INSTITUTE OF HYDRAULIC RESEARCH approach has been validated such that it should be used for
(IIHR), THE LTNIVERSITY OF IOWA design trade-off studies and relative design evaluation.

by F. Stern The work on the complete viscous-solution method

[5,6] has primarily concerned extensions for practical pro-
PROPULSOR FLOW AND PROPULSOR-BODY peller geometries, i.e., the SR-7 turboprop. Various techni-
INTERACTION cal issues have been investigated, including the following:
grid-generation techniques for complex propeller geometries
The work at IIHR on propulsor flow and propulsor- with regard to both grid quality and flow-field resolution;
body interactions involves the development of both interac- development of numerical techniques to enhance conver-
tive and "complete" viscous-flow computational methods. In gence and accuracy; and the explication of the blade-flow
the former case, the propeller is represented in the viscous- physics, i.e., leading-edge and tip and root vortices, span-
flow method by a body-force distribution which is obtained wise flow and trailing-edge separations, blade boundary
interactively from an inviscid propeller-performance method layers and wakes, blade-to-blade flow, and overall propeller
with specified effective inflow. In the latter case, the actual walce. The principal conclusions are as follows: the design
propeller geometry is embedded in the viscous-flow method condition performance and flow field are accurately pre-
and the detaileci flow about the rotating blades is simulated. dicted, however, only very limited experimental information
The focus of the current research is on the complete viscous- is available for detailed validation; off-design condition
flow method. isidelitionally, experimental studies have been performance and fkrw-field trends are aLso a.ccurately pre-
conducted both as an aid in explicating the flow physics and dicted, but further work is required for quantitative predic-
to provide test cases for validating the computational tions. Presently, calculations are in progress for the marine-
methods. propeller geometry, DTRC Model Propeller P4119, for
which a more complete set of experimental information is
The work on the interactive method has primarily con- available [7] for validation. The preliminary results are very
cerned validation studies for practical axisymmetric [1] and encouraging.
three-dimensional [2] propeller-body configurations, includ-
ing comparisons with the experimental studies [3,4] and Some additional on-going studies include: the devel-
inviscid-flow methods. Various technical issues have been opment of zonal methods in conjunction with advanced tur-
investigated, including the following: relative influence of bulence modeling for the detailed resolution of the blade
turbulence modeling and grid density on the resolution of flow; and extensions for unsteady-flow simulations, includ-
the propeller inflow; development of propeller-hull interac- ing the use of overlaid grids and studies concerning the
tion procedures for axisymmetric and three- dimensional interaction of natural and forced unsteady flows.
geometries, i.e., utilization of radially and angularly varying
body-force distributions and appropriate definitions of the
effective propeller inflow; and the explication of the physics ATLANTIC APPLIED RESEARCH CORP. (AARC)
of propeller-hull interaction, including the differences of the Burlington, Massachusetts
flow for the with and without-propeller conditions and the
differences of the steady and unsteady propeller perfor- by D.S. Greeley and H.A. Brown
mance for the nominal and effective propeller inflows. The
principal conclusions are as follows: modified turbulence NON-CAVITATING PROPELLER NOISE
models and grid refinement enable accurate prediction of the
Oth harmonic and some improvement for the higher harmon- For undervvater vehicles propeller cavitation can usually
ics of the nominal inflow, however, the current limitations be eliminated, so that the remaining propeller noise sources
of computational fluid mechanics preclude a more detailed of concern are related to unsteady fluid dynarnic pressures
description of the flow; the differences of the higher har- acting on the blades. These unsteady pressures arise from
monics of the nominal and effective propeller inflows and of the interaction of the propeller blades with inflow velocity
the higher harmonics of t.he resulting unsteady propeller spatial non-uniformity and turbulence, and the interaction
forces and moments are of the same order as the differences between the blades and the turbulent boundary layers that
of the higher harmonics between the calculations and the form on them. Since these phenomena involve only single
experimental data; the steady flow field and propeller per- phase fluid flow, they can be studied in any fluid media.
formance are accurately predicted; the propeller-hull interac- We have chosen to study these phenomena in air for both
tion is explicable in terms of the consequences of the pro- economic and acoustic reasons.
peller loading resulting from the propeller operation in the
effective propeller inflow; significant differences are The AARC Acoustic Wind Tunnel (AWT) was
observed between the present viscous-flow predictions and designed and built in 1989 specifically for the measurement
those from the inviscid-flow methods, which are indicative and study of noncavitating propulsor noise on submersibles.
of the limitations of the latter; and, finally, the interactive Figure 1 shows the facility, which is an open jet wind tunnel
with the jet exhausting into an acoustic measurement cham-

141
 123 f L.

Muffler Bank

Return Air Passage

Screens
Acoustic Test Chanaber
Control Room (Anechoic)
ting

Model Propeller(s)
o

diame ter
Storage Room 8,8' square Model of Submersible
mull

Return Air Passage

Labyrinth Silencer Expansion Joint Honeycomb and Screens

PLAN

11111111111111111 11 1111 1 1 11111 111 1111111111111111 111

6 ft.-

Sting AATAAAAJALIA
Foundation Pit with Anechoic Treatment
ELEVATION

Figure 1. AARC Acoustic Wind Tunnel

ber. A variable speed 200-hp electric motor drives an low, to enable measurements to be made of very quiet
upstream centrifugal blower to move the air through the propulsors. Vehicle models are cantilevered in the open jet
tunnel. Extensive noise control treatments have been built by a nose sting, so that no model suspension struts or guy
into the tunnel so that the facility background noise is very wires are needed to support the model in the flow. With a
typical model in the 4 ft. diameter exit duct, the flow veloci-
AXIAL vimocraes - NO PROPELLER
ties in the annulus around the model reach 120 ft/sec. The
model propulsors (either single or contrarotating) are driven
by variable speed electric motors mounted within the model
afterbody, and can spin model propulsors up to 6000 RPM.

j#11 ritWitiW
0.1
The model propulsor drives have also been extensively qui-
eted to avoid interfering with propulsor noise measurements.
as

04 /111111FAINII
v y
The AWT facility is primarily used for measuring fluid
dynamic and acoustic parameters of interest. Of particular
interest are walce surveys done to measure propulsor inflow,
which are made with 5-hole Pitot tubes (for the 3 compo-
02 nents of the time average flow) and hot wire anemometry
for measuring turbulence characteristics of the propulsor
0 .0
010.0.00100101111110110,11001101.001010000,000001010,111...11000010101010/11110001,1111101000001.1.0.0.001M,00000
SOO
inflow. These measurements are made using an automated
Lso ura goo
IO lO
POSITION (DEGREES) walce survey gear, driven directly by the data acquisition
3.5" 4.5 6.5" ___. 6.5" 7.5" 8.6" computer. Figure 2 shows an example of the axial velocities
measured behind a squared-off cross section submersible:
Figure 2. Axial Velocity Behind Submersible

142
 note the lack of wake features caused by model support currently performed at DREA using lifting surface methods.
struts. Boundary layer measurements and flow visualization Poor agreement with experiments has been found when the
studies are also routinely done in this facility. Acoustic cavitation approaches the root of the blade. This is likely
measurements of propeller noise are made using both fixed due to neglect of the propeller hub and the inability of the
and movable microphones located outside the open jet shear singularity distributions at the mean blade surface to model
layer, over the frequency range 100 Hz - 40,000 Hz. the thick sections near the root. To address both these prob-
lems, DREA is developing (under contract to Aerospace
Over the last three years, this facility has seen extensive Engineering Consultants Ltd. (AERCOL), Toronto and
use in performing baseline studies of non-cavitating noise MIT) a surface panel singularity code which will also
for various propulsors, as well as supporting propulsor include hub and root fillet geometries. The steady inflow
design efforts in the areas of afterbody flow visualization, version of this code is complete [4,5]. It is a modification of
propulsor wake surveys, and verification of the acoustic the MIT code PSF-10 and is called ASP-1. The unsteady
performance of propulsors designed for particular vehicles. version is still under development

In order to support propulsor design efforts, extensive Research on a reliable method for predicting TVC is
development of analytical and numerical techniques for the being performed under contract by George Engineering Ltd.
prediction and reduction of non-cavitating noise has also [6,7]. To date, only planar lifting surfaces have been
taken place in conjunction with wind tunnel measurement addressed. Although vortex locations have reasonable cor-
programs. We can now routinely predict, using physics- relation with model test data obtained at MARIN, the vortex
based models (within +4 dB) the broadband non-cavitating velocities and pressures cannot be predicted reliably.
noise that will be measured from a given propulsor when Improvements to the method are continuing.
tested, given a measurement of the inflow turbulence
characteristics. WAKE MODELING

The DREA wake modeling programs, HLLFLO

DEFENSE RESEARCH ESTABLISHMENT [8,9,10], have been enhanced to include the effects of a free
ATLANTIC (DREA) surface using a Kelvin/Havelock source method. Validation
Dartmouth, N.S., Canada and documentation of the free surface code is in progress, as
well as studies of the effect of the free surface on propeller
by David HaIly inflow. Work has begun on the development of flow codes
which solve the Reynolds-Averaged Navier-Stokes (RANS)
INTRODUCTION equations with an appropriate turbulence model. A litera-
Propeller research at the DREA is currently focused on ture survey of methods currently in use was completed in
the reduction of propeller noise. Research is being con- 1991 [11]. A program to generate grids for ship geometries
ducted in three main areas: fundamental studies of the way has recently been completed [12]. DREA has also obtained
in which noise is produced by cavitation; prediction of cavi- the wake scaling code WAICE86 from BMT and modified it
tation inception, extent, and noise production; and walce for use with our vessel wake database. An experimental
modeling. This brief note describes propeller research program to conduct model submarine wake surveys, propul-
undertaken at DREA during the past three years. Nethercote sion tests, and cavitation tests will begin later this year.
et al. [1] have provided a more detailed survey of DREA
propeller research over the past decade. INTERNATIONAL COOPERATION

CAVITATION NOISE PRODUCTION MECHANISMS During the past three years, DREA has been involve,d in
three international cooperative propeller research ventures.
Cavitation noise experiments have been performed at
the DREA Acoustics Barge [2,3]. Analysis of these data is The NSMB Cooperative Research Ships Working
still in progress. The data expand upon previous observa- Group on Propeller Tolerances: This group completed
tions that tip vortex cavitation (TVC) can be an intense its project in 1991 [13,14,15].
source of relatively narrow bandwidth, low frequency noise. The NATO Research Study Group on Cavitation
Numerical studies of the dynamics and noise production of Noise Scaling: The final report is in press.
single cavitation bubbles have been completed and docu- The cooperative Naval Research Project on In-
mentation is in progress. Studies of clusters of air and cavi- Service Propeller Performance: A DREA inhouse
tation bubbles have begun. study [16] has determined that cavitation noise is
important at all sea-states. The aim of the interna-
PREDICTION OF CAVITATION tional study is to determine the effects of motion in a
seaway on propeller performance. A literature survey
Prediction of the extent of unsteady back sheet cavita- and supporting numerical calculations have been per-
tion, surface pressures, and cavitation inception speeds is formed [17]. Experiments on DREA's research ves-

143
 sel Quest have been performed in July 1991 and April Coney, W.B., A Method for the Design of a Class of
1992; flow measurements near the propeller were Optimum Marine Propulsors, Ph.D. Thesis, MIT De-
made with an LDV system developed at MARIN, in partment of Ocean Engineering, August 1989.
conjunction with propeller viewing provided by the
DREA video system [18]. Analysis of the data is Chen, B.Y.-H. and A.M. Reed, "A Design Method and
underway. an Application for Contrarotating Propellers," DTRC-
90/003 (Jan 1990).
OTHER WORK Chen, B.Y.-H., "Postswirl Propulsors - A Design
Method and an Application," Proceedings, Interna-
A study of the effects of synchrophasing twin propellers tional Symposium on Propulsors and Cavitation Con-
on the levels of radiated noise and on propeller-induced hull ference, Hamburg, Germany (June 1992).
pressures has been completed. Documentation is in
progress. Neely, S.K. and B.Y.-H. Chen, "Asymmetric Preswirl
Stator Design for the U. S. Coast Guard Island Class
Patrol Boat," DTRC/SHD-1335-02, (Mar 1991).
REFERENCES Neely, S.K., J. McMahon, and B.Y.-H. Chen, "Design
Method and Application of an Asymmetric Stator Up-
Introduction stream of an Inclined Shaft Propeller," Proceedings,
23rd American Towing Tank Conference, New Or-
Hsin, C.-Y., Development and Analysis of Panel leans, LA (June 1992).
Methods for Propellers in Unsteady Flow, Ph.D. The-
sis, MIT Department of Ocean Engineering, January Hughes, M.J. and S.A. Kinnas, "Analysis Method for
1991. a Ducted Propeller with Preswirl Stator Blades," Pro-
Hsin, C.-Y., J.E. Kerwin, and S.A. Kinnas, "A Panel ceedings, Propellers/Shafting '91 Symposium,
SNAME, Virginia Beach, VA, September 1991.
Method for the Analysis of the Flow Around Highly
Skewed Propellers," Proceedings, Propellers/Shafting Brockett, T., "Hydrodynamic Analysis of Cycloidal
'91 Symposium, SNAME, Virginia Beach, VA, Propulsors," Proceedings, Propellers/Shafting '91
September 1991. Symposium, SNAME, Virginia Beach, VA, September
1991.
ICinnas, S.A., M.J. Hughes, and I.E. Kerwin, "Com-
parison of Analysis and Experiment for a Ducted Pro- Keenan, D.F., "Computer Animation of a Cyclic Pitch
peller," Proceedings, 2nd International Symposium on Propeller," Marine Computers '91, The 2nd
Performance Enhancement for Marine Applications, Symposium on Computer Applications in the Marine
Newport, Rhode Island, October 1990. Industry, Burlington, Mass. September 1991.
Kinnas, S.A. and N.E. Fine, "A Nonlinear Boundary
Element Method for the Analysis of Unsteady Pro-
peller Sheet Cavitation," To be presented at the 19th David Taylor Mode Basin
Symposium on Naval Hydrodynamics, August 1992, Yang, C.-I., "Prediction of Propeller Blade Pressure
Seoul, Korea. Distribution with a Panel Method," DTRC-90/013
Yang, C.-I., "Prediction of Propeller Blade Pressure (May 1990).
Distribution with a Panel Method," DTRC-90/013 Chen, B.Y.-H. and A.M. Reed, "A Design Method and
(May 1990). an Application for Contrarotating Propellers," DTRC-
Stern, F., Y. Toda, and H.T. Kim, "Computation of 90/003 (Jan 1990).
Viscous Flow Around Propeller-Body Configurations: Chen, B.Y.-H., "Postswirl Propulsors - A Design
Iowa Axisymmetric Body," J. Ship Research, Vol. 35, Method and an Application," Proceedings, Interna-
No. 2, June 1991, pp. 151-161. tional Symposium on Propulsors and Cavitation Con-
Yang, C.-I., P.M. Hartwich, and P. Sundaram, "A ference, Hamburg, Germany (June 1992).
Navier-Stokes Solution of Hull-Ring Wing-Thruster Coney, W.B., A Method for the Design of a Class of
Interaction," Proceedings, 18th Symposium on Naval Optimum Marine Propulsors, Ph.D. Thesis, MIT De-
Hydrodynamics, Ann Arbor (1991). partment of Ocean Engine,ering, August 1989.
Dai, C.M.H., "Computation of an Integrated Ducted Neely, S.K. and B.Y.-H. Chen, "Asymmetric Preswirl
Propulsor/Stern Performance in Axisymmetric Flow," Stator Design for the U. S. Coast Guard Island Class
Proceedings, 23rd American Towing Tank Confer- Patrol Boat," DTRC/SHD-1335-02, (Mar 1991).
ence, New Orleans, LA (June 1992).

144
 Neely, S.K., J. McMahon, and B.Y.-H. Chen, "Design Hughes, Mi. and S.A. Kinnas, "Analysis Method for
Method and Application of an Asymmetric Stator Up- a Ducted Propeller with Preswirl Stator Blades," Pro-
stream of an Inclined Shaft Propeller," Proceedings, ceedings, Propellers/Shafting '91 Symposium,
23rd American Towing Tank Conference, New Or- SNAME, Virginia Beach, VA, September 1991.
le,ans, LA (June 1992).
Jessup, S.D., An Experimental Investigation of Vis-
Nguyen, P.N. and Gorski, Ji., "Navier-Stokes Analy- cous Aspects of Propeller Blade Flow, Ph.D. Thesis,
sis of Turbulent Boundary-Layer and Wakes for 2-D The Catholic University of America, 1989.
Lifting Bodies," Proceedings, 18th Symposium on
Naval Hydrodynamics, Ann Arbor (Aug 1991). Keenan, D.F., "Computer Animation of a Cyclic Pitch
Propeller," Marine Computers '91, The 2nd
Nguyen, P.N., "Use of Navier-Stokes Analysis in Symposium on Computer Applications in the Marine
Section Design," DTRC/SHD-1262-04 (Dec 1990). Industry, Burlington, Mass. September 1991.
Yang, C.-I., P.M. Hartwich, and P. Sundararn, "A Kinnas, S.A., C.-Y. Hsin, and D. P. Keenan, "A Po-
Navier-Stokes Solution of Hull-Ring Wing-Thruster tential-Based Panel Method for the Unsteady Flow
Interaction," Proceedings, 18th Symposium on Naval Around Open and Ducted Propellers," Proceedings,
Hydrodynamics, Ann Arbor (1991). 18th Symposium on Naval Hydrodynamics, Ann Ar-
Dai, C.M.H., "Computation of an Integrated Ducted bor, Michigan (1991).
Propu/sor/S terrr Performance irr Axisymmetric Flow," Kinnas, S.A., "A General Theory for the Coupling
Proceedings, 23rd American Towing Tank Confer- Between Thickness and Loading for Wings and Pro-
ence, New Orleans, LA (June 1992). pellers," Journal of Ship Research, 36(1): pp. 59-68,
March 1992.

Massachusetts Institute of Technology Kinnas, S.A. and W.B. Coney, "The Generalized Im-
age Model - an Application to the Design of Ducted
Coney, W.B., A Method for the Design of a Class of Propellers," Journal of Ship Research, 36(3),
Optimum Marine Propulsors, Ph.D. Thesis, MIT De- September 1992. To appear.
partment of Ocean Engineering, August 1989.
Kinnas, S.A. and N.E. Fine, "A Nonlinear Boundary
Delpero, P.M., Investigation of Flows Around a Two Element Method for the Analysis of Unsteady Pro-
Dimensional Hydrofoil Subject to a High Reduced peller Sheet Cavitation," To be presented at the 19th
Frequency Gust Loading, Master's Thesis, MIT De- Symposium on Naval Hydrodynamics, August 1992,
partment of Ocean Engineering, February 1992. Seoul, Korea.
Drela, M., "XFOlL: An Analysis and Design System Kinnas, S.A. and N.E. Fine, "A Numerical Nonlinear
for Low Reynolds Number Airfoils," Lecture Notes in Analysis of the Flow Around 2-D and 3-D Partially
Engineering (Vol. 54, Low Reynolds Number Cavitating Hydrofoils," Submitted for Publication.
Aerodynamics), Springier-Verlag, New York, 1989.
Kinnas, S.A. and N.E. Fine, "Non-Linear Analysis of
Fine, N.E. and S.A. Kinnas, "A Boundary Element the Flow Around Partially or Super-Cavitating Hydro-
Method for the Analysis of the Flow Around 3-D foils by a Potential Based Panel Method," Boundary
Cavitating Hydrofoils," (submitted for publication). Integral Methods - Theory and Applications, Proceed-
Hsin, C.-Y., Development and Analysis of Panel ings, IABEM- 90 Symposium, Rome, Italy, October
15-19, 1990, pp. 989-300 Springer-Verlag, Heidelberg,
Methods for Propellers in Unsteady Flow, Ph.D. The- 1991.
sis, MIT Department of Ocean Engineering, January
1991. Kinnas, S.A. and C.-Y. Hsin, "A Boundary Element
Hsin, C.-Y., J.E. Kerwin, and S.A. Kinnas, "A Panel Method for the Analysis of the Unsteady Flow Around
Method for the Analysis of the Flow Around Highly Extreme Propeller Geometries," AIAA Journal,
30(3):688-696, March 1992.
Skewed Propellers," Proceedings, Propellers/Shafting
'91 Symposium, SNAME, Virginia Beach, VA, Kinnas, S.A., Mi. Hughes, and I.E. Kerwin, "Com-
September 1991. parison of Analysis and Experiment for a Ducted Pro-
Hufford, G., Viscous Flow Around Marine Propellers peller," Proceedings, 2nd International Symposium on
Using Boundary Layer Strip Theory, Master's Thesis, Performance Enhancement for Marine Applications,
MIT Department of Ocean Engineering, May 1992. Newport, Rhode Island, October 1990.

Hugel, M.A, An Evaluation of Propulsors for Several Kinnas, S.A. and C.H. Mazel, "Numerical vs Experi-
Navy Ships, Master's Thesis, MIT Department of mental Cavitation Tunnel," Proceedings, 23rd ATTC
Ocean Engineering, June 1992. Conference, New Orleans, June 1992.

145
[21] Rice, J.Q., Investigation of a Two Dimensional Hydro- Johnson, G.W., S. Hulshof, A. Samji, and R. Bowser,
foil in Steady and Unsteady Flows, Master's Thesis, "A Three-Dimensional Combined-Singularity Potential
MIT Department of Ocean Engineering, June 1991. Code for Marine Propellers, ASP Version 1.0 User's
Guide," DREA Contractor Report CR1911449/VOL. II,
July 1991.
Iowa Institute of Hydraulic Research (IIHR) George, M.F., "Analytical and Experimental Investiga-
Stern, F., Y. Toda, and H.T. Kim, "Computation of tion of Tip Vortices Associated with Planar Lifting
Viscous Flow Around Propeller-Body Configurations: Surfaces," DREA Contractor Report. CR/89/430,
Iowa Axisymmetric Body," J. Ship Research, Vol. 35, 1989.
No. 2, June 1991, pp. 151-161. George, M.F., "Development and Validation of Nu-
Stern, F., H.T. ICim, D.H. Zhang, Y. Toda, J.E. Ker- merical Methods on Tip Vordces, Phase 1: Planar Lift-
win, and S. Jessup, "Computation of Viscous Flow ing Surfaces," DREA Contractor Report CR/91/450,
Around Propeller-Body Configurations: Series 60 CB June 1991.
= .6 Ship Model," to be submitted J. Ship Research. Hally, D., "POTFLO: A Suite of Programs for Calcu-
Toda, Y., F. Stern, I. Tanaka, and V.C. Patel, "Mean- lating Potential Flow about Ship Hulls," DREA Tech-
Flow Measurements in the Boundary Layer and Wake nical Memorandum 891210, 1989.
of a Series 60 CO = .6 Model Ship With and Without. Haily, D., An Integral Method for the Calculation of
Propeller," J. Ship Research, Vol. 34, No. 4, Dec. Boundary Layer Growth on a Ship Hull," DREA Re-
1990, pp. 225-252. port 85/107, 1985.
Hyun, B.S., Measurements of the Flow Around a Ma- Hally, D., "HLLFLO User's Guide," DREA Technical
rine Propeller at the Stern of an Axisymmetric Body, Communication 86/306, 1986.
PhD. Thesis, The University of Iowa, Iowa City, Iowa
Ethier, C.R., "Navier-Stokes Solvers for Ship Flows:
(1989).
A Literature Survey," DREA Contractor Report
Kim, H.T. and F. Stern, "Viscous Flow Around a Pro- CR1911409, 1991.
peller-Shaft Configuration With Infinite-Pitch Rectan-
gular Blades," AIAA J. Propulsion and Power, Vol. 6, Hall, Di., D.W. Zingg, and C.R. Ethier, "An Elliptic
No. 4, July-August 1990, pp. 434-444. Grid Generator for Ship Hulls," DREA Contractor Re-
port (in press).
Kim, H.T. and F. Stern, "Computation of Viscous
Flow Around Propeller-Shaft Configurations: SR-7 Luttmer, R.I., "Measurements of Performance Varia-
Turboprop," IIHR Report (in preparation). tions on a Series of Propellers and Blades Differing by
Tolerance Errors," Report 48236-5-SP, MARN, June
Jessup, S.D., An Experimental Investigation of Vis- 1991.
cous Aspects of Propeller Blade Flow, Ph.D. Thesis,
The Catholic University of America, 1989. Chevalier, Y.D. and P.F. Genoux, "Effect of Tolerance
Errors on the Performance of a Marine Propeller," Re-
port 1.03.33.2042, Bassin d'Essais des Carenes, Apr.
Defense Research Establishment Atlantic (DREA). Canada 1991.
Carlton, J.S., "MARIN Cooperative Research Pro-
Net.hercote, W.C.E., D. Hally, M. Mackay, Di. Noble,
peller Tolerances Working Group Summary Report,"
and N.C. Sponagle, "DREA's Propeller Design and Report PPEE.532.90, Lloyd's Register Engineering
Analysis Experience," in MARIN Jubilee Workshops
Services, May 1991.
(May 1992).
Nethercote, W.C.E. and N.C. Sponagle, "The Relative
Sponagle, N.C., "The Sound from Different Types of Importance of Cavitation and Ambient Noise," DREA
Cavitation," Proceedings of the Canadian Acoustical
Note W91/4, 1991. (Informal Report).
Society (1989).
Kuiper, G., "Velocities in the Propeller Plane due to
Sponagle, N.C., "Noise from Tip Vortex and Bubble
Waves and Manoeuvers," Report 48885-1-SS,
Cavitation," DREA Technical Memorandum 90/202,
MARIN, May 1991.
March 1990.
Kennedy, J.L., N.C. Sponagle, D.W. Wheaton,
Johnson, G.W., S. Hulshof, A. Samji, and R. Bowser, P.MacDonald, and R. W. Creaser, "Video Systems for
"A Three-Dimensional Combined-Singularity Potential Propeller Viewing," Proceedings, 22nd American
Code for Marine Propellers, Phase 1 (Steady Flow)," Towing Tank Conference (1989).
DREA Contractor Report CR1911449/VOL. I, July
1991.

146
 Development of a Waterjet Propulsor for the
Marine Corps High Water Speed
Landing Craft Application
J. Stricker, A.J. Becnel, J.G. Purnell
(Carderock Division, Naval Surface Warfare Center, USA)
ABSTRACT the PSD defined the speed-thrust
An requirements for each of these vehicles
advanced high performance and provided
waterjet propulsor was developed as part preliminary powering
of the Marine Corps estimates for a full-scale AAAV planing
Surface Mobility hull concept. Waterjet propulsor design
technology development effort. Design goals were
optimization, performance estimating, and established during this
hydrodynamic development process, and propulsor design
design -procedures are
described for a propulsion unit required parametric analyses performed to define
to meet specified thrust, power, and minimum-envelope waterjet designs capable
of meeting the established thrust-speed-
envelope characteristics. Testing of a
10-inch scale model and trials of the full powering requirements. These requirements
led toward the application of cavitating
scale propulsion units in both the HWSTD
(High Water Speed Technology Demonstrator) inducer technology at much higher flow
and PSD (Propulsion System Demonstrator) coefficients than current state-of-the-art
developmental vehicles are described and practice to provide a minimum volume and
performance data correlations presented. weight propulsion pump design.

NOMENCLATURE Before proceeding to the detailed

pump and nozzle design stage, definition
C. Ratio of NPSH to NPSH at Complete of the inlet loss and drag characteristics
Head Breakdown was made based on a design series created
by Mr. John Hoyt of CARDEROCKDIV, NSWC,
Gravitational Constant, ft/sec2 Code 1562 (formerly the David Taylor
Incidence angle, degrees Research Center). A baseline inlet design
JVR Jet Velocity Ratio Vi / V.
( ) by Jack Roper(5) was included in model
Cavitation Number (2 g P V2rei) testing of the series at the Davidson
NB Number of Blades Laboratory(6). A semi-flush inlet design
NPSH Net Positive Suction Head, ft using an elliptical lip contour was
Pst Static Pressure, ft selected for the full-scale HWSTD
Utip Blade Tangential Velocity, ft/sec application, based on measured drags and
Axial Velocity , ft/sec internal losses for minimum-envelope
Vi Jet Velocity, ft/sec configurations. This design was used in
Craft Velocity, ft/sec both towing-tank propulsor model tests and
Vrei in the full-scale propulsors as installed
Relative Velocity, ft/sec
in the HWSTD vehicle. For further savings
flin Blade Metal Angle, degrees in propulsor length for the PSD
Inducer Cavitation Coefficient installation, a shortened version of this
0 Flow Coefficient inlet was designed and constructed.
INTRODUCTION Figure 1 shows the propulsor cross-section
for the PSD system.
Several developmental vehicles have
been designed to establish a data base and Optimum final design parameters were
developed using the inlet and bare hull
demonstrate critical system capabilities model data described above and detail
for the Marine Corps Advanced Assault designs of the rotors, stator, and
Amphibious Vehicle (AAAV) program. nozzle
Propulsor studies and model tests have were performed using an assemblage of
defined the retractable planing surface design methods covering both cavitating
and non-cavitating
mounted waterjet design blade section
concept and hydrodynamics. Initial blade root
established the feasibility of this system stress
estimates were made and materials selected
for heavily-loaded planing hulls(1, 2, 3, 4) for
Scale model tests of both the HWSTD and
.
10-inch diameter scale model
components which were manufactured at the
Naval Surface Warfare
Center. Model
147
components were assembled and propulsor calculates the cavity length and height
performance tests completed in the NSWC based on a cascade of flat plates at an
towing tank facility. These tests angle of attack. The incidence angle
verified the predicted powering, thrust, causes a vapor cavity to form on the
and cavitation performance characteristics suction surface of the blade that will
of the inlet and propulsion pump designs. eventually collapse in the blade channel
if properly designed. The cavity
A total of seven full-scale, 16.1 dimensions are a function of the blade
inch diameter, propulsors were angle, stagger angle, relative velocities,
manufactured by Northern Research and incidence angle, and cavitation number.
Engineering Corp. (NREC) in accordance Figure 4 shows the cavity shape for the
with designs developed by NSWC. Three inducer tip section. Blade thickness and
propulsors were used to propel the HWSTD, chord lengths were determined on the basis
using a diesel-driven hydrostatic drive of the calculated cavity dimensions for
system. The PSD vehicle uses four the cavitating section of the blade.
propulsors powered by a combined diesel
and gas turbine-driven hydrostatic drive. Noncavitating Blade Row Design
Testing of the HWSTD was completed at the The noncavitating blade sections
NSWC Special Trials Unit (STU) located on were designed with the mean-streamline
the Patuxent River in Miryland. Tests of method. The method allows the designer to
the PSD have been completed at STU and are quickly layout blade sections with
continuing at" the Marine Corps Camp arbitrary thickness and chordwise loading
Pendleton California base. distributions that have a specified amount
of fluid turning. Also, the method can
PUMP DESIGN METHODS AND THEORY account for three dimensional flow
effects, such as changes in axial velocity
1-D Design Optimization along the blade chord.
A 1-D design optimization computer
program was developed, based on a The method divides the blade section
Rocketdyne design method(7), to determine into equal segments with the mean
the size, cavitation, and performance streamline laid out along the relative
characteristics of axial-flow waterjet velocity vector for each constant radius
systems with free-vortex blading. The segment. The mean camberline is obtained
method sizes the waterjet for minimum by applying deviation to the mean
diameter and length using cavitating streamline because the flow exits from the
inducer design theory. The inducer blade at a smaller angle than the metal
suction performance limit curve is shown angle. Deviation is calculated by an
in Figure 2 for a 3 degree tip incidence empirical formula which relates it to the
angle. The inducer cavitation coefficient section lift coefficient and solidity(10).
(7) versus flow coefficient (0) is a non- The blade surface coordinates are obtained
dimensional plot of the suction by applying the thickness normal to the
performance. A 20 percent cavitation mean camberline. Leading and trailing
margin was used for the current design. edges are blended into the existing blade
shapes using double circle arcs.
The required thrust at 18 miles per
hour was 3767 pounds for each waterjet. The blade surface pressure
Figure 3 shows all possible design points distributions are estimated by using the
for a incidence angle of 3 degrees and a impulse-momentum relationship which
JVR of 1.6 to 2.6. The stator cavitation relates the average pressure difference
margin was also held to a minimum of 20 across the blade segment to the fluid
percent, which occurs at the hub section turning. The suction surface pressure is
for a free vortex design. The design point estimated by subtracting two thirds of the
for the waterjet was chosen at a JVR of pressure difference from the mean static
2.4 and flow coefficient of .44, since it pressure at that blade segment. The NACA
best met the required size and powering diffusion factor was held to .5 or less
specifications. Design point parameters for each section to avoid flow separation.
are shown in Table 1.
A 2-D Navier-Stokes flow solver(11)
Cavitating Blade Row Design was used to check the final design for
The cavitating blade sections were each section. The code was used to
designed using a method based on Stripling compare flow turning and surface pressure
and Acosta free-streamline cavity model(8) distributions. The mean streamline and
in conjunction with the mean-streamline Navier-Stokes codes agreed to within less
method(9). The free-streamline model than one degree on flow turning. The

148
 pressure differences across the blade The parent thickness distribution used was
generally were in good agreement; however, the same as that in Ref. 9 with the
as expected, the mean streamline method maximum thickness at 60 percent chord.
always predicted a lower minimum pressure. The number of blades were double that of
the inducer to minimize the axial length.
PUMP DETAIL DESIGN The blade thickness was also held to a
minimum in order to control blockage in
Once the design-point the blade channel. The total axial length
characteristics of Table 1 were of the kicker is 5.5 inches at the hub
established, an interactive procedure was compared to 8.45 inches for the inducer.
initiated in which materials strength,
overall pump and nozzle length, and
producability issues were considered Stator and Nozzle Design
simultaneously. For inviscid hydrodynamic Design of minimum length stator
similarity, blade dimensions are inversely blades was found to be constrained by
related and stresses directly related to blade and boundary layer blockage effects
numbers of blades per row. Minimum length coupled with 5-axis machining limitations
blade rows are also influenced by boundary for thin section, closely spaced blades.
layer blockage effects which become more Ten blades were chosen for the final
severe as numbers of 'blades increase. design and sections were designed by the
These blockage effects will in the /imit mean streamline method to operate with a
req-uire relatively thinner blade sections, 20 percent cavitation margin at the design
so that blade stresses may increase point speed of 18 mph. Because of support
steeply with increasing numbers of blades. given by the stator housing,
Stress considerations directed this design stress
toward a four-bladed inducer followed by considerations were much less important
than for the inducer and kicker. A unique
a separate eight-bladed non-cavitating
booster impeller, feature of this propulsor is the fully-
or kicker. This ventilated stator hub and blade trailing
combination provided the minimum edge design, configured to further limit
achievable overall length rotors and met overall length.
requirements for 5-axis machining of each The full-scale nozzle
length is .80 inches, with a diametral
blade row.
contraction of 1.0 inches over that
length. The overall stator and nozzle
Inducer Design length was 5.5 inches.
The inducer is a low solidity 4 The flat blade
trailing edges were designed to provide
bladed impeller that raises the pressure ventilating paths to the flush hub surface
to a sufficient level so the noncavitating
kicker blade row can input the majority of
from the atmosphere surrounding the jet
the headrise. The tip incidence angle was while the vehicle is on plane. As seen in
chosen at 3 degrees with a cavitation Figure 1, this combined stator and nozzle
margin 20 percent above complete head design is very compact. Overall length
breakdown. The incidence angles at other reduction on the order of eight inches or
more was realized compared to a
radii were calculated to maintain the 20
-percent margin from hub to tip. conventional tapered hub, fully wetted
This design.
required that sections other than the tip
needed to also have blade camber, in Stresses and Materials Selection
addition to the incidence, to meet the Initial
design headrise. blade root stress
The noncavitating part calculations were used in the development
of the inducer sections were designed with of radial thickness distributions and
the mean streamline method. The thickness
selection of inducer, kicker, and stator
distribution of the cavitating part of the materials. A nickel-aluminum bronze
inducer blade was held to 90 percent of material (AMPCO 45) was selected for
the cavity height and the chord length for strength, machinability, and corrosion
the cavitating section was chosen to be 25
resistance in seawater. Ten inch-scale
percent longer than the cavity length. fully machined inducer and kicker blade
This provides a margin of safety for the rows were manufactured at NSWC,
cavitating design. and
initial tests showed excessive deflection
Kicker Design of inducer blades resulting in housing
interference. At that point, finite
The kicker is a non-cavitating high element models of all three blade rows
solidity 8 bladed impeller that inputs the
were defined and NASTRAN stress analyses
majority of the required headrise using run by Mr. Ray Mikesell at AA1 Corp. Data
only blade camber. The blade row was
designed using the mean streamline method. confirmed that stresses and deflections in
the inducer were high, requiring a design

149
and material change. Blade sections were established as the optimum design at 18
restacked and the material changed to 17-4 mph.
PH H1100 precipitation hardening stainless
steel. Tests of the model components were HWSTD and PSD Trials
completed with the revised inducer, and Torque-RPM curves for the full-scale
adequacy of the 17-4 material was pumps are shown in Figure 5, and are
demonstrated. For the full-scale essentially the same as the scaled 10-inch
components, precision castings of 15-5 PH measured curve. The shortened inlet and
were specified (inducer, kicker, and one- transition design used on PSD is the
piece stator blade row/housing). Castings probable cause for some bollard test
of this alloy, provided by NREC, are torque loss at over 1300 RPM as seen in
solution annealed at 1950 F for 4.5 hours Figure 5, where inlet energy losses and
and heat treated at 1150 F for 4 hours to velocity distortions have most likely
provide required strength and corrosion initiated inducer cavitation breakdown.
resistance. Underway measurements show generally that
torque loss does not occur in the vicinity
Design Integration and Construction of the 18 mph hump, but that air ingestion
Design of full scale housings and effects cause some seaway-dependent torque
driveline components, including selection and thrust loss at higher speeds in both
and mounting of the hubLmounted spherical vehicles. The measured maximum speeds
roller thrust bearing, was performed by recorded for the HWSTD and PSD vehicles
Mr. Steve Miller of AAI Corp. The pump are 29.4 and 28.5 knots, respectively.
inlet transition/impeller housing design These values match quite closely predicted
features graphite epoxy and glass fiber top speeds for each vehicle. Test
composite construction for strength, light experience to date has seen no component
weight, corrosion resistance, and failures, excessive wear, or apparent
dimensional stability. Inlets were corrosion-erosion pitting or metal loss in
fabricated as an integral part of the the 15-5 PH impellers and stators.
tailgate aluminum weldment. Assembly of
the propulsor system (hydraulic drive Conclusions
motors, shafts, housings, impellers, and Effective, compact waterjet systems
stator/nozzle units) was performed by AAI may be designed using carefully-applied
Corp., manufacturer of the HWSTD and PSD cavitating blade row theory, coupled with
vehicles. Figure 6 shows the stackup of a detailed development of non-cavitating
precision cast inducer, kicker, and blade sections in both impellers and
stator/housing components. stators. The most compact propulsor
designs require inducer flow coefficients
TEST EXPERIENCE AND CONCLUSIONS which are typically higher than present
practice dictates, optimum values being
10-Inch Model Tests dependent on individual vehicle powering
Tests were performed using a load- and thrust-speed characteristics. High
cell mounted narrow hull form model flow coefficient designs may become
mounted in the high speed carriage at limited by blade root stresses, requiring
NSWC. Full load power was 150 horsepower extra care in analyzing the design and
at 2015 RPM, and design scaled thrust was selecting adequate-strength materials.
1450 lbs at 18 mph. Tests verified
torque-shaft speed predictions within Further waterjet system development
about 2 % and the design thrust within a is needed to address vehicle-propulsor
similar margin. Underway thrust was interaction knowledge gaps and to develop
determined by subtracting bare hull designs that more effectively deal with
(covered inlet) measured drags from air ingestion-related phenomena.
measured loads with the pump operating. Ventilated compact nozzle design
The bollard thrusts fell short of development is needed, and a better
calculated values by some 20 percent, and understanding of the interaction of
a major portion of this discrepancy is distorted inflows and impeller blade
attributable to the unventilated operating unsteady performance should be developed.
condition of the nozzle at very low Continued development of waterjet
forward speeds. Model tests uncovered the technology offers high payoff in terms of
need to change the material and stacking cost reduction and performance
geometry of the inducer to avoid improvement, particularly in the current
unacceptable blade deflections. Several environment where waterjet propulsors are
nozzle plate exit sizes were evaluated finding wider acceptance for both military
during towing tank experiments, with the and commercial applications.
15.1 inch full scale nozzle exit diameter

150
ACKNOWLEDGEMENTS Wong, G. S., "Waterjet Technology
Study', Rockwell International Final
The authors would like to thank Report R-9632, June 1975.
Messrs. Rich Swanek, HWSTD Program
Manager, Steve Ouimette, PSD Program Stripling, L. B. and A. J. Acosta,
Manager, and Walter Zeitfuss, Head of the 'Cavitation in Turbopumps - Part l',
Marine Corps Program Office at NSWC, for Transactions of the ASME, pp. 326-338,
their confidence and support throughout Sept 1962.
the development process of the waterjet
propulsors. We also thank John Hoyt of Bruce, E. P., W. S. Gearhart, J. R.
the Special Ship and Ocean Systems Ross and A. L. Treaster, 'The Design
Dynamics Branch at NSWC for his support of Pumpjets for Hydrodynamic
and also his sense of humor that was Propulsion", Proceedings of the
severely tested on more than one occasion International Symposium on Fluid
during the course of this effort to Mechanics and Design of
develop jets for these vehicles. Finally, Turbomachinery, Sept. 1970.
we thank AAI Corp. for their role in
designing the critical driveline and McBride, M. W., 'Refinement of the
interface hardware, and in assembling and Mean Streamline Method of Blade
installing the test vehicle waterjet Section Design", ASME Journal of
systems Fluids Engineering, pp. 561-566, Sept.
1977.
REFERENCES
Gorski, J. J., 'Solutions of the
Wendel, A., R. Etter, and E. Miller, Incompressible Navier-Stokes
'Concept Study of Transom Flap Mounted Equations using an Upwind-Differenced
Propulsors for High Speed Tracked TVD Scheme", Lecture Notes in Physics,
Amphibious Vehicles', Hydronautics pp. 278-282, Proceedings of the llth
Inc., Technical Report 8210.06-1, International Conference on Numerical
April 1982. Methods in Fluid Dynamics,
Williamburg, USA, 1988.
Lueders, D. and E. Numata, "Waterjet
Propulsion System Performance in a
Manned Testcraft in Calm Water',
Davidson Laboratory, Stevens
Institute of Technology Report SIT-DL-
85-9-2519, March 1985.

Anonymous, 'Large High Water Speed

Technology Demonstration and
Automotive Test Rig Concept
Design Analyses", MAR Inc.
Technical Report No. 513, Oct 1985.

Hoyt, John G., Ernest E. Zarnick,

Roland S. Young, and Kathryn E.
Wilson, "Calm and Rough Water
Performance Evaluation on a .295 Scale
Model of the High Water Speed
Technology Demonstrator" David Taylor
Research Center Report DTRC-89/SHD-
1300-01, Jan 1989.

Roper, J., 'Development of Waterjet

Propulsion Unit" , Davidson Laboratory,
Stevens Institute of Technology Report
SIT-DL-85-9-2328, March 1985.
Burtness, M. N., "Pressure
Distributions and Losses in a Waterjet
Propulsion System', Davidson
Laboratory, Stevens Institute of
Technology Report SIT-DL-87-9-2566,
Sept 1989.

151
 HYDRAULIC MOTOR

COUPLING
KICKER

SHAFT INDUCER

---e412,Z14,1

INLET

NOZZLE

TRANSITION/PUMP HOUSING
Figure 1. Cross Section of PSD Propulsor System

3.00
Tip Incidence Angle = 3 degrees
2.75 - Design Point

2.50 - 2 g NPSH
y2 ax
2.25 -7:
V
ax

2.00 Utl

1.75

1.50 C0=1.2

Co=1.1

1.25 Co=1.0

1.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20

Figure 2. Inducer Suction Performance

152
 430

THRUST = 3767 LBS at 18 MPH -

JVR = 2.6
410 --) DESIGN POINT

0
ILL
__-,
, ,
JVR = 2,4
1

LJ 390-
(f) s1,.
-,,
JVR = 2.2
CC
o __
JVR =
JVR = 1.6

JVR = 1.8
370 -

CC STATOR CAVITATION LIMIT

CO 350 -

330
15.0 17.0 19.0 21.0 23.0 25.0
DIAMETER ( inches )

Figure 3. Design Point Map for 3 Degree Incidence Angle

Thrust 3767 lbs

Craft Speed 18 mph
Headrise 57.5 ft
Flowrate 22994 GPM
NPSH 37.2 ft
RPM 1251
Diameter 16.1 inches
Utip 88 fps
Vax 41.6 fps
Torque 1633 ft-lb
Horsepower 389
Pump Efficiency 88%

Table 1. Waterjet Design Point Characteristics

153
 2.00

cri g = 28.3 degrees

cu
vrei = 97.6 fps
u 1.50 NB = 4
i = 3 degrees
k = .06844

1.00
CD
FA
LIJ

Max. Cavity Length = 6.62 inches

Max. Cavity Height = .363 inches
0.50 -
F-i

0.00
00 1.0 2.0 3.0 4.0 5.0 6.0 70
CAVITY LENGTH ( inches )

Figure 4. Cavity Shape for Inducer Tip Section

2000

Predacted
Measured Bollard HWSTD
Measured Bollard PSD
Measured Underway HUSTD

.../

I I I F

200 400 600 800 1000 1200 1400

RPM

Figure 5. Comparison of Predicted and Measured Torque

154
tr)
 Experience with Hot-Film Anemometry for
Ship Model Nominal Wake Survey in a Towing Tank
S. Wu, N. Bose (Memorial University of Newfoundland, Canada)

the most commonly reported problems encountered were

sensor contamination, large drift of signals and air bub-
ABSTRACT bles on the probe (3,4). The quality of the water in a
towing tank, and the conditions under which a towing
Problems with the calibration and operation of tank is operated, are usually uncontrolled. Existence
hot-film anemometers in water have prevented their of a temperature gradient below the water surface, and
widespread application for towing tank measurements. of drift in the water temperature, pose additional dif-
The potential advantages of their use, in terms of their ficulties. Repeatable calibrations have been difficult to
accuracy compared with pitot tubes and their low cost obtain.
compared with laser-doppler anemometry, have yet to In comparison with pitot tubes, traditionally used for
be fully exploited. Calibration experiments with several ship model work, hot-film probes have potential advan-
hot-film probes were done in a towing tank to investigate tages due to their high spatial and frequency resolu-
their characteristics. After an initial 'burn in' period of tions. Among their advantages, records of the turbu-
about 100 hours of operation, their calibration was found lence can be obtained in addition to mean flow compo-
to be stable and repeatable, although it varied with the nents; the probes are less intrusive to the flow than pitot
ambient water temperature. tubes; and signals from the instruments can be easily
Nominal wake survey experiments were conducted be- recorded on computer based systems leading to rapid
hind a model of a Newfoundland fishing vessel form by and accurate analysis of the data. Compared with laser
using a wedge-shaped single sensor hot-film probe and a doppler anemometry (LDA), these instruments are much
dual sensor vee hot-film probe respectively. The model cheaper; initial purchase costs of a hot-film anemometry
was towed at a speed of 1.45m/s, which corresponded to system are about 1/10 those of an LDA system. It is
a Froude number of 0.33 and full scale vessel speed of possible to envisage a bank of these probes used in a
6.4 knots. Levels of turbulence as well as mean velocity similar manner to the pitot tube rakes for wake survey
values in the wake were obtained. A comparison is made work.
between sample results obtained by the use of the two The behaviour of these instruments was studied in
different probes. experiments conducted in a towing tank with three hot-
film probes. Following this, two nominal wake survey
INTRODUCTION experiments were done with single sensor and dual sen-
sor wedge-shaped hot-film probes behind a model of a
Constant temperature anemometry has been used Newfoundland fishing vessel form. Fishing vessels com-
successfully and extensively for fluid mechanics mea- monly used in Newfoundland have a length/beam ratio
surements in air. Use of these instruments in water, of about 3; their propellers are often poorly designed and
which requires hot-film probes with thick quartz coating, may operate under non-optimal conditions often behind
has been problematical even under controlled conditions a wide skeg. The turbulent as well as the mean velocity
(e.g. 1). A recent review of work done on the calibra-
components of the wake behind the model were obtained
tion and use of hot-film probes in water (2) showed that from both experiments.

157
 itself is usually negligibly small if they are switched on
CALIBRATION OF THE PROBES long enough before tests are done (normally about two
hours).
General Description
The probes were calibrated in a towing tank over a Steady Drift. After about 70-80 hours of opera-
velocity range up to 3 m/s. Change of flow velocity of tion, the drift reduced and became relatively steady.
the water relative to the sensors was achieved by adjust- Calibration curves obtained were now smooth and well-
ing the carriage speed; variation in the direction of the behaved. During this period, calibration runs were
velocity relative to the sensors was achieved by changing done regularly to monitor the variation in the output.
the orientation of the probes. In contrast to most cali- The anemorneter output voltages decreased steadily over
brations done in air, calibration in a towing tank is done time. The ratio of the anemometer output voltages ob-
with the fluid medium (water) still, and the probe in tained at two different times was nearly constant over
steady motion; advantages of this method are that tur- the velocity range (see figure 2). Typically, the differ-
bulence is usually low and the calibration velocity can ence in anemometer output voltages obtained 7 hours
be measured extremely accurately. apart was about 2-3%.
The probes used included two wedge-shaped vee hot-
Low Drift. After about 100 hours of operation, the
film probes (Dantec model 55R72 having two sensors set
at a 900 angle to each other) and one wedge-shaped hot-
level of drift between the anemometer output vo/tages
film probe (Dantec model 55R.32 having one sensor only),
obtained 7 hours apart dropped to below 1%. Eventu-
ally, for one of the vee probes, the calibration was found
all of which had been newly purchased. The anemome-
ters used included one DISA 55M01 and one DISA 55D05 to be repeatable within 0.2%; for the other probes, the
type; both of these were operated in the constant tem- drift level over a two-day period was under 1%. The
perature mode. The output from the anemometers were probes had reached a stable operating condition.
recorded through a Keithley 575 A/D convertor onto a This long 'burn in' period required for the sensors at
micro-computer; the analogue signals were sampled at a the early stages of calibration of a new probe is unlike
rate of 300Hz. The carriage speed was recorded in the
the behaviour shown by hot-wire probes; the irregular
nature of the output can be misleading in the search for
same way.
a stable calibration prior to experimentation. Obviously,
sufficient time has to be allowed for the output of these
Calibration Behaviour probes to stabilize before attempting calibrations and
Output from a new probe exhibited an unsteady be-
measurements. The time required may vary with local
haviour when first used. This initial period can be di-
water quality.
vided roughly into three phases: irregular, steady, and
low drift.
Calibration Models
The calibration curve between anemometer output
Irregular Drift. In the initial setup of all three
voltage, E, and calibration velocity, V, is usually fitted
probes, the sensor operating temperature was set at with King's law (e.g. 5) or some form of modified King's
about 45°C; the anemometers were left in the OPER-
law (e.g. 6); polynomial models have also been used
ATE mode at zero velocity most of the time. Air bub- (7). For these probes, we have found that curve fitting
bles on the tip of the probe were observed persistently accuracy is increased by using a general form of modified
during the first few hours of operation; few bubbles were
King's law (8):
observed after this period. During the first few days, 171

the anemometer output voltages at zero velocity drifted E2 = E AkVk" (1)

significantly: in one of the calibrations involving a vee k.0
probe in continuous operation over a three day period,
the output for one sensor at zero velocity drifted from with Ak and n being calibration constants. A value of
3.17 volts to 2.40 volts; for the other sensor the output m < 3 usually provides sufficient accuracy.
drifted from 3.00 volts to 1.34 volts. During this pe- Yaw sensitivity calibrations are necessary for vee (or
riod, trial calibration runs were recorded over a velocity X) sensors and yawed probes. In these the velocity is
confined to the plane defined by the two sensor axes, or
range up to 2.0m/s. The calibration curves were irregu-
lar. Figure 1 shows an example of the output from one of the sensor and probe axes, and is directed at an angle to
the direction normal to the sensor axis (usually referred
the vee probes for calibrations during this period. Sim-
ilar behaviour was observed with the other two probes. to as the yaw angle and denoted as a). The conventional
calibration model for a yawed sensor is the effective ve-
The large and irregular drift was attributed to deterio-
locity/yaw factor model (5). Again for these sensors,
ration of the sensors. Drift of the anemometer circuitry

158
we have found that a two-variable nonlinear polynomial ities associated with variation of the thermal properties
gives considerably bett,er curve fitting accuracy to the of the water with the ambient temperature could be ne-
data (2); this model is written as glected; the anemometer output was linearly related to
variation of the temperature. A correction was applied
m k
to the calibrations based on the difference between the
E2 = EEAkiVinak-i (2) actual water temperature and the temperature at which
k.0j=0 the calibration was done. The variation of the sensor
in which Akj and n are experimentally derived constants output with temperature was determined from two cali-
from the calibration. A typical yaw sensitivity calibra- brations done at different temperatures by assuming that
tion curve is given in Figure 3; this shows that over a the variation between the calibrations was entirely due
wide range of yaw angles, the sensor output fell below to variations in temperature; the calibrations were usu-
the cosine law curve representing the minimum of the ally taken before and after a series of tests.
yaw factor model. As a result, curve fitting errors at in-
dividual calibration points over this range can amount to NOMINAL WAKE SURVEY EXPERIMENTS
as much as 12%. With the two-variable nonlinear poly-
nomial model, the maxirnum errors at individual points The fishing vessel model had been used previously for
can be reduced by as much as 85%. Figure 4 shows the resistance and self-propulsion tests (9); it was typical of
accuracy of the two-variable nonlinear polynomial model vessels of its size used in the Newfoundland fishery. The
compared with that of the conventional model. body plan and profile can be found in references (9,10).
The model had a scale ratio of 1/5; waterline length
Water Quality and Temperature of 2.03m; waterline beam of 0.67m at mid-ships; and a
The most important fa,ctors concerning the towing design draught of 0.21m.
tank conditions are the quality of the water in the tank, The hot-film probe was mounted on a support that
and its temperature fluctuation. The water in the towing was fastened to a dummy propeller shaft. Circumferen-
tank was tap water from the city main and its quality was tial movements of the probe were made by rotating the
relatively poor. Dirt and algae were visible. No filter- shaft, while radial movements were made by reposition-
ing or chemical treatment of the water was done. Despite ing the probe in the support.
this, during the calibration of these three probes, no sen- The single sensor probe was used in the first wake
sor contamination was observed; these tests were done survey. In this, the tangential velocity component in the
separately and they each lasted about three weeks. Low propeller disk, Vt, and the radial component, 17,., were
contamination may have been due to the wedge shape of both assumed to be small in comparison with the ax-
the probes. ial component, Va, and the probe was used to estimate
The water temperature fluctuated with the laboratory the axial velocity component of the wake. As the actual
air temperature, which varied with the open air temper- velocity components, Vt and 17,. are nonzero, two orien-
ature. To minimize the effect of temperature variation, tations of the probe were considered to try to minimize
the probes were submerged 0.35 m (1.2ft) below the sur- the effect of these components: a) in the plane defined
face at which level the temperature was relatively steady. by V, and Vi, assuming that V, was smaller than Vt; or
However, there was a temperature gradient over depth b) in the plane defined by Va and V,., assuming that V*
during the first wake survey experiment; one measure- was smaller than V,.; the latter was chosen in these tests.
ment of this showed a temperature of 17°C at the sur- One of the dual sensor probes was used in the second
face, 16°C at 0.25m below the surface, 15.5°C at 0.5m wake survey experiment, and its plane (that defined by
below the surface, and 15°C at the bottom, 1.75m be- the two sensor axes) was oriented as in the first wake
low the surface. The tank had been refilled just prior survey experiment. In both tests, measurements were
to the experiment and this exacerbated the temperature made within the half propeller disk on the port side,
gradient. along five half circles within a 12 cm diameter propeller
Output from these thermal sensors is sensitive to disk. The model was towed at 1.45m/s.
changes in the ambient temperature. To account for this, For clarity of the comparison, results from both exper-
periodical recalibrations are necessary, or a temperature iments of the mean Taylor wake fraction along the half
correction can be applied to the original calibration. Al- circle of radius 4 cm only, are shown in Figure 5; the
ternatively, the effect can be minimized by using a bank mean Taylor wake fraction is defined as (Vm - VA )1Vm
of probes and completing the experiment over a short with Vm and VA being the speed of the model and mean
period of time. During these experiments, temperature speed of advance respectively. Results over the whole
variations were within 1.5°C. This temperature variation propeller disc for the experiment with the single sensor
was small and led to the assumption that the nonlinear- probe are presented in reference (10); for the experiment

159
with the dual sensor probe, results are given in reference hour period. Sufficient time must be allowed for the
(11). output from these probes to stabilize before use.
In the wake of the skeg, the flow was complex: re- Output from the probes varied with water temper-
verse flows and vortex shedding occurred, which were ature. Water temperature variation was accounted
evidenced by the low mean velocity component and com- for by monitoring the water temperature and ap-
paratively high standard deviation of the velocity records plying corrections to the calibrations. Alternative
(see Figures 6 and 7). In this region, the wake fraction methods would be to periodically recalibrate the
results from both experiments are only qualitatively cor- probe, or to use a bank of probes to reduce the
rect; they show very high values of the wake fraction time period for the wake survey to be completed.
(greater than 0.5). Outside of this central region the
flow was more streamlined. The assumption of small The yaw sensitivity of the probes did not fit the
radial and/or tangential velocity components compared conventional yaw model for hot-wire probes. Use
with the axial component, necessary for the analysis of of a direct two-variable nonlinear polynomial fit is
velocity components from the sensor output, is more re- recommended.
alistic. Results of the simultaneous measurement of the If radial and tangential wake components are low
radial and axial velocity components from the second in comparison with axial components, single sen-
wake survey experiment (found by assuming the tangen- sor probes may be used to obtain the axial wake
tial component of velocity to be small) confirm this as- distribution. More generally, results show that sig-
sumption; the radial velocity component has a maximum nificant differences occur between measurements of
value of about 23% of the maximum axial velocity com- wake done with single and twin sensor probes, espe-
ponent. The axial velocities from the second experiment cially in regions at high turbulence and separation;
were expectedly lower in general than than those from multi-sensor probes should be used when possible.
the first experiment. For a single sensor probe, nonzero
radial and tangential velocity components result in an Hot-film sensors can be used to obtain measure-
increase in the anemometer output voltage from that ments of the turbulence spectrum in the wake. For
induced by the axial velocity component alone. As a re- these tests, significant energy levels of the turbu-
sult, the actual axial velocity component measured by a lence occurred at frequencies below 20 Hz; very low
single sensor probe, is overestimated. levels of turbulence existed in the flow outside of
The frequency content of the axial wake velocity at the wake of the bluff skeg.
several positions along the half circle of 4 cm radius is
shown in Figure 8; these are taken from the experiment ACKNOWLEDGEMENTS
with the twin sensor probe. The maximum turbulence
occurs near to the edge of the skeg. Significant energy is This work was co-sponsored by the Canadian Centre
concentrated at frequencies below 20 Hz, but these lev- for Fisheries Innovation, St. John's; Natural Sciences
els are much lower than the zero frequency component and Engineering Research Council, Canada; Avalon
(which is off the scale on these plots). Outside of the Propulsion Systems Ltd., St. John's; the Institute for
central region far away from the skeg, there is little tur- Marine Dynamics, National Research Council; and ad-
bulence in the axial direction. ministered by Seabright Corporation. We thank the
Ocean Engineering Research Centre, Faculty of Engi-
CONCLUSIONS neering and Applied Science, Memorial University of
Newfoundland, for providing the tank facilities; Mr.
The calibration behaviour of three newly-purchased Thomas Hofmann for building the ship model; and all
wedge-shaped hot-film probes was studied to in- the technicians and research staff who contributed to the
vestigate their feasibility for towing tank applica- work.
tions. Repeatable calibrations were obtained and
two nominal wake survey experiments were con- REFERENCES
ducted.
Air bubbles were observed during the first few hours 1 Fingerson, L.M. and Freymuth, P. 1983, " Thermal
of operation. Irregular drift in the anemometer out- anemometry" in Fluid Mechanics Measurements ed.
put voltages was encountered during the first 70-80 R. J. Goldstein, Hemisphere Publishing Co..
hours of operation, followed by a period of steady, 2 Wu, S. and Bose, N., 1991, "Calibration of a wedge-
but still significant, drift. After about 100 hours, shaped vee hot film probe in a towing tank" , submit-
the calibrations were repeatable within 1% over a 7 ted to J. of Phys. E: Sci. Instrum.

160
3 Goodman, C.H. and Sogin, H.H., 1974, " Calibration
of a hot fiLm anemometer in water over the velocity
range 0.5 to 200cm/s", Flow - Its Measurement and 5

Control in Science and Industry. Part Two - Flow

Measuring Devices edited by Wendt, R.E., Jr., In-
strument Society of America, pp. 589 - 598.
4 Okuno, T. and Himeno, Y., 1982, "'Turbulence mea- o

surements in ship model flow", International Union

of Theoretical and Applied Mechanics Symposium et
on Three-Dimensional Turbulent Boundary Layers
edited by Fernholz, H.H and Krause, E., Berlin, Ger-
manY, PP. 55 65.
5 Jorgensen, F.E. 1971, "Directional sensitivity of wire
and fibre-film probes", DISA Information, no. 11.
6 Win H. and Prahl, J.M., 1986, "Calibration of 0.5 1 1.5

constant-temperature hot-film anemometers at low calibration velocity (m/s)

velocities in water of uniform temperature", Inter.
Comm. Heat Sz Mass Transfer, vol. 13, 567. Figure 1 Calibration data taken over a two-hour period during the
7 George, W.K., Beuther, P.D., and Shabbir, A., initial stage of operation (irregular drift); the marked points are the
1987, "Polynomial Calibrations for Hot Wires actual calibration data points from the three calibrations.
in Thermally-Varying Flows", Symp. Thermal
Anemometry, D.E. Stock (ed), ASME FED-Vol. 53,
ppl 6.
8 Wu, S. and Bose, N., 1991, "Methods of fitting cal-
ibration curves to hot film/hot wire constant tem-
perature anemometer data", Ocean Engineering Re-
search Centre R,eport OERC91-W'TT-TR003, Fac- 7.5
ulty of Engineering and Applied science, Memorial
University of Newfoundland.
9 Wu, S. and Bose, N., 1991, "An experimental inves-
tigation into the propulsive efficiency of two New-
foundland fishing vessel forms", Ocean Engineer-
ing R,esearch Centre Report OERC91-WTT-TR001,
Faculty of Engineering and Applied Science, Memo-
rial University of Newfoundland.
10 Wu, S., and Bose, N., 1992, "Axial wake survey be- taken on amsday
hind a fishing vessel model by using a wedge-shaped o taken on VVednesday
hot-film probe", to be presented at the Kansai Soci-
ety of Naval Architects Spring Meeting 1992, Kobe, 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
J apan. calibration velocity (m/s)
11 Wu, S., and Bose, N., 1992, "Nominal wake survey
behind a ship model by using a dual sensor hot-film Figure 2 Calibration data for one sensor of a twin sensor probe taken
eight hours apart during the steady drift stage.
probe", Ocean Engineering Research Centre Report
in preparation, Faculty of Engineering and Applied
Science, Memorial University of Newfoundland.

161
 6

L.

a ,
o 5'
-. 6 , "a
7.;
I

á !

-, 1

3. i

o maximum and minimum calibration valu. s'o 4r

+ mean calibration values calibration data points
1
... (2) .g
0.21_ (1) cosine law: cosa
- - - - effective velocity /yaw factor model
i 3.51- two-variable polynomial model
(2) modified cosine law: (c0820 + 0.242 x sin2cr)1/2
;

C1) E
I

40 50 60 70 SO 90
53
'E

0 0.5 1.5
10 20 30 1

yaw angle a (degrees) calibration velocity (m/s)

Figure 3 Variation of yaw function (ratio of the effective and cal-

ibration velocities) with yaw angle for one sensor of a dual sensor Figure 4 Calibration data points and curve fits by using the con-
wedge-shaped hot-film probe; results for the other sensor were similar. ventional effective velocity / yaw factor model and a tiro-variable non-
linear polynomial model respectively for the calibration of one sensor
of a ves probe at a yaw angle of 30°.

1.4

so.

2 7
o mean component
standard deviation
-5 0.4 - - - with the single sensor probe
4 with the dual sensor probe

0.2

e ........ _ .......
(cm)
o .

120 140 160 180

20 40 60 80 100
6 4 2 o 2
position angle (degrees)

Figure 5 Wake fraction (in percent) from two nominal wave survey Figure 6 Measured mean and standard deviation of the axial velocity
experiments along a half circle of radius of 4cm; the values in paren- component of the wake along a half circle of 4cm radius (see Figure
theses are from an experiment with a dual sensor probe, and the others 5) frorn both experiments; positions of the measurement are shown in
are from an experiment with a single sensor probe. Figure 5.

162
 Q3

3-
Q25
position angle 30°
1-

o
3r

2+ position angle 90°

-o
o'

'711,
n.
1
position angle 150°

o 10 20 30 40 50 60

o frequency (Hz)

Figure 8 Energy spectrum of the measured axial velocity components

80 100 120 140 160 180 of the wake at several positions along the 4cm half circle (obtained
20 40 60
from the experiment with the dual sensor probe).
position angle (degrees)

Figure 7 Measured mean and standard deviation of the radial veloc-

ity component of the wake along a half circle of 4 cm radius from the
experiment with the dual sensor probe; positions of the measurements
are shown in Figure 5.

163
 Numerical vs. Experimental Cavitation Tunnel
(A supercavitating hydrofoil experiment)
S.A. Kinnas, C.H. Mazel (Massachusetts Institute of Technology, USA)

Abstract
A supercavitating hydrofoil experiment is performed at
the MIT Variable Pressure water tunnel, at several an-
gles of attack and cavitation numbers. The velocity is
measured on a rectangular contour surrounding the foil
and cavity, and the results are compared to those from
an analysis method_ Forces are inferred from momentum
integrations of the measured velocities and also compared
\
\
Pivot Itcd
.../
Stainless Steel.
x Housing

a .. washer

to those predicted from the analysis.

1 Introduction and Objectives

Numerous experiments have been performed of different
bodies in cavitating flow and in particular of hydrofoils
and propellers. The majority of these experiments have
focused on visualizing the cavity shapes, measuring forces Figure 1: A perspective view of the foil and its supports
and making comparisons against predictions of different
analysis methods for the same flow conditions (cavitation
number). presented by Kerwin [3]. The maximum foil thickness was
In the present experiment, even though we considered 11/16". The foil' was attached to the acrylic windows of
the previous aspects, we have mainly been concerned with the water tunnel at two points on either end. The foil
measuring the detailed flow field around a supercavitat- was free to pivot about the upstream attachment points,
ing hydrofoil and then making direct comparisons against while the downstream attachments (pins) were fixed but
adjustable. This arrangement, shown schematically in
the velocity flow field predicted from analysis. In the de-
scribed analysis the effects of the tunnel walls have been Figure 1, enabled us to change the angle of attack as
desired.
accounted for by imaging the singularities representing
the cavity and the foil with respect to the walls. The A narrow slot was milled into each end of the foil, and a
forces acting on the cavitating hydrofoil have been de- length of rubber 0-ring material (gasket) was fitted into
duced from momentum integrations of the measured ve- this slot. When the foil was mounted in the water tunnel
locity flow field. the rubber .was squeezed between the foil and the window,
preventing cross-flow from the pressure to the suction side
of the foil.
2 Experiment
2.2 Measurement of angle of attack
2.1 The foil and mounting mechanism
The angle of attack, a, of the foil was measured using
The test foil, a supercavitating section with high loading the laser beams from the velocimeter system. The laser
at the trailing edge, was machined from aluminum and was positioned so that the horizontal and vertical beam
anodized. The chord length was 8", with a 20" span pairs were aligned with the foil leading edge at midspan,
to traverse the 20" x 20" test section of the MIT water and the coordinates of the 3-axis positioning system were
tunnel. A description of the MIT water tunnel and its set to zero at this location. The laser support table was
Laser Doppler Velocimetry- (LDV) system was recently then moved until the beams were aligned with the trailing
Principal Research Engineer 8, Lecturer
edge, and the new coordinates were used to compute a.
2 Research Engineer
3Its complete geometry is given in [7].
165
Figure 2: A photograph of the foil at a = 1.85° with a
supercavity of / = 3

Velocity measurements vvere made at a = 1.85, 3.10, and

4.85 degrees.
Figure 3: A photograph of the foil at a = 1.85° with a
supercavity of / = 1.5
2.3 Notes on cavity characteristics
At each a, measurements were made with cavity lengths, tended downstream approximately 1 inch for 1 = 2, and
/, of 1.5, 2, 2.5, and 3, defined as the ratio of the cav- approximately 2-3 inches for / = 2.5 and / = 3.
ity length (the distance from the leading egde of the foil At a = 3.1° the cavitation on the suction side started
to the trailing edge of the cavity) to the chord length. approximately 2 inches from the leading edge for all cavity
Photographs of the supercavitating foil inside the tunnel lengths. There was no established face cavitation for any
are shown in Figures 2 and 3. Several test runs were cavity length. At / = 3 there was occasional, short-lived
made at a = 3.1° with the foil fully wetted, in order to cavitation on the pressure side, starting at isolated points
test the numerical techniques against a more predictable
along the leading edge.
condition. At a = 4.85° the cavitation on the suction side started
The cavity length was set by running the water tunnel at the leading edge for all cavity lengths. There was no
at the desired test speed and adjusting the tunnel pres- face cavitation.
sure. The cavities were approximately two-dimensional
(cavity length constant across the span) under all test
conditions. By making small adjustments to the tunnel 2.4 Experimental procedures and special
pressure it was possible to maintain the cavity length at conditions
or acceptably close to the desired value. In a few cases the
system was more sensitive, and the cavity length would Fluid velocity measurements were made with a two-
spontaneously shrink or grow. This seemed to be related component laser doppler velocimeter system[3], permit-
and ver-
to the amount of bubbles in the water. In these cases data ting for simultaneous acquisition of streamwise
measured at
runs along a leg were repeated after the cavity length had tical velocity components. Velocities were
points in a rectangular box pattern surrounding the foil
stabilized.
For each condition the extent of cavitation on the pres- and cavity. The sampling rectangle was 68cm in the hor-
izontal (streamwise) direction and 20 cm in the vertical
sure side (face cavitation), and the location of cavity de-
tachment on the suction side, were determined visually. direction. The sample spacing along the horizontal por-
At a = 1.85° the cavitation on the suction side started at tions of the rectangle, above and below the foil, was
10
the vertical portions of the rect-
about mid-chord for all cavity lengths. There was some mm. The spacing along
face cavitation in all cases, starting at the leading edge. angle was 5 mm, except in the wake region, where the
The face cavitation was barely visible at / = 1.5, and ex- sample spacing was reduced to 2 mm.

166
 At one set of operating conditions (a = 5.21°, I = 1.75, The input (also shown in Figure 4):
= 0.05, a- = 0.43, U = 19.7 ftls), in an attempt to
measure the fluid velocity on the cavity surface for com- The foil section geometry and the angle of at-
parison with numerical prediction, the horizontal velocity tack a.
component was measured at points along a vertical path The cavity length 1.
approaching the cavity. This was done at two locations
above the cavity, and one location below. The cavity detachment point on the upper sur-
face,
2.5 Data acquisition and quality control The length of the cavity termination model, A.
We have the options to apply a "curved plate
The laser system was programmed to acquire 750 velocity model", as shown in Figure 4, or a prescribed
readings at each sample location. Under optimal condi- pressure law model (the same on both sides of
tions these samples could be acquired within a few sec- the cavity-) followed by an open cavity wake. In
onds, but in the presence of a large number of bubbles the case of the "curved plate model" the plate is
it took 20 seconds or more. Within the data acquisi- allowed to move up and down but is restricted
tion program the array of 750 velocity values was pro- from rotating. The models will be described in
cessed to determine the mean and standard deviation for more detail in a future report.
each velocity component. The program then discarded
any points which fell more than three standard deviations The output:
from the mean, and recomputed the means and standard
deviations from the reduced data set. The corresponding cavitation number, u, de-
It was necessary to add an additional data quality check fined as usual:
for this experiment. For reasons unknown to us and to
the equipment manufacturer, when large numbers of cav-
o.= P. Pc
(1)
itation bubbles were present the laser system sometimes U2
2 0.
produced anomalously high velocity readings. A software
filter that discarded any points with a velocity greater The cavity shape. It is determined in an iter-
than 35 feet per second, well above the range of real val- ative manner with the first iteration being the
ues, was added to the data processing software. cavity shape from linearized cavity theory.
The pressure distribution and forces (lift L and
3 Numerical Method drag D) acting on the foil. The lift is deter-
mined by integrating the pressure forces in the
direction normal to the inflow. The drag con-
3.1 Cavitating hydrofoil in unbounded sists of two components: (a) the ideal flow cav-
flow ity drag which is determined by integrating the
The fundamentals of the numerical method are described pressure forces in the direction of the inflow and,
in detail by Kinnas and Fine in [6]. A supercavitating (b) the viscous drag which is determined by ap-
hydrofoil is depicted in Figure 4 (for convenience we take plying a uniform friction coefficient, Cf , over the
the inflow U,,, -= 1 and the foil chord length c = 1). The wetted part of the foil.
main characteristics of the method may be summarized The velocity field at any point in the flow field.
as follows:
The numerics of the method have been validated exten-
It is a completely non-linear perturbation potential sively in [6]. In general, for supercavitating foils at moder-
based panel method. The foil and cavity are modeled ate angles of attack, linear theory, compared against the
with constant strength source and dipole panels. The panel method, predicts the cavity shape and the forces
method solves for the unknown perturbation poten- with acceptable accuracy[5,2]. However, the velocity- flow
tials 0 (dipole strengths) on the fully wetted hydro- field, especially close to the cavity or foil, is computed
foil and for the unknown source strengths, 1?-2, on the more accurately by the panel method than by the linear
cavity surface. The source strengths are known on theory. This is due to the fact that in linear cavity theory
the wetted foil via the kinematic boundary condition the upper and lower surfaces of the foil or cavity are as-
and the potentials are known on the cavity via the sumed to collapse in one segment (the branch cut in the
dynamic boundary condition. The dy,-narnic bound- complex plane), as opposed to the panel method in which
ary condition requires the pressure on the cavity to the singularities lie on the exact foil and cavity surface.
be constant, p pc. Since one of the objectives of the present work was to

167
 /.,///// /ly
uc,0 u

/7 ////// //////// /7
Figure 4: Supercavitating foil

compare the velocities from experiment and analysis, it Figure 5: Supercavitating foil inside of a tunnel. The wall
was decided to use the panel method. is represented with images of the foil and cavity singular-
In the case of supercavitation, the forces as well as the ities. The case with number of reflections NIMAGE= 1
cavity shapes have been found to depend very little on is shown.
A[1]. Even though we could determine A from LDV mea-
surements at the end of the cavity, it is a fortunate result
NIMAGE CL CD
that its value does not affect the solution substantially.
In all the upcoming computations we used the same value 0 0.429 0.038
of A, A =0.1.
10 0.445 0.040
3.2 Cavitating hydrofoil inside of a tun-
nel Table 1: Dependence of forces on the number of images;
a = 1.85°, / = 1.5, hIc = 2.5. NIMAGE = 0 corre-
A supercavitating hydrofoil inside of a tunnel is shown sponds to unbounded flow.
in Figure 5. The height of the tunnel is h. The effect
of the tunnel walls can be accounted for in the previ-
ously described method by including in the formulation smaller in unbounded flow, as can be seen by comparing
the influence of the infinite set of images of all the sin- parts (b) and (c) of Figure 6. Therefore, the tunnel wall
gularities representing the foil and cavity with respect to effects must be included when making comparisons be-
both walls. In practice, a set of ten images (i.e. number tween analysis and experiment. The effect of walls on the
of reflections NIMAGE = 10) is enough in order for the forces is shown in Table 1.
zero normal velocity condition on the walls to be satis-
fied within acceptable accuracy. The effect of the tun-
nel walls on the cavity solution can be seen in Figure 6. 4 Numerical vs. Experimental
The cavity shapes and pressure distributions are shown
as predicted by the present method: (a) in unbounded
Results
flow, (b) when the images are included and for the same The hydrofoil which was tested in the experiment de-
cavity length as in (a) and, (c) in unbounded flow with scribed in Section 2, was analyzed at the same angles of
the cavity length corresponding to the same cavitation attack and cavity lengths to those in the experiment. The
number as in (b). Comparing (a) with (b) notice that the non-linear panel method, with the wall effects included,
cavity shapes do not differ substantially from each other. as described in Section 3, was utilized. Comparisons of
However, the corresponding cavitation number4 is higher the velocities and forces are given next. The cavity length
when the tunnel walls are included. Thus, for the same of / = 2.5 was found to be the most unstable during the
cavitation number the cavity length and volume will be experiment and thus, the corresponding results are not
4Remember that c = Cp on the cavity. shown.

168
 0 50

0 30
I.--' a
.70,..................sree.....", 4.1 Velocities
6
At one set of conditions we measured horizontal veloci-
01
6
i ties at several points on a vertical path approaching the
cavity. The measured velocities are shown against those
.; predicted from the described analysis in Figure 7. The
io overall comparison is very good. The velocities in the
analysis were computed as the sum of the contributions
-0.30 of all dipoles and sources on the foil and cavity as well as
those of their images. Notice that the computed velocities
smoothly extrapolate to the cavity velocity qc which has
-0 59 co
0.50 1. OD
XeC
1- 50 2_00 been computed from the formula q, = 7, where a is
the cavitation number predicted from the analysis. This
provides a very strong test of the validity of the numerical
(a) computations. On the other hand, the quality of the mea-
surements deteriorates as the laser beams get very close to
the cavity. This may be attributed to the unsteadiness as
well as variations in the spanwise direction' of the cavity
surface. Figure 7 suggests that in order to find the cav-
ity velocity, the measurements should be extrapolated to
the location of cavity surface (determined visually with
0.10 the help of the laser) from the measurements away from
the cavity. Since the measurements do not vary- smoothly
.; with vertical distance the numerical curve (scaled by a
;. -o to
constant factor which is determined from minimizing the
error bewteen this curve and the measurements in a least
-O. 30 squares sense) may be used in the extrapolation. This ex-
trapolation will be more successful the smaller the slope
of the velocity with distance (du/dz) is. It can be proven
that the slope duldz is proportional to q,1R where R is
the radius of curvature of the cavity. For example, the
(b) curvature of the cavity shown in Figure 7 is largest at the
bottom, where du/dz is larger than those corresponding
to the other two locations.
A representative comparison between the numerical vs.
experimental velocities is shown in Figures 8 and 9 for
a = 3.1° and I = 1.5 and I = 3.0. The absolute value
of the inflow velocity U.0 (given at the caption of the
figures) was determined by scaling the numerical veloci-
ties through the left side of the rectangle until their mean
value was the same to the mean value of the corresponding
experimental velocities. The complete set of comparisons
for all tested angles and cavity lengths may- be found in
[7]. From these figures we may notice the following:
The overall agreement is very good.
The analysis, which only models the potential flow,
(e) is not able to predict the velocity defect in the wake
Figure 6: Effect of walls (represented with images) on
of the cavity. However, it predicts the flow outside
the predicted cavity shapes: (a) unbounded flow, 1 = 1.5 the wake boundary layer quite accurately.
( b) with images, I = 1.5 (c) unbounded flow at the same
cavitation number as in b. In all cases a = 1.85° and The agreement seems to deteriorate with increasing
I, = 0.3. The height of the tunnel is IzIc = 2.5. cavity length. In particular, for 1 = 3 an appreciable
5 The cavity velocity is measured with the set of horizontal beams
vvhich are in effect "grazing" the cavity surface.

169
 shift (in the order of 10% of the inflow) between the disagreement than the velocities, simply because of the
numerical and the measured velocities is observed at magnified error in the difference between the velocities
the right side of the rectangle. In an attempt to above and under the foil and cavity.
reduce this shift we applied an open cavity model
in our analysis. The ultimate openness of the cav-
ity was determined from the computed cavity drag 6 Acknowledgments
by applying momentum theory. A main parameter The experimental part of this work was sponsored by
in this model is the length of a transition region in the Propulsion Group at Volvo-Penta of Sweden. The
which the cavity wake thickness reduces gradually analytical part was sponsored by the Applied Hydrome-
from its value at the cavity end to its ultimate wake chanics Research Program administered by the Office of
value. By applying this model and by changing the Naval Research (Contract: N00014-90-J-1086). The au-
mentioned parameter, we were able to improve the thors would like to thank Thor Hendrickson, a junior at
comparison in the vicinity of the cavity end without, Mechanical Engineering, Neal Fine, a doctoral graduate
however, reducing the shift at the right end of the student and, Matt Knapp, a freshman at MIT, for their
rectangle substantially. A possible explanation for help with the experimental and numerical aspects of this
this shift may be the reduced effective density of the
work.
fluid due to the presence of dispersed cavity bubbles,
especially behind the cavity end. We did not pursue
this issue any further. References
N. E. Fine. Non-linear Analysis of Unsteady Cavitating
4.2 Forces Flows Around Hydrofoils and Propellers. PhD thesis, De-
The experimental forces were computed by integrating partment of Ocean Engineering, MIT, 1992. Under prepa-
the velocities along the rectangular contour as described ration.
in [4]. The forces from the analysis have been computed N.E. Fine and S.A. Kinnas. A Boundary Element Method
as described in Section 3. In computing the viscous drag a for the Analysis of the Flow Around 3-D Cavitating Hy-
uniform friction coefficient. C f = 0.002, has been applied. drofoils. March 1992. Submitted for Publication.
A representative comparison between the forces from ex-
periment and analysis is shown in Figure 10 for a = 3.1°. J.E. Kerwin. The MIT Marine Hydrodynamics Water
The complete set of comparisons for the other angles and Tunnel - A 53rd Anniversary Celebration. May 21 1991.
cavity lengths may be found in [7]. The lift and drag Presented at the New England Section SNAME Meeting,
MIT.
coefficients are defined as usual from CL = L/(01)c
and CD =Di (WC). Even though the analysis seems S.A. Kinnas. Hydrofoil Lift and Drag from Momentum
to predict the cOrrect trends (especially for the lift), the Integrations. Technical Report No. 91-4, Department of
comparisons are not as favorable as those of the veloci- Ocean Engineering, MIT, November 1991.
ties. For the lift this may be attributed to the fact that its S.A. Kinnas. Inversion of the source and vorticity equa-
value is essentially proportional to the difference between tions for supercavitating hydrofoils. Journa/ of Engineer-
the horizontal velocities at the top and bottom sides of ing Mathematics, 26:349-361, 1992.
the rectangle. The error in predicting the lift is of the
same order as the error in predicting the mentioned dif- S.A. Kinnas and N.E. Fine. Non-Linear Analysis of the
ference. The relatively small error between the numerical Flow Around Partially or Super-Cavitating Hydrofoils by
and measured velocities however, is magnified when ex- a Potential Based Panel Method. In Boundary Inte-
gral Methods-Theory and Applications, Proceedings of the
pressed in terms of the difference between the velocities IABEM-90 Symposium, Rorne, Italy, October 15-19, 1990,
at the top and the bottom of the rectangle. pages 289-300, Springer-Verlag, Heidelberg, 1991.
S.A. Kinnas and C.H. Mazel. Numerical vs. Experimental
5 Conclusions Cavitation Tunne/. Technical Report No. 91-6, Depart-
ment of Ocean Engineering, MIT, December 1991.
A systematic comparison between experiment and anal-
ysis has been performed for a supercavitating hydrofoil.
The effects of the tunnel walls were modeled completely
in the analysis method. The velocities were found to be
in very good agreement, especially for cavities extending
no more than one chord length behind the trailing edge
of the foil. However, the forces seemed to be in larger
170
 1.20

1.00

0. 00

0. 60

0.40

0.20

0. °°0 so 1.00 L 10
X/ C =O. 730
1.0e 1.10 1.30 1. 40
u/ro
X/ C =1. 204

1.10 1.20 1.40

./0
X/ C =1. 2154 scrr)

Figure 7: Numerical vs. experimental velocities as a function of the vertical distance from the cavity boundary at
different chordwise locations; a = 5.21°, 1 = 1.75, lo = 0.05, c = 0.43.

171
 1.20 -

ti.) 0.80 above foil experimental

below foil experimental
analytical
1.20

0.60

LOO

etz
0.40 1 11414 )1.)

1.00 0.00 -1.00 -2.00 -3.00 -4.00

-X C re
a) 0.80 above foil experimental
Q) 00000 below foil experimental
- analytical
0.60 -
0.60
0.40 - Downstream
- Analytical Upstream
- - Analytical Downstream
Upstream
0.20 - 0.40 I1,11,111111111,11114.4411/1,111411,
1.00 0.00 -1.00 -2.00 -3.00 -4.00
-X C

0.00 -
Figure 9: Numerical vs. experimental horizontal veloc-
ities. Top and bottom sides of the rectangle; Uo, =
0.20 - 22.4 ft/s, ce = 3.1°, / = 3.0

0.40 -_

0.60
0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10
Vx/free stream

Figure 8: Numerical vs. experimental horizontal veloc-

ities at each side of the rectangle; //co = 23.8 ft/s,
= 3.1°, 1 = 1.5

172
 0.6

0.5
CL vs I

o 0.4

0.3
C)
o
o
0.2
7.1

0.1

o 721111,
1.5 2 2.5 3

0.05
0.045
0.04
0 0.035
"ff
0.03
Ea) 0.025
o
c.) 0.02
co
al 0.015
0.01
0.005
o
1.5 2 2.5 3
Cavity Length/Chord

110 Experimental an Numerical

Figure 10: Numerical vs. experimental lift and drag; a = 3.1°

173
 Design Method and Application of an Asymmetric
Stator Upstream of an Inclined Shaft Propeller
S.K. Neely, J. McMahon, B.Y-H. Chen
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT Propeller radius

R, Stator radius
A method is presented for the design of an asymmetric Thickness
stator upstream of a propeller in an inclined flow. Two VR Relative velocity
contrasting applications of this method are presented. The
Vs Ship speed
first application is an asymmetric stator and a matched pro-
peller design for a typical surface ship with an inclined shaft Tangential velocity
propeller. The primary purpose of this design is simply to Vx Axial velocity
validate the design method. A new propeller and a large 0 Pitch angle
span asymmetric stator are designed which increase propul- o. Cavitation number, normalized by VR:
sive efficiency relative to an existing propeller. Model test = pV;
results are provided which show that significant flow
smoothing is achieved across the entire span of the pro- as Cavitation number, normalized by V,:
peller. The model tests show that the predicted increase in = (P. - PVs2
propulsive efficiency of 6-7% was achieved. The second
Fluid density
application is a backfit to a high speed U.S. Coast Guard pa-
trol boat. The purpose of this design is to eliminate severe Circulation
cavitation erosion near the root of the existing propeller. A
small span asymmetric stator is designed to operate with the
existing propeller. Details of the design and cavitation pre- INTRODUCTION
dictions are provide,d. Cavitation erosion tests in the DT/V1B
water tunnel demonstrate that the severe erosion can be Propellers often are required to operate in an oblique
eliminated. flow, usually due to an inclined shaft. The result is a once
per revolution variation in tangential velocity into the pro-
peller which degrades cavitation performance. The purpose
NOMENCLATURE of an asymmetric stator is to reduce the nonuniforrnity in the
propeller inflow, and therefore increase the cavitation in-
C Chord length ception speed. At the same time, preswirl is added to pro-
C,
peller inflow which can increase the efficiency of the
Lift coefficient: C, = (dL/dr)I-f pVR2e propulsor.
C, Pressure coefficient, normalized by VR: A historical review of the stator vane development and
= (p - p,)1-1pV, an asymmetric stator design philosophy has been presented
by Gearhart and Marboe [1]. Gearhart's method was used to
Cps Pressure coefficient, normalized by Vs: design an asymmetric stator for a U.S. Coast Guard 41 ft
C =(p-P-)1+ Pl7s2 utility boat with results presented by Larimer, et al [2]. An
alternate method has also been presented by Kerwin, et al
CT, Thrust loading coefficient: Cr, = T14pV,27ER2 [3]. Kerwin's method, which employs lifting-line and lifting
Diameter surface theory, was demonstrated for an axisymmetric and
Maximum camber an asymmetric stator designed to operate with an existing
Non-dimensional circulation: G = TI27ERV, propeller. Water tunnel test data were presented.
J, Advance coefficient: J = V, I nD The objective of this paper is to present two distinct
Lift applications of an asymmetric stator upstream of an inclined
Shaft speed (rev/sec) shaft propeller. A brief summary of the design method is
given first, followe,d by two design applications. A brief
Local pressure
P- description of a large span stator design and a matched pro-
Ambient pressure peller design is presented first, followed by a more detailed
r Local radius

175
 the propeller plane is determined as the sum of the nominal
description of a small span stator design which operates
wake and the stator induced velocity.
with an existing propeller. The objective of the asymmetric design phase is to
minimize the nonuniformity of the inflow to the propeller.
The total circulation from all of the axisymmetric stator
DESIGN METHOD blades is redistributed circumferentially in order to achieve
this goal. Since the same total circulation is maintained, the
A design method for an asymmetric stator upstream of propeller will se,e the same circumferential mean velocity as
Kerwin, et the axisymmetric case, and the propeller forces will be ap-
an inclined shaft propeller has been presented by
al [3]. The same general approach was used for the two ap- proximately the same. The stator forces, however, may be
plications presente,d in this paper. This section presents a somewhat different due to the asymmetry of the inflow. If
brief summary of this design method. this difference is large, the propeller circulation may need to
The design process involves two stages. The first stage be adjusted in order to obtain the desired total thrust or
uses lifting-line theory to determine the optimum circulation torque.
distributions for the propeller-stator combination and u) de-
The
termine the appropriate asymmetric stator arrangement. Lifting-Surface Stage
second stage uses lifting-surface theory to determine the
which will produce the de-
propeller and stator geometry Once propeller and stator load distributions have been
sired circulation distributions. determined, lifting-surface methods are used to determine
the pitch and camber which will produce this loading. The
Lifting-Line Stage lifting-surface method used for the analysis of an asymmet-
ric stator has been described by Hsin [5]. It is an "analysis"
The lifting-line method described by Kerwin, et al [4] is method rather than a "design" method. In other words, it
used to determine the propeller and stator load distributions determines the loading for a given geometry, rather than
which satisfy the powering and cavitation design require- determining the geometry for a given loading. The "design"
ments. If applicable, several other quantities can be deter- geometry is one which develops the desired spanwise circu-
mined from the lifting-line model. For example, the diame- lation with each se,ction operating at ideal angle of attack.
is
ter, RPM, and the blade chord and thickness distributions An iterative method to determine the design geometry"de-
and practice, one can sacrifice the
can be determined, usually in combination with stress also presented in [3]. In
cavitation analysis tools. sign" load distribution to some degree in order to allow
some stator blades to have the same geometry. This may be
Axisyrnmetric Stator Model desirable to reduce manufacturing costs.

The stator is initially assumed to be axisymmetric; the

blades are uniformly distributed around the hub, and
they LARGE SPAN ASYMMETRIC STATOR
each have the same load distribution. The inflow to the sta-
uniform. Also, only
tor and propeller is circumferentially A large span asymmetric stator was designed for a par-
the circumferential mean induced velocity from one compo-
The wake ticular surface ship with an inclined shaft propeller. The
nent on the other component must be evaluated. primary objective of this design was to validate the design
from the stator is modeled in the usual way with constant method. A propeller/stator combination was designed to in-
the while
pitch helical vortices, where the pitch is deterrnined from crease efficiency relative to the reference propeller
local relative velocity at the line. providing considerable flow smoothing across the entire
de-
The objective of the axisymmetric design phase is to span of the propeller. A brief description of the design
circulation which produces the
termine the propeller/stator along with model test results is presented in this section.
desired thrust or torque.
Design Summary
Asymmetric Stator Model
Since design re,quirements were not provided, there was
The lifting-line model is also used to design the asym- considerable flexibility in the selection of a design point for
metric stator. In this case, each stator blade can have a dif- the new design. The reference propeller had been
designe,d
before, only self
ferent inflow velocity and load distribution. As for an advance coefficient, .1, , of 0.745, although
induced velocity from the pro-
the circumferential mean
required. The induced velocity from propulsion occurred at 0.801. Lifting-line calculations pre-
peller on the stator is opti-
num- dicted an efficiency of 0.644 at the design J, and an
the stator on the rotor, however, is now computed at a since the
ber of radial and circumferential locations in the propeller mum efficiency of 0.676 at J5=0.960. However,
the stator operated at J 5=0.801, the effi-
plane. To simplify this computation, the wake of reference propeller actually
semi-infinite line vortices with a ciency for the operating condition is 3.2% higher than that
blades is modeled using
The resultinz nonuniform velocity in of the design point. Thisworks out to be an efficiency of
pitch of 90 degrees.

176
0.665 which will be considered to be the reference effi-
ciency. The optimum RPM for a propeller/stator combina- 0.498R
tion occurred at a significantly lower RPM (J5=1.07),
where the predicted efficiency is 0.712. This is a 5.3% in-
crease over an optimum single propeller and a 7.1% increase
over the reference efficiency.
Since the reference propeller was designed for a fairly
high RPM (therefore low torque), the rotational kinetic en-
ergy lost in the slipstream is minimal. Thus, a preswirl sta-
tor can not improve efficiency significantly at the given
propeller RPM. The lifting-line calculations showe,d only a -0.1
1.5% increase in efficiency at the reference propeller design ,('' Original Wake (Measured)
RPM. To demonstrate a measurable effect on powering, the -0.2
new design point was selected at the lower RPM (.15=1.07),
where a significant increase in efficiency could be obtained_ -0.3
The stator diameter was selected to be 20% greater than o 60 120 180 240 300 360
the propeller diameter. This was done in order that signifi- Angular Position, 14 (deg)
cant flow smoothing could be obtained over the entire span
of the propeller and to prevent the stator tip vortex from im- Figure 1. Tangential Velocity in Propeller Plane, r/R=0.50
pinging on the rotor. The circulation distribution on the
stator was initially selected to be the optimum (based on ef- 0.650R
0.3
ficiency) and was later modified somewhat to improve flow Calculated
smoothing. Seven stator blades distributed with 30 degre,e
0.2
spacing from 0 to 180 degrees (the port side of the shaft)
provided the desired flow smoothing with reasonable lift
coefficients. 0.1 Measured
The propeller chord and circulation distributions were V
selected to be the same as for the reference propeller. The 0.0
V,
circulation was scaled in order to obtain the same total
thrust The thrust loading coefficient, for the reference . ------ r-
propeller is 0.625. Original Wake (Measured)

Model Test Results

Model tests were performed for the baseline propeller 60 120 180 240 300 360
and the new stator/propeller design. The powering tests Angular Position, O (deg)
were performed in the towing basin at DTMB, and the wa.ke
survey tests were performed in the 48-inch variable pressure Figure 2. Tangential Velocity in Propeller Plane, r/R=0.65
water tunnel at Applied Research Laboratory (ARL) at Penn 0.813R
State University. 0.3
The powering tests showed that the efficiency for the
stator/propeller combination increased 6.6% percent over 0.2
the reference propeller at the full power condition. This is
in good agreement with the predicted 7.1% increase. 0.1
Wake survey tests were performed with the propeller
shifted downstream and driven from a downstream dy- V
0.0
namometer. This allowed wake measurements to be taken
in the original propeller plane, with the propeller operating.
The tangential velocity distributions obtained at three radii
are shown in Figures 1 to 3. The asymmetric stator does in- Original Walce (Measured)
deed reduce the once per revolution variation in tangential
velocity as predicted.

60 120 180 240 300 360

Angular Position, O (deg)

Figure 3. Tangential Velocity in Propeller Plane, r/R=0.81

177
 SMALL SPAN ASYMMETRIC STATOR Wake Estimate

The second applic,ation of an asymmetric preswirl stator Wake survey data were not available for this boat and it
is the use of small span stator blades designed to operate was necessary to estimate the wake based on a previous
with existing propellers. The purpose of the stator, in this wake survey for a similar hull form. The Coast Guard boat
case, is to eliminate a cavitation erosion problem which oc- is 110 feet (33.54 m) long and has a maximum speed of 30
curred near the root of the propellers on U.S. Coast Guard knots. It has an open stem with twin shafts each supported
patrol boats. After a relatively short period of time of oper- by a single vertical strut. The shafts are inclined downward
ation at the full power condition, deep pits appear near the approximately 11 degrees relative to the hull buttock lines.
root on the suction surface of each blade. The erosion is be- The U.S. Navy boat R/V Athena was found to have similar
lieved to be caused by leading edge suction side cavitation features. Athena is 160 feet (48.78 m) long and has a
due to the high angle of attack associated with an inclined maximum speed of 30 knots. Athena also has an open stem,
shaft. This form of cavitation can be reduced by the installa- twin shafts and approximately 11 degrees shaft inclination.
tion of an asymmetric stator. Since wake survey data were available for Athena, the pre-
The following sections describe the steps taken to de- dicted walce for the Coast Guard boat was based primarily
sign the asymmetric stator, beginning with prediction of on the model wake survey for Athena [7]. The predicted
cavitation for the existing propeller. The design of the stator nominal wake is shown in Figure 5.
is then described, followed by the prediction of powering
and cavitation for the modified configuration. Some of the 03 1.2
cavitation erosion test results are also provided. A more de-
tailed account of the design is given in [6]. 0.2 1.1

Design Constraints 1.0

Design constraints imposed by the USCG played an 0.9 K.

important role in the design of the stator. These constraints,
which were imposed to minimize the cost and complication 0.8
of construction, are listed below.
Existing propeller cannot be modifie,d. 0.7

Stator is limited to the region of the existing 0.6

rope g-uard, a ring located on the hub directly
60 120 180 240 300 360
forward of the propeller. The length of the
rope guard is only 3% inches (0.079 m). Angular Position, 6 (deg)

Each stator blade must have identical ge- Figure 5. Nominal Walce Estimate
ometry (however, each blade may be rotate,d
to obtain the appropriate pitch setting). Existing Propeller Analysis

The first step in solving the erosion problem was to re-

produce the conditions for the existing propeller using pro-
peller analysis tools. The cavitation at the full power condi-
tion was predicted using two methods as described in the
following paragraphs.
The first procedure used for pre,diction of cavitation was
the traditional "cavitation bucket" method described by
Brockett [8]. This procedure calculates the minimum sur-
face pressure for a 2D airfoil at various steady angles of at-
tack. The result is a cavitation bucket which defines regions
for various types of cavitation. The cavitation bucket is
fixed for a fixed section geometry. The propeller operating
condition is represented by the cavitation index, a. The
c,avitation index is plotted as a function of angle of attack
for a propeller section as it rotates through the unsteady in-
flow. Points which fall outside of the bucket indicate cavita-
tion over that portion of the propeller rotation. An angle of
attack reduction factor is applied to account for steady 3D
Figure 4. U.S. Coast Guard Patrol Boat effects. However, since the flow is assumed quasi-steady,

178
 the unsteady effect on pressure and angle of attack is ne-
glecte,d.
Cavitation buckets for the USCG boat were calculated
for the 30 knot, full power condition. The result for the
0.3R radius, shown in Figure 6, indicates that when the
blade is in the 2700 position, the propeller experiences
leading edge suction side cavitation. At the 90° position, in-
ception of leading edge pressure side cavitation is evident.
There is no indication of back bubble cavitation at 30 knots.
The second method for prediction of cavitation was
based on a panel method developed by Lee [9]. Using this
method, the propeller blade pressure distribution was de-
termined for steady flow conditions. Hub boundary and
steady 3D effects are included. The inflow must be consid-
ered quasi-steady in order to examine the circumferential
variation in angle of attack. Results are shown in Figures 7 0.2 0.3 0.4 0.5 0.6 0.7
and 8 for a single radius near the hub (0.325R) with the -CP m. and O-
blade at 900 and 270°. In this case, the pressure coefficient
and the cavitation index are normalized by the ship speed. Figure 6. Cavitation Bucket With Original Inflow
At 30 knots, the value of o-, is 0.925. This figure indicates
leading edge suction side cavitation at 30 knots when the 2.0
blade is at 270° and leading edge pressure side cavitation at
90'. Thus both cavitation prediction methods indicate simi- Original Condition
lar forms of cavitation at the full power condition. 1.5 Radius = 0.325R
Pressure Side Blade at 90 degrees
Since back bubble cavitation was not predicted in either
case, it was concluded that leading edge suction side cavita- 1.0
0-5=0.925
tion was the cause of the erosion. Although this cavitation
was predicted to be much more extensive at larger radii, the
erosion occurred only near the hub. The most promising
-C0.5
way to eliminate the erosion was to eliminate the cavitation. 0.0 Suction Side
Since leading edge cavitation was caused by excessive angle
of attack, a preswirl stator was designed to reduce the cir- -0.5
cumferential variation in tangential velocity.
-1.0
Asymmetric Stator Design
0.0 0.2 0.4 0.6 0.8 1.0
Chordwise Position, xdc
This section presents the rationale used in the design of
the asymmetric preswirl stator. The basic configuration is Figure 7. Chordwise Pressure Distribution, Blade at 90°
first explained, which includes the sele,ction of diameter,
number of blades, and location of blades. The spanwise dis- 2.0
tribution of loading is then explained and its effect on the
Original Condition
downstream velocity distribution. Finally, selection of the 1.5 Radius = 0.325R
blade geometry, which includes spanwise distribution of Blade at 270 degrees
chord, thickness, carnber, and pitch, is explained.
1.0
Basic Configuration

The stator diameter was selected to be large enough

-C0.5
such that the stator could adequately affect the inflow to the 0.0
propeller in the region of the erosion. The erosion occurred
on the propeller near the 0.3R radius, which was approxi- -0.5
mately 2.5 inches away from the hub. Thus, a 5.0 inch span
was selected which, given a 10 inch hub diameter, lecl to a -1.0
20 inch stator diameter. This is approximately 40% of the 0.0 0.2 0.4 0.6 0.8 1.0
49.6 inch propeller diameter. A larger diameter would have Chordwise Position, xdc
led to an unnecessary amount of drag and a higher stress
level at the root of the blades. Figure 8. Chordwise Pressure Distribution, Blade at 270°

179
 The selection of blade number and blade locations was Load Distribution
based on a variety of reasons. Since propeller leading edge
suction side cavitation was a problem in the 2700 region, The shape of the spanwise circulation distribution, G,
stator blades were necessary in this region to eliminate the greatly affected the stator induced velocity in the plane of
existing erosion problem. These blades needed to turn the the propeller. The magnitude of the loading was selected in
flow downward which required that they be lifting upward. conjunction with the blade ge,ometry and is discussed later.
This configuration would have been adequate. However, A finite hub loading with a significant extent of zero slope
upwash from these blades occurred in the 90° region which near the hub led to the most desirable distribution of veloc-
aggravated propeller leading e,dge pressure side cavitation. ity downstream. This distribution is shown in Figure 10.
This form of cavitation is known to be more violent than Since there was no slope near the hub, there was no shed
leading edge suction side cavitation and erosion on the pres- vorticity in this region. The load near the tip was reduced in
sure side may occur. To prevent leading edge pressure side order to prevent tip vortex cavitation. The modified wake at
cavitation, stator blades were also required in the 900 re- the 0.759R5 (0.306R), shown in Figure 11, represents the
gion. Since these blades must aLso lift upward, the camber modified walce at the radius where the erosion occurred.
must be in the reverse direction of the blades at 270°. Thus This modified wake provides a 40% reduction of the tan-
at least two blade geometries were required, which violate,d gential velocity at 90° and 270°.
the design requirement of only one blade geometry. It was
quickly realized, however, that a mirror image geometry
0.020
would be necessary for the port stator and there were actu-
ally two blade geometries to work with. Thus the blades in
the 90° region were designed to be a mirror image of those
in the 270° region. In order to eliminate possible cavitation 0.015
due to the strut viscous effect, a blade was also positioned at
00. The selected load distribution, which is discussed later,
had a finite value at the hub. Since this can lead to a strong G 0.010
hub vortex and a drag penalty, a blade was also positioned at
1800 to cancel the swirl of the blade at 00.
As stated in the design constraints, the stator was lim- 0.005
ited to the region of the existing rope guard. This rope
guard was only 3X inches long. This restriction limited the
chord length of the stator blades which in turn limited the 0.000
amount of load each blade could carry. In order to produce 0.5 0.6 0.7 0.8 0.9 1.0
enough swirl, the number of blades was maximized. For
ease of construction, the minimum blade spacing was se-
lected to be 30°. For these reasons, four blades were posi- Figure 10. Stator Design Load Distribution
tioned around 270° and 90°. This basic configuration is
shown in Figure 9. 0.3
r = 0.759R5 = 0.306R Original Vel6city

0.2

0.1
Vc
0.0
285°
75°
-0.1

-0.2
255'
105' -0.3
60 120 180 240 300 360
Angular Position, e (deg)

Figure 11. Modified Tangential Wake in the Propeller Plane

Figure 9. Stator Configuration

180
 Blade Georrzetry
0.25 0.05
The chord length was maximized in order to achieve the
greatest amount of loading. Given a hub length of 3X
inches, the root chord of the blade was selecte,d to be 2.9 0.20 0.04
inches. In order to provide a margin against stall, the lift
coefficient was selected to be no more than 0.5. Since the 0.15 0.03
cID,
load distribution was constant ne,ar the hub and the relative
velocity was nearly constant, the chord was held constant D,
near the hub in order to maintain a CL equal to 0.5. The 0.10 0.02
chord was rounded down to 1.5 inches at the tip where the
load distribution goes to zero. A finite chord was main- 0.05
tained at the tip in order to help prevent tip vortex cavita-
tion. The distribution of chord and thicicness are provided in
Figure 12. 0.00
The thicicness, t, was selected to be large enough to 0.5 0.6 0.7 0.8 0.9
provide strength while not inducing back bubble cavitation. r/Rs
The following approximate formula based on the cavitation
buckets for a MACA 16 section vias used ta check the mar- Figure 12. Stator Chord and Thickness
gin against back bubble cavitation:
= 2.28(t / c)+ 0.56C, 0.06

Standard practice for maximum allowable stress for a pro- 0.05

f lc (with lifting surface correction)
peller is 12,500 psi (for nickel-aluminum-bronze). The
maximum stress at the hub of the stator was only 2,800 psi, 0.04
well below this limit.
The camberj, and pitch, 0, were determined using lift- 0.03
ing surface design methods. Since a lifting surface code f Ic (2D)
was not available for asymmetric stator design, blades in the 0.02
270° region were simulated by computing the lifting surface
geometry for an axisymmetric 12 bladed stator. For this
0.01
computation, the lifting surface design method by Wang
[10] with extensions for hub boundary effects was used.
The results are shown in Figure 13 for camber. There was 0.00
essentially no lifting surface correction for pitch. 0.5 0.6 0.7 0.8 0.9 1.0
From the lifting line analysis, the hydrodynamic pitch rIRs
for each blade was determined. However, the design re-
quirements stated that only a single blade geometry could be Figure 13 Stator Camber Distribution
designed. Since the load distribution for each blade was the
same, the distribution of pitch for each blade was similar. 20
Thus it was expected that if the average pitch distribution
15
for all the blades was used, the original design load distribu-
tion could be nearly sustaine,d. The average design pitch 10
distribution is shown in Figure 14. Using the lifting surface
analysis method for asymmetric stators described by Hsin 5
Pitch
[5], each blade was then rotate,c1 in an attempt to obtain the Angle 0
design loading. This required a simultaneous iteration of the (deg)
pitch of each blade until the original design loading was ob- -5
tained.
-10
ResultS 15

-20
The effect of the stator on powering and cavitation
performance is now presented. From the lifting line analy- 0.5 0.6 0.7 0.8 0.9 1.0
sis, the net forces from the stator were computed. The lon- rIR,
gitudinal, vertical, and lateral forces are listed in Table 1.
The longitudinal drag of 11 lbs. per shaft is negligible com- Figure 14. Stator Pitch Distribution

181
pared to the 20,000+ lbs. of thrust each propeller produces.
Thus no significant effect on powering was expected. 2
Radius= 0.3R eading Edge
Suction Side
Table 1. Net Stator Forces (per shaft) Cavitation

Back
Longitudinal: -11 lbs. (drag) Bubble
avitatio
Estimated Strut
Vertical: 746 lbs. 900 Viscous Effect
Modified
224 lbs. Operating Points (0-)
Lateral:

The modified flow behind the stator was used to de- Bucket (-Cp..) Leading Edge
termine the impact on cavitation performance of the pro- Pressure Side
Cavitation
peller. The two cavitation prediction methods were repeated
2
for the new condition. The cavitation bucket, shown in 0.4 0.5 0.6 0.7
Figure 15, indicates that leading edge suction side cavitation 0.2 0.3
is near inception at 270°. While the panel method, shown
-C °-

in Figures 16 and 17, shows a reduction of the suction peak,

Figure 15. Cavitation Bucket with Modified Inflow
cavitation is still evident. At 90° the cavitatinn bucket
shows a large margin against pressure side cavitation while
the panel method shows cavitation inc,eption. The cavitation 2.0
inception speed was determined for both methods for vari- Modified Condition
ous types of cavitation at the 0.3R radius and is shown in 1.5 Radius = 0.325R
Table 2. The panel method pre,dicts lower inception speeds Blade at 90 degrees
Pressure Side
in all cases. The principal result is the increase in le,ading
edge suction side cavitation inception speed at the 0.3R ra-
1.0 - 0.925-1
dius of the propeller of approximately 3.5 to 5.5 knots. 0.5
Cps
Table 2. Predicted Cavitation Inception Spe,eds ;- Original -
0.0
Original - Modified Delta MOdified
Prediction Cavitation
0.5
Method Type Suction Side
Leading edge
24.6 30.0 5.4 1.0
suction side: 0.8 1.0
0.0 0.2 0.4 0.6
Cavitation Leading edge
30.0 38.1 8.1 Chordwise Position, xdc
buckets pressure side:
Figure 16. Modified Pressure Distribution, Blade at 90°
Back Bubble: 37.5 36.8 -0.7
Leading e,dge
Panel suction side: 22.0 25.6 3.6 2.0
code Leading edge Modified Condition
Suction Side
pressure side: 24.0 28.0 4.0 1.5 Radius = 0.325R -
Blade at 270 degrees
Model tests were conducted in the DTMB 36-inch vari- 1.0
able pressure water tunnel to determine cavitation erosion
characteristics of the propeller alone and with the stator 0.5
[11]. The erosion tendency was determined using an erod- Cps
able coating applied to the propeller blades. Results for the 0.0 Original
propeller alone, Figure 18, show severe erosion spots on the
Modified
suction and pressure side near the root. The erosion was so -0.5
severe that it wore through the coating and the anodized Pressure Side
finish, down to the bare metal. With the addition of a stator, -1.0
the cavitation erosion tendency was greatly diminished. 0.4 0.6 0.8 1.0
0.0 0.2
Some blades had no erosion spots, but a few had a new, less
Chordwise Position, xdc
severe spot located closer to the trailing edge, as shown in
Figure 19. Figure 17. Modified Pressure Distribution, Blade at 270°

182
Figure 18. Simulated Cavitation Erosion, Propeller Alone Figure 19. Simulated Cavitation Erosion, With Stator

Figure 20. Cavitation of Propeller Alone Figure 21. Cavitation of Propeller With Stator

183
 For the propeller alone, the cavitation covered the entire [4] LE. Kerwin, W.B. Coney, and C-Y. Hsin, "Optimum
blade, as shown in Figure 20. With the addition of a stator, Circulation Distributions for Single and Multi-
the cavitation was greatly reduced in the root region, as Component Propulsors," Twenty-First American
shown in Figure 21. A small amount of cavitation is still ev- Towing Tank Conference, National Academy Press,
ident. 1986.
C-Y. Hsin, "Analysis of the Performance of a Non-
Axisymmetric Stator by Lifting Surface Theory,"
CONCLUSIONS Massachusetts Institute of Technology, Department of
Ocean Engineering, 1987.
A brief description of the design method and two dis-
S.K. Neely, and B.Y-H. Chen, "Asymmetric Preswirl
tinct applications of an asymmetric stator upstream of an
Stator Design for U.S. Coast Guard Island Class Patrol
inclined shaft propeller have been described. The first ap-
plication shows that with a large span asymmetric stator Boats," David Taylor Research Center Report SHD-
significant flow smoothing is achieved across the entire span 1335-02, March 1991.
of the propeller. A 6.6% increase in efficiency is also R.B. Hurwitz and L.B. Crook, "Analysis of Wake
achieved, which ag-rees well with the predicted increase of Survey Experimental Data for Model 5365
7.1%. Note that an efficiency increase of only 3.0% is pre- Representing the R/V Athena in the D'INSRDC
dicted if the reference propeller had been designed for the Towing Tank," David Taylor Research Center, Ship
optimum RPM. It is expected that this type of asymmetric Performance Department Report DTNSRDC/SPD-
stator can significantly improve the cavitation performance 0833-04, October 1980.
of inclined shaft propellers while also improving the effi-
ciency. This design served to validate the design method. T. Brockett, "Minimum Pressure Envelopes for
The second application shows that the root erosion Modified NACA-66 Se,ctions with NACA a=0.8
Camber and BUSHIPS Type I and II Sections," David
problem of the U.S Coast Guard patrol boat propellers might
be eliminated with use of a small span asymmetric stator. Taylor Research Center Report 1780, February 1966.
Sufficient flow smoothing was achieve,d in the root area to J-T. Lee, "A Potential Based Panel Method for the
reduce the tendency for cavitation erosion Analysis of Marine Propellers in Steady Flow,"
Massachusetts Institute of Technology, Department of
Ocean Engineering Report 87-13, 1987.
ACKNOWLEDGMENTS M.H. Wang, "Hub Effects in Propeller Design and
Analysis," Massachusetts Institute of Teclmology,
Funding for the design of the large span asymmetric Department of Ocean Engineering Report 85-14, 1985.
stator was provide,d by the Office of Naval Technology, un-
der the Surface Ship Technology Program, Program T.B. Smith and K.D. Remmers, "Propeller Erosion
Element 62621N. Powering test results were provided by Re,duction with an Asymmetric Preswirl Stator," To be
Richard Bell of DTMB, and wake survey test results were presented at the Twenty-Third American Towing Tank
provided by Fred Smith and John Eaton of ARL. Conference, 1992.
The design of the asymmetric stator for the large patrol
boat was funded the USCG Research and Development
Center, Groton, CT, through a project directed by Edward
Purcell. The model test results were provided by Kenneth
Remmers, Douglas Jenlcins, and Tim Smith of DTMB.

REFERENCES
W.S., Gearhart and R.C. Marboe, "Asymmetric
Reaction Vane Application," Proceedings of the
Symposium on Hydrodynamic Performance
Enhancement for Marine Applications, 1988.
G. Larimer, W. Gearhart, D. Van Liew, and C. Shields,
''Reaction Fin Applications in Marine Propulsion,"
Presented to SNAME Hampton Roads Section, 1987.
J.E. Kerwin, "Hydrodynamic Aspects of
Propeller/Stator Design," Propellers '88 Symposium,
SNAME, 1988.

184
 Improvement of Surface Ship Propeller Cavitation
Performance Using Advanced Blade Sections
J.W. Bailar, S.D. Jessup, Y.T. Shen
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT Water density

cY Cavitation number
A propeller design procedure
incorporating new propeller blade INTRODUCT ION
sections has been developed to improve
propeller cavitation inception. Two- At high speeds, propeller
dimensional foil design procedures of cavitation can be responsible for noise,
Shen and Eppler provide geometry for a vibration, and thrust breakdown.
specified blade pressure distribution. Therefore, the propeller designer's
Advanced propeller lifting surface primary goal is to design propellers
design procedures then generate which meet powering requirements while
propeller geometry. Panel method operating cavitation-free at as high a
analysis verifies final design changes. speed as possible, and with a minimum
Resulting propeller geometries produce a amount of cavitation at higher speeds.
specified blade pressure distribution General propeller parameters, such as
with improved blade surface cavitation blade chordlength, maximum thickness,
inception. maximum camber and skew have
A series of new blade section traditionally been the main variables
model propellers has been designed and used to control cavitation and
tested for cavitation and powering efficiency in the final design process.
performance. The new propellers show a The shapes of the blade thickness and
2-3 knot improvement in blade surface camber sections have been selected from
cavitation inception speed. The new standard families of shapes developed in
section propellers also show some the 1940's for aeronautical
Reynolds number effects on powering, applications. Only a few of these
requiring higher speed testing than standard section shapes have proved to
conventional blade section propellers. be applicable to marine propellers.
These sections have relatively uniform
NOMENCLATURE chordwise pressure distributions to
CD Drag coefficient avoid low pressure peaks which lead to
CL Lift coefficient cavitation. Thus, control of the blade
Cp Pressure coefficient pressure distribution is limited to
Chord length variation of the maximum section
Propeller diameter thickness, maximum section camber and
Camber angle of attack.
JA Advance coefficient, (1-wT)Vs/(nD) Recently, advances have been made
KT Thrust coefficient, T/(p n2 D4) in the design of two-dimensional (2-D)
Revolutions per second foil shapes with a specified pressure
Propeller radius distribution. These design techniques
Rn Reynolds number have been applied to marine hydrofoils
Local propeller radius by Shen and Eppler(1). In these studies
Thickness 2-D hydrofoil sections were developed
that showed marked improvements over
Distance along chord
Vs Ship speed conventional geometries. With the
Taylor wake fraction emergence of these 2-D techniques, there
wT has been obvious interest in their
a Angle of attack, degrees application to marine propellers. Some

185
propeller designers have merely inserted The new 2-D section design is
the 2-D camber and thickness performed using the Eppler-Shen
distributions of the Eppler-Shen procedure. For each radius, a new
sections into propeller geometry, section is designed to delay cavitation
similar to what is done traditionally over as much of the defined operating
with the MACA section shapes range as possible. The new section
(corrections only being made to the design procedure is iterative in nature
magnitude of thickness and camber due to in which regions of constant velocity
propeller lifting surface effects). (or pressure) on the foil are specified
In this paper an alternative for specified angles of attack. Also
approach is used which accounts for the trailing edge pressure recovery and
propeller lifting surface effects more closure is specified. The minimum
precisely. This method will be presented pressure envelope, referred to as the
along with a description of a series of cavitation bucket is then calculated
new blade section propeller designs and using potential flow theory. Boundary
layer characteristics are also
model evaluations.
calculated. The final foil section is
PROCEDURE TWO-DIMENSIONAL iteratively designed to avoid boundary
DESIGN -
layer separation at full scale Reynolds
SECTION DESIGN
number and to produce the most favorable
The new blade section design cavitation bucket. For this case, where
procedure includes the design of two- the lift coefficient increases with ship
dimensional foil sections and their speed, a bucket favoring lower a and
into final propeller higher CL is desired.
integration typical
geometry using lifting surface design Figure 1 compares
procedures. conventional and new section cavitation
The main goal in the 2-D design buckets. Cavitation occurs outside the
phase is to generate propeller sections enclosed area of the bucket. The
which have desirable cavitation buckets conventional section was a NACA 66(TMB
such as those shown in Figure 1. A MOD.) with a thickness to chord ratio,
lifting line design must be performed to t/c=0.0662, and an a=0.8 meanline, with
determine initial geometric parameters. a camber to chord ratio, f/c=0.0141. The
From this design, the propeller resulting new section maximum thickness
operating conditions are determined as a and camber ratios were t/c=0.05 and
function of local position in the inflow f/c=0.01198. The new foil thickness was
significantly reduced over the
wake. These conditions are represented
as blade section lift coefficient (CL), conventional section and yet resulted in
or angle of attack (a) and cavitation an enlarged bucket. The upper left
"corner" of the bucket has been expanded
number (6). The cavitation number is
to improve inception while the less
based on the ship speed and the important lower left "corner" has been
submergence depth of the blade section
cropped. Figure 2 shows a comparison of
at the appropriate angular position. The thickness and camber distributions. The
variation in inflow angle is obtained
from wake survey measurements. The mean
new thickness is increased toward the
leading edge, giving leading edge
lift coefficient is obtained from a performance more like that of a thick
lifting line calculation. Lift at section. The camber loading is reduced
varying angles of attack is determined
line and wake test at the leading edge and shifted aft.
from lifting Figure 3 shows the conventional and new
results(2,3). Once the operating points
section pressure distributions at the
are determined, the radii for blade design ship speed, each with equivalent
section designs are selected. Selecting lift coefficients. The standard section
many radii will give a design which is
well tailored to the inflow wake at the
operates at ideal angle of attack, so-
called shock free entry, while the new
expense of increased complexity and section operates at a negative angle
difficult geometry fairing. Selecting producing negative leading edge loading.
only a few radii will give a simpler As speed is increased, the section angle
design which will not have the fully
optimum cavitation performance.
of attack increases which produces an

186
 earlier leading edge suction peak than interpolate and extrapolate the 2-D
on the conventional section. Optimally, section thickness distributions across
the new section will attain a flat the span of the blade. The computed
pressure distribution at the highest camber surface is shown in Figure 6.
speed possible with a pressure just This surface shows some unfairness, and
above vapor pressure to avoid very large camber at the tip. Later
cavitation. To maintain the specified designs have incorporated foil sections
lift, the new section relies on loading at the hub and near the tip to better
the aft end of the section. This large control the hub and tip geometry.
aft loading is produced by high trailing
edge camber, which can produce a
distinctive trailing edge hook or TNCORPORATTON OF PANEL CODP, INTO
cupping. This increases the potential PROPFUJER DESIGN
for flow separation and decreased Typical new blade section
efficiency. The new section is designed propellers generated by the lifting
to separate on the suction side near 98% surface code can show unacceptable
chord without significant performance amounts of unfairness and extreme
degradation at full-scale Reynolds geometry at the ends. Varying degrees of
number. fairing have been used to smooth the
blade surface and to modify the loading
INCORPORATION OF NEW SECTIONS INTO at the root and tip. Any geometry
PROPELLER DESIGN changes can reduce the original foil
Final design calculations are section performance; therefore,
performed with a lifting surface design modifications were checked by
code(4). With specified blade parameters calculating the blade surface pressure
such as circulation, chord length, distribution using a surface panel
thickness, skew, and rake, the program method(5). To demonstrate the general
computes the blade section pitch and capability of the design process, Figure
camber surface to produce the required 7 compares the original 2-D pressure
radial and chordwise blade loading. distributions against the calculated
Modifications were made in the code to distributions using the panel code.
permit the specification of a completely Through the mid-span region the
arbitrary three dimensionally varying comparisons are very good, but, at the
blade loading distribution. hub and tip, the lifting surface design
From the pressure distributions method does not adequately represent
selecteded for each propeller radius, flow relative to the paneled blade and
typically shown in Figure 3, the hub surface results.
differences in upper and lower surface
Cr
p AC P' were calculated and are shown DESCRIPTION OF NEW BLADE SECTION DESIGN
in Figure 4. For this particular design, AND EVALUATION EFFORT
foil sections were designed for 4 radii,
0.5, 0.6, 0.7, and 0.9. These An initial new section design was
distributions were used to specify the performed for an existing controllable
chordwise and radial ACp loading pitch propeller (DTMB Propeller 4990)
distribution for input into the lifting designed for a shaft-and-strut
surface code. Interpolations and arrangement with NACA 66, a=0.8 meanline
extrapolations were performed to provide section shapes. This was intended as a
input across the span of the blade. direct replacement of the blade sections
Figure 5 shows the interpolated and for the existing propeller with no
extrapolated ACp loading surface used as changes in overall geometric parameters
input. The loading at the hub and tip or loading distribution except for blade
were assumed to be zero, thus driving thickness. This design, designated as
the ACp loading surface extrapolation DTMB Propeller 5072, was an ideal
near the ends. candidate for new sections because of
The 2-D thickness was used the large increase in blade loading
directly for final propeller geometry. (section CL) with increasing ship speed.
The radial distribution of maximum This design improved blade surface
sect).on thickness was used to cavitation, but required refinements for

187
improved root cavitation and powering with the panel code at the blade angular
position of maximum and minimum loading
performance.
Later, four parametric propellers due to the shaft inclination. Figure 11
(DTMB Propellers 5131, 5132, 5133, and shows the suction side distribution for
5143) were designed and model tested to Propellers 5072, 5133 and the original
refine the new section propeller design. propeller 4990. The leading-edge suction
Variations were made in the root and tip peak can be seen on the original
areas to improve cavitation, and in the propeller causing earlier inception.
chordwise loading to improve powering. At high speeds there is
The chord length on the latter three substantial reduction in cavitation
extent with the new section designs, as
propellers was increased to improve
cavitation performance. Figure 8 shows shown in Figure 12. The character of the
the distribution of thickness and camber cavitation on the standard and new
for Propeller 5132. Figure 9 compares section propellers is quite different.
the blade outline of the Propeller 5132 Due to the strong suction peak,
cavitation initiates at the leading edge
and the original Propeller 4990. The The new
increased aft loading of the new on the original propeller.
propellers required a forward shift in section designs operate with a

the blade skew to balance blade spindle relatively flat suction side pressure
distribution, so the cavitation starts
torque.
Cavitation tests were conducted in as streaks occurring somewhat randomly
the DTMB 36-inch water tunnel. The test over the blade surface. The number of
shown in Figure 10, streaks increases as speed is increased,
configuration, while the standard propeller produces an
included shaft inclination, shaft, shaft
barrel, and struts. This resulted in a increasingly large cavitation sheet
relatively accurate representation of starting from the leading edge. Of the
new section designs, Propeller 5133 has
the straight ahead full scale wake. followed by
Model propeller diameter was 15.86 the least cavitation
Propeller 5132, then Propeller 5072.
inches (402.9 mm) run at full scale this agrees with the
speeds. This resulted in test Reynolds Qualitatively,
numbers up to 5 million, based on 0.7 panel code calculations.
radius inflow. Cavitation inception was
measured and cavitation extent and pat- ROOT CAVITATION PERFORMANCE
terns were observed through the speed is also
The types of blade surface Root cavitation
range.
cavitation observed were suction-side significant for all the propellers
tested. The inclined flow, along with
leading edge, suction-side streak, and the large hubs and shaft barrel shapes
suction- and pressure-side root
cavitation. Pressure-side blade surface of CRP propellers, produce large angle
cavitation was not observed throughout of attack variations at the root which
results in suction- and pressure-side
the operating range. Test conditions
were set by thrust identity, matching root cavitation and contributes to
thrust breakdown at high speeds.
original propeller propulsion test Propeller 5072 was designed with 2-D
thrust with measured propeller thrust in sections from 0.5 radius to 0.9 radius.
the tunnel. Powering performance was
assessed with open water testing up to a The extrapolation of specified section
Reynolds number of 2.4 million. geometry to the root resulted in blunt
sections which produced premature
suction-side root cavitation and
BLADE SURFACE CAVITATION PERFORMANCE increased thrust breakdown. To improve
Blade surface cavitation was root performance, 2-D sections were
designed at 0.3, 0.35, and 0.4 radii for
observed in the water tunnel initiating
the latter propellers. Table 2 shows the
on all propellers at the suction side
leading edge at 0.5 radius behind one of measured root cavitation inception
the struts. Table 1 shows approximately
speeds. Improvements were made in the
suction side inception speeds for Pro-
a 2 knot improvement with the new pellers 5132 and 5133 over the original
section propellers. The quasi-steady 4990, but there was a
pressure distribution was calculated Propeller

188
reduction in pressure side root speed open water dynamometer/drives at
cavitation inception speed. These DTMB. At low Reynolds number, both
differences could be attributed to propellers suffer efficiency loss
global parameters of root pitch and presumably due to the inner radii
maximum camber. It appears that the separation. A sharp increase in
extrapolation of two-dimensional efficiency occurs for both propellers up
analysis to the complex three- to Rn = 1 x 106. The transitional
dimensional flow conditions at the root increase in efficiency occurs at a
have only limited benefit. slightly higher Reynolds number for the
new section Propeller 5133. Propeller
POWERING ISSUES AND EFFICIENCY 4990 reaches a maximum efficiency and
then gradually decreases, while
An important trade-off in the new Propeller 5133 continues to gradually
blade section propeller design is increase its efficiency. Both propellers
cavitation improvement balanced against attain the same efficiency at the
the risk of efficiency loss due to flow highest Reynolds number tested. It
separation. As seen in the predicted appears that the inner radius separation
blade section pressure distributions, subsides on both propellers at around Rn
the use of aft section loading results
in extreme adverse pressure gradients at = 1 x 106. The trailing edge separation
the trailing edge with potential for on Propeller 5133 may be gradually
flow separation. All foil sections were subsiding with increasing Reynolds
designed to avoid separation at full- number.
scale Reynolds numbers where the blade Fig. 15 shows the open water
section boundary layer is fully turbu- thrust coefficient, KT for Propellers
lent. Unfortunately small- scale open- 5132 and 5133 operating at their design
water tests using standard dynamometry advance coefficient, J = 1.27. The
are often limited to a blade Reynolds thrust coefficient steadily increases up
number of 7 x 105 for defining open to the Reynolds number of 1 x 106. At
water characteristics. At these higher Reynolds number the thrust of
conditions, laminar boundary layer flow both propellers is relatively constant
occurs which will easily separate at the and approaches the thrust predicted
start of pressure recovery near the using the panel code. The prediction
trailing edge. using the potential based panel code re-
To investigate these effects, open quires a viscous correction, typically
water tests were conducted over a range in the form of an applied section drag
of Reynolds numbers, along with blade coefficient. A value of CD = 0.007 was
surface oil flow visualization. Figure used, which is typical for standard
13 shows the surface flow lines on the sections. The target thrust coefficient
suction side of Propellers 4990 and 5133 from the preliminary lifting-line design
at a Reynolds number of 1 x 106. On was specified at 0.23, indicating some
Propeller 5133 at the 0.5 to 0.9 radius, inaccuracies for Propeller 5132.
there is some outward flow of the
surface streamlines at the trailing CONCLUSIONS
edge, which is indicative of flow
separation. The original Propeller 4990 A propeller design procedure
shows no tendency of separation at the incorporating new propeller blade
trailing edge. Also shown is laminar sections has been developed to increase
separation that occurred on both propeller cavitation inception speed.
propellers at lower Reynolds number, 5 X Two-dimensional foil design procedures
o 5. This inner radius separation are utilized along with advanced
appears to be common for controllable propeller lifting surface design
pitch propellers with relatively large procedures and panel method analysis.
hubs, thick root sections and strong a After completion of a series of pro-
circulation gradient at the inner radii. peller designs and model cavitation and
Figure 14 shows the measured open powering experiments, the following
water efficiency for the two propellers conclusions have been reached.
tested on the low speed and new high

189
 The design procedure is capable Table 1. Suction side leading edge cavitation inception.
of producing a specified blade pressure
distribution in the mid-span region of PROP NO. Measured CALCULATED
36" WT PSF10
the blade. Local root and tip geometries 4990 +2.2
cannot be sufficiently defined to match 5072 +2.2 +2.8
a specified pressure distribution. 5132 +1.7 +2.1
5133 +1.9 +5.4
The new blade section
geometries can improve blade surface
inception (excluding root cavitation) by
2-3 knots by direct blade section Table 2. Root cavitation inception.
replacement for typical open shaft,
transom-stern surface ships. Very PROP NO. SUCTION PRESSURE
in cavitation SIDE SIDE
significant reduction +0.4
4990
volume can be achieved over most of the 5072 -2.8 -1.6
cavitating speed range. 5132 +1.2 -1.0
5133 +0.1 -5.1
The potential-based surface
panel method for steady propeller flow
analysis has become a powerful tool for
iteratively evaluating propeller Fig. 1. Cavitation bucket and Operating points
cavitation performance through the
design process. 0,-

The efficiency of the aft-

loaded blade sections is variable with 0,- NEW SECTION

Reynolds number, but only slightly more

than traditional section geometries.
Powering evaluations at blade Reynolds 0.3 -

number greater than 1 x 106 are required C1

to avoid large Reynolds number effects. 0.2

REFERENCES
0.1 -

Y.T. and Eppler, R., "Wing

Shen, STANDARD SECTION

Sections for Hydrofoils - Part II:

Nonsymetrical Profiles," Journal of Ship 0

Research, Vol. 25, Sept. 1981, pp. 191- O , Cavitation numb.

200.
Rader, H.P., "Cavitation Phenomena in
Non-Uniform Flows," Appendix II of the Fig. 2. 2-D section.

Cavitation Committee Report,

0.08
Proceedings, 12th International Towing
Tank Conference, Rome, Italy, pp. 251-
364, September 1969.
Brockett, T., "Steady Two-Dimensional 0.06

Pressure Distributions on Arbitrary

Profiles," David Taylor Model Basin
Report 1821, October 1965. t/c0.04

Greeley, D.S. and J.E. Kerwin,

"Numerical Methods for Propeller Design
and Analysis in Steady Flow,"
0.02
Transactions SNAME, Vol. 90 (1982).
Kerwin, J.E., et. al., "A Surface
Panel Method for the Hydrodynamic
Analysis of Ducted Propellers," SNAME 0.6 0.8
0.2 0.4
Transactions Volume 95, 1987. x/c

190
 Fig. 2(continued). 2-D section Fig. 4. Chordwise A Cp loading

0.02

0.015

0.005

o 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8

x/c x/c

Fig. S. Prop 5022 A Cp loading surface for input to the lifting

surface code.

NEW SECTION, t/c-.0.05, E/C..0.0120

---------------

SACA 66, A-0.6, t/c..0.0662, f/c0.0141

Fig. 3. 2-D section pressure for Cl - 0.208

0.4 FA.

Fig. 6. Camber surface from the lifting surface code for Prop 507;

0.2

-oF

-0.4
0.2 0.4 0.6 0.8
x/c
Lea., Edge

191
 Fig. 8. Camber and thickness for Prop 5132
Fig. 7. Comparison of specified 2-D and calculated
panel code pressure distributions for Prop 5133 at
design condition.

f/c

t 0.5

x/c

' -0.5
t/c

0.2 0.4 0.6 0.8

x/c 03 0.4 0.5 0.6 0.7 0.8 0.9

.358 r/R

Fig. 9. Blade shape for P4990 and P5132.

.5

0.2 0.4 0.6 0.8

x/c
.70

Fig.10. Inclined shaft test configuration for DTAC 36" Water Tunne:

Artificial wake
deficit from shaft
0.5 angle joint
Inclined shaft
and struts

.5

Rydrophones for
0.6 0.e acoustic measurements
0.2 0.4
x/c
.9R

192
 Fig. 11. Quasi-steady maximum blade loading Fig. 13. Prop 5133 flow visualization at Rn - 1 x 106.
pressure.

-0p

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

x/c
Suction side pressure ..stribution at .50R.

Fig. 14. Measured open water efficiency at design

advance coefficient, Ja-1.27.
6
0.75 -
£4990 high .peed dynamometer

so.
- P5133 high sp., dynamometer
2 -
P5133 0.7
as.
-
/ P5133 lo. .peed dynamometer
Tle

P4990

0.65
:7!

-6 - r ;

0 3 0.4 0.5 0.6 0.6

0.7 0.8 0.9 1
rfR 5 10 15 20 25 30
Suction side Cpmin represented as inception .peed. Itoynolcla number, Re x 10-5

Fig. 12. Measured cavitation patterns in DTRC 36" Water Tunnel. Fig. 15. Measured open water thrust at design
advance coefficient, JA - 1.27.

0.3

0.25
P5133
P5133

P5132 P5132
Prop 4990, Standard Section Prop 5132, Now Section CT 0.2

0.15

0.1
0 10 15 20 25 30
Reynolds number, Re x 10-5
Prop 5072, Neo Section Prop 5133, New Section

193
 Propeller Parametric Study for
Mid-Term Fast Sealift Ships
K-H. Kim, J.F. Fraas
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT P Fluid density

F Circulation
A program was initiated in 1990 to assess new and
alternative technologies for application to mid-term fast
sealift ships that could be under contract in 1998 or later. A INTRODUCTION
variety of baseline hull configurations have been developed
in this program. This paper p..scats a parametric study of A technology development program was initiated in
powering performance of three baseline hull configurations. 1990 for mid-term fast sealift ships which would be
They are a 30-knot, twin-skeg ship, a 20-1cnot single-shaft contracted for 1998 or beyond. It is cturently being carried
ship, and a 35-Icnot triple-shaft ship. Comparisons of out jointly by the Naval Sea Systems Command
propulsive performance between open-single and (NAVSEA), the David Taylor Model Basin of the
contrarotating (CR) propellers are presented. The Carderock Division of the Naval Surface Warfare Center
predictions indicate that CR propellers will improve (CDNSWC), the Maritime Administration (MARAD), and
propulsive efficiency by about 7 to 9 percent for the three private contractors under the sponsorship of the Strategic
baselines. It is concluded that the 30-knot, twin-skeg ship is Sealift Division of the Chief of Naval Operations. The
the most promising candidate for the mid-term fast sealift major goal of the program is to identify and assess various
ship. A model test program is underway to confirm the technologies in the areas of hydrodynarnics, machinery
predicted powering performance for this ship. Single and systems, cargo handling, effective manning, operational
CR propellers will be designed, manufactured, and tested on analysis, survivability, hull structures, produceability and
the model hull. fleet management for applications to mid-term fast sealift
ships. This program is not directed to any specific ship
acquisition program. An overview of the program was
NOMENCLATURE presented by Bauman [1] and preliminary results on the
development of baseline ship concepts and various
c Chord length promising technologies and their potential application to the
C, Lift coefficient: C, = (dLldr)lipV;c baselines were presented by Kramer, et al [2].
This paper presents an overall powering performance
CT, Thrust loading coefficient: CT, = Ti+py527r.R2 prediction for single open and contrarotating (CR)
Diameter propellers for three baseline fast sealift configurations, as
Non-dimensional circulation: G = 1-127r.RVs part of the effort led by the hullform and propulsor studies
J, Advance coefficient: Js = Vs I rzD group in support of the technology development program.
Lift There are growing interests in power-saving ship propulsion
Shaft speed (rev/sec) devices in the commercial and the military sectors as well,
PC Propulsive coefficient: PC = PEI PD especially when affordability is becoming increasingly
important.
PE Effective Power
Pp Delivered Power
A substantial research and development work on CR
propulsion in the U.S. has been accomplished through
Local radius MARAD sponsorship in the period of 1950-1980 [3].
R Propeller radius Under the MARAD sponsorship, the Curtiss-Wright
t Thickness or thrust deduction factor designe-d and developed a family of marine transmissions
T Thrust suitable for serving a broad field of applications within a
VR Relative velocity 20,000 to 60,000 horsepower (hp) range. The basic systems
Vs Ship speed evolved into a two-stage planetary gear with mechanical
Vx Axial velocity reversing transmitting 40,000 hp at the propeller rotational
wr Taylor wake fraction speed of 105 rpm and a similar system having a
contrarotating second stage rated at 60,000 hp at 105 rpm

195
dual output speed. Second stage planetary units with and MID-TERM FAST SEALIFT BASELINES
without CR shafts were built and successfully tested up to
40,000 hp for 450 hours. The one unit was accepted by the At the outset of the technology development program
American Bureau of Shipping for installation on the effort, a design approach was established for five baseline
PONCE DE LEON RO/RO (roll on - roll off) vessel. The concepts (Baselines Oa, 1, 2a, 3, 4a). The design philosophy
ship renovation, however, never took place. for Baselines Oa, 1, and 4a was predicated on a purely
More recently, CR propeller and propulsion systems military mission, while Baseline 2a, an independent but
have been successfully installed to large conunercial vessels parallel effort by MARAD, was approached as a
in Japan. The II-II (Ishikawajima Harima Heavy Industries commercially viable ship with potential for military utility.
Co., Ltd.) has built a 37,000 DWT bulk carrier JUNO [4] Baseline 3, though having a military mission, was treated
and MHI (Mitsubishi Heavy Industries) a 3,500 (Metric slightly differently between mid-term sealift and future far-
Ton) car carrier, TOYOFUJI 5 [5] with CR propulsion term Combat Logistics Force ships. Detailed description of
system. It is also reported that an MI CR propeller will be these five baseline concepts are presented by Kramer, et al
installed on a 254,000-dwt VLCC scheduled for in Ref. 2. As these baselines were being refined, a twin-
commissioning in 1993 in Japan. The Japanese innovations skeg variation of the open-stern Baseline Oa was developed
[4,5] removed many of the reliability concerns related to and named Baseline Ob.
gears, shafting, and seal problems. In this paper, three baseline configurations (Ob, 1, and
After a series of discussions and technology 4a) were evaluated for powering performance. Fig. 1 shows
assessments by members of tlae NAVSEA. hallform and the body plans of the three baseline vanfigurations. The
propulsor group and the machinery systems group, it was ship and propulsion characteristics of Baselines Oa, Ob, 1,
concluded that the CR propulsion should be seriously and 4a are presented in Table 1. The ship and propulsion
considered for the mid-term fast sealift ships. Although the characteristics of other baselines are presented in Ref. 2.
CR propulsion approach appears sound and has been Baseline Ob is a twin-skeg ship with a notional ship
generally proven with the engineering experiences to date speed of 30 knots, developed by Ingvason of SI Shipping
by the U.S. and the Japanese, the great need is for Konsult AB (SISAB) of Goteborg, Sweden under NAVSEA
operational experience at higher power ratings. sponsorship. The advantages of the twin skeg are that a
Construction and operation of a fast sealift ship built to large diameter propeller can be installed without penalty of
commercial standards would provide the necessary large appendage resistance and that a high hull efficiency
experience to make this technology a significant step can be achieved due to the presence of the skeg.
forward for U.S. industry.

Table 1 - Ship and Propulsion Characteristics

Baseline Oa Baseline Ob Baseline 1 Baseline 4a

SHIP TYPE RO/RO RO/RO MCP RO/RO

NOMINAL SPEED (KTS) 30 30 20 35

SHIP DIMENSIONS
LOA (fi) 948.0 948.0 927.0 948.0
LBP (ft) 928.0 926.0 890.0 928.0
Beam (B), Molded (ft) 105.5 105.5 105.5 105.5
Depth, to Main Deck (ft) 77.5 92.75 77.5 77.5
Draft (T), Design (fi) 33.0 32.5 35.1 32.5
Displacement, Design (LT) 49,600 56,000 50,700 49,900

PROPULSION PLANT
No. of Shafts 2 2 1 3
Prime Mover M.S. Diesel GT or Diesel M.S. Diesel ICR GT
Engine Manufacturer Colt-Pielstick GE or CP Colt-Pielstick W/RR
No. of Engines 4 4 2 6
Installed Power, BHP 117,280 117,280 32,580 174,000
Notional Prop Diameter (ft) 24.0 32.0 27.4 Center - 30
Wing - 24
Notional Prop RPM at F.P. 115 65 75 Center - 95
Wing - 120

196
 11111WINNInanst.- Baseline 1 is a single-screw, 20 knot multi-cargo
11111MIEMINIX prepositioning (MCP) ship that carries a combination of
vehicles and containers. The hull form is basically a stretch
and upscaling of a Hogner-stern design represented by
NSRDC Model 4883 [6].

LJrt
Baseline 4a is a 35-knot, triple-shaft ship with an open
stern. It was developed by CDNSWC. Twin-skeg
variations for this baseline were also investigated
independently by BMT and NAVSEA and it was concluded
that the design is feasible but would require considerable
hullform development to insure low resistance. In this paper
only the open stern concept for Baseline 4a is evaluated.
I / ji
\\*KulaworA_AINWAIWANFAIIPW PROPELLER PERFORMANCE
Baseline Ob
PREDICTIONS
For each of the three baseline hulls, the performance of
single and CR propellers were predicted over a range of
design RPM, propeller diameter, and blade area by using the
lifting-line computer code developed at MIT [7]. The
circulation distribution was determined by unloading the
optimum distribution near the root and tip region to reduce
k
mom/
hub and tip vortex cavitation. For single propellers, the
blade number was 5, and for CR propellers, the forward and
aft blade numbers were 5 and 7, respectively. For CR
propellers, the diameter of the aft propeller was 83% of the
forward propeller diameter, and the axial distance between
the forward and aft propellers was 25% of the forward
propeller diameter. The torque ratio between the forward
and aft propellers was set to 1.0.

___imtgwoli The thickness (t/c) and chordlength (c/D) distributions,

as shown in Fig. 2, were determined based on the existing
database. The thickness distribution, t/c, was kept the same
for all designs, while t/D and c/D were adjusted for each
Baseline 1 EAR to produce the given t/c as shown in Fig. 2. For each
design rpm and diameter, the blade area was minimized
such that the following criteria are mee
No thrust breakdown due to cavitation, according to
Gawn-Burrill criteria,
Maximum blade sectional lift coefficient less than
0.5, and
Maximum blade static stress less than 12,500 psi.
For ranges of design parameters considered, the thrust
breakdown was predicted to be well within the limit. The
lift coefficient and stress criteria control the blade area for
different range of design advance coefficients, J. For lower
design Js, the stress limit determines the blade area, and for
higher design Js, the lift coefficient limit controls the blade
area.

Baseline Ob (SIS AB)

Baseline 4a
The performance of single and CR propellers was
Fig. 1. Body Plans of Three Baselines predicted for an estimated mean axial velocity shown in Fig.

197
3. The resistance and propeller/hull interaction coefficients 0.020
were estimated by the Danish Maritime Institute. At 30
knots, the estimated effective power (Pe) was 74,634 hp, the
thrust deduction (1-t) and wake fraction (1-wT) were 0.82 0.015
and 0.80, respectively, for 32-ft diameter propellers. For
different propeller diameter, the thrust deduction value was Optirnurn
assumed to be constant, but the wake fraction was scaled 0 0.010
based on Fig. 3.
Figs. 4 and 5 show a comparison of optimum and N,

unloaded circulation shapes for single and CR propellers, 0.005

respectively. For each design condition, the circulation
distribution was scaled to produce the required thrust The
hub and tip unloading reduces the optimum propulsive 0.000
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
efficiency by 1 to 2 points. Fig. 6 shows the propulsive
efficiency of single and CR propellers at 30 knots as a Nondimensional Radius, r/R
function of design rpm for three diameters. The propulsive
efficiency (PC) is defined as the ratio of the effective power Fig. 4. Circulation of Baseline Ob Single Propeller
(PE) to the delivered power (PD). For single propellers, the
optimum efficiency increases slightly with increasing
diameter and decreasing rpm. For CR propellers, however,
the optimum efficiency is essentially constant for the three 0.015
Unloaded
0.40
AFT PROP
0.010 Optimum
0.30
/f/

0.20 0.005 Optirn

Unloaded
/ FWD PROP
/ I
0.10 i
0.000
0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.00
1.0
Nondimensional Radius, r/R.
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Nondimensional Radius, r/R Fig. 5. Circulation of Baseline Ob CR Propeller
Fig. 2. Chord and Thickness

0.90
1.2
Baseline 4a CR Propellers
0.85 24119.92 = Dfwdeaft

/ \
Wing - 32126.56 28/23.74
1.0

0.80
0.8 - Center Shaft
Baseline 1
0.75
0.6 32 ft
Baseline Ob (SISAB)
0.70 28 ft
0.4 24 ft
Single Propellers
0.65
0.2
40 50 60 70 80 90 100 110 120 130
0 2 4 6 8 10 12 14 16 18
Design RPM at 30 lcnots
Radius, ft
Fig. 6. Propulsive Efficiency of Baseline Ob
Fig. 3. Circumferential Mean Axial Velocity
198
 diameters. It is mainly because the thrust loading 0.030
coefficients are very small and as the diameter decreases,
the hull efficiency increases due to the lower wake fraction 0.025 Unloaded,____,
(1-wT) scaled from Fig. 3. It is to be noted that for each
design point a different expanded area ratio (EAR) has been 0.020 S.

selected to meet the stress and lift coefficient requirements S.

shown above. For each diameter, the optimum rpm and 0 0.015 Optimum
corresponding EAR, PC and CT. are presented in Table 2.
0.010
Table 2. Summary of Baseline Ob Propeller Performance
0.005
D (ft) RPM PC I EAR C,., 1 -w,. 1-c
,, ,,
I I

SINGLE PROPELLER
24 110 0.762
0.000
0.55 0.427 0.748 0.82
28 90 0.770 0.50 0.314 0.783 0.82 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
32 70 0.779 0.46 0.241 0.800 0.82
CR FROPELLER Nondimensional Radius, r/R
24/19.9 ±100 0.832 0.45/0.55 .212/.313 .7481.697 0.82
28/23.2 +70 0.836 0.43/0.58 .153/.234 .783/.745 0.82 Fig. 7. Circulation of Baseline 1 Single Propeller
32126.6 +60 0.832. 0.35/0.47 .118/.178 .800/.780 0.82.

0.030
Baseline 1 Unloaded
0.025 AFT PROP
The effective power and the propeller/hull interaction Optimum
coefficients were estimated based on existing ships similar 0.020
to Baseline 1. At 20 knots, PE was estimated to be 18,433
hp, (1-w.1) and (1-t) were 0.755 and 0.845, respectively, for 0 0.015 Optimum
28-ft diameter propellers. An estimated circumferential
mean wake distribution as shown in Fig. 3 was used for 0.010 FWD PROP Unloaded
propeller performance predictions.
Figs. 7 and 8 show a comparison of an optimum and 0.005 I
unloaded circulation shapes for single and CR propellers,
respectively. Fig. 9 shows the propulsive efficiency of 0.000
single and CR propellers at 20 lcnots as a function of design 0.2 03 0.4 0.5 0.6 0.7 0.8 0.9 1.0
rpm for four diameters. The optimum efficiency increases
with increasing diameter and decreasing rpm for both single Nondimensiona1 Radius, r/R
and CR propellers. The incre,ase in optimum efficiency is
Fig. 8. Circluation of baseline 1 CR Propeller
more pronounced than that of Baseline Ob. It is because the
thrust loading coefficients for Baseline 1 propellers are
about twice as large as those of Baseline Ob. As before, the
EAR is adjusted for each design point to meet the stress and 0.85
lift coefficient requirements shown above. For each
diameter, the optimum rpm and corresponding EAR, PC and Df= 2s ft CR Propellers
CT, are presented in Table 3. 0.80
- -----
24 ft
Df= 22 ft
Table 3. Summary of Baseline 1 Propeller Performance 0.75 , ....
.."
D (ft) RPM PC EAR I CT, 1-w, 1-t
..
D= 28 ft
SINGLE PROPELLER 0.70 26 ft
22 110 0.705 0.52 0.823 0.716 0.845 E. 24 ft
24 100 0.725 0.48 0.691 0.732 0.845 Single Propellers D= 22 ft
26 80 0.744 0.46 0.589 0.744 0.845
28 70 0.760 0.44
0.65
0.508 0.755 0.845
CR PROPELLER 40 50 60 70 80 90 100 110 120 130
22/18.26 +90 0.766 0.45/0.54 .428/.573 .716/.700 0.845
24/19.92 +80 0.783 0.39/0.49 .357/.485
Design RPM at 20 knots
.732/.699 0.845
26121.58 +70 0.798 036/0.46 .303/.417 .7441.717 0.845
28123.24 +60 0.811 0.35/0.45 .259/.362 .7551.732 0.845
Fig. 9. Propulsive Efficiency of Baseline 1

199
Baseline 4a 0.015

Baseline 4a is a 35-knot, triple-shaft ship with an open

stern. The maximum diameters of the centerline and two
wing shafts were estimated to be 30 and 24-ft, respectively, 0.010
to allow a tip clearance of 25% propeller diameter. It was
assumed that two diesel engines, each producing 29,300 hp,
will be used to drive the center shaft, and three gas turbines
(LM 2500), each producing 27,000 hp, to drive each wing 0.005
shaft.
The effective power was estimate-d to be 130,000 hp at
35 lcnots. Propeller/hull interaction coefficients were
estimated based on the exsiting database. For the center 0.000
shaft, (1-t) and (1-wT) were estimated to be 0.85 and 0.89, 0.2 03 0.4 0.5 0.6 0.7 0.8 0.9 1.0
respectively, for 30 ft diameter propellers. For the wing Nondimensional Radius, r/R
shafts, (1-t) and (1-wT) were estimated to be 0.85 and 0.92,
respectively, for 24-ft diameter propellers. It was assumed
Fig. 10. Circulation of Baseline 4a Center Shaft Propeller
that the required thrust for each shaft was proportional to the
available power, i.e., 26.6. percent for the center shaft and
36.7 percent for the wing shaft.
Only single-open propellers were investigated for 0.030
Baseline 4a. Figs. 10 and 11 show a comparison of opti-
mum and unloaded circulation shapes for the single pro- 0.025 Unloaded -----.
pellers on the center and the wing shafts, respectively. Fig.
12 shows the propulsive efficiency of single propellers at 35 0.020
knots as a function of design rpm and diameters. The opti-
mum efficiency increases with increasing diameter and 0.015
decreasing rpm. The EAR changes for each design point to
meet the stress, lift coefficient, and the thrust breakdown 0.010
requirements shown above. Because of high thrust loading
coefficients of the wing propellers, the EARs are substan- 0.005
tially higher than the other propellers to avoid the thrust
breakdown. The range of EARs for 24-ft diameter was 0.000
between 0.7 and 0.9 and for 22-ft diameter between 0.8 and 0.2 03 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1.0. For diameters smaller than 22-ft, the EARs were larger Nondimensional Radius, r/R
than 1.0, which is considered to be out of the practical limit
for open propellers. For each diameter, the optimum rpm Fig. 11. Circulation of Baseline 4a Wing Shaft Propeller
and corresponding EAR, PC and CT. are presented in Table
4. The PCs in Table 4 are defined by PE/PD for each shaft.
The total PC for a combination of 30-ft diameter center
propeller and 24-ft diameter wing propellers is 0.704, 0.85
Caner Shaft
requiring 184,700 hp with 44,800 hp at the center shaft and 0. 80
69,960 hp at each wing shaft. D = 30 ft D = 2811
0.75 D = 26 ft

0.70
/
Table 4. Summary of Baseline 4a Single Propeller 0.65 /
Performance D= 24ft ,-**
0.60
Wing Shaft /I
D (fi) RPM PC EAR CT. l -WT I 14
0.55
CENTER SHAFT D =22 ft
90 0.752 0.48 0.205 0.895 0.85 0.50
26 160
0.46 0.177 0.893 0.85 40 60 80 100 120 140
28 80 0.762
30 70 0.772 0.44 0.154 0.890 0.85 Design RPM at 35 knots
WING SHAFT
22 130 0.659 0.80 0.395 0.92 0.85 Fig. 12. Propulsive Effciciency of Baseline 4a
24 110 0.682 0.70 0.332 0.92 0.85

200
 CONCLUSIONS AND FUTURE WORK REFERENCES
The NAVSEA hull form and propulsor group made the Bauman, W.D., "Fast Sealift Technology Development
following conclusions: Plan (TDP)," Proceedings of the Military Sealift Technology
The SIS AB hull form is the best candidate for the 30 Conference, 1990, Washington, D.C.
knot design which may possibly be extended to 35
knots. Kramer, R.P., A.C. Landsburg, C.F. Snyder III, and S.B.
Operational Re-quirements studies inclicate that ships Wynn, "Mid-Term Sealift Technology Development
of 25 to 30 knots are most likely needed. Program: A Progress Report," presented to the Symposium,
The propulsive efficiency of the CR propeller is Naval Ship Design: Setting Course for the 21st Century,
approximately 7 to 9 percent higher than that of the Arlington, VA, February 1992.
single propeller.
"Application of Contrarotating Propulsion to U.S. Flag
Based on these conclusions and funding considerations, Merchant Vessels: Executive Summary," by the Office of
it was decided that only the twin-skeg SISAB hull form Commercial Development, MARAD, July 1980.
developed by Ingvason of Sweden will be built and tested at
model scale. A 30-ft hull model is currently being Nishiyama, S., Y. Sakamoto, S. Ishida, R. Fujino, and
constructed. Single and CR propellers are being designed M. Oshima, "Development of Contrarotating-Propeller
for powering experiments in the CDNSWC towing tank in System for Juno - a 37000-DWT Class Bulk Carrier,"
September 1992. A significant risk associated with the SNAME Transactions, Vol. 98, 1990.
design of high speed (30 to 35 knot range) ships is in the
area of hull vibration and propeller cavitation. After the Nakamura, S., T. Ohta, K. Yonekura, T. Sasajima, and
powering experiments, it is recommended that cavitation K. Saki, "World's First Contrarotating Propeller System
and hull vibration model tests be carried out in the Successfully Fitted to a Merchant Ship," The Motor Ship,
CDNSWC's Large Cavitation Channel in order to reduce 1 1 th International Marine Propulsion Conference and
these uncertainties and confirm the viability of the concept. Exhibition, March 1989.

Dillon, E.S., F.G. Ebel, and A.R. Goobeck, "Ship Design

ACKNOWLEDGMENTS for Improved Cargo Handling," Transactions SNAME 1962.

This work was funded by NAVSEA 55W3 under Work Coney, W.B., "A Method for the Design of a Class of
Request N00249-WR-15032, Program Element 63564N. Optimum Marine Propeller," Doctoral Dissertation,
The work was performed under Work Unit Number 1-1521- Department of Ocean Engineering, MIT, 1989.
853. Estimates of the wake and propeller/hull interaction
coefficients were provided by William E. Beaver of DTMB.

201
SESSION II-C

CAVITATION
 The Cavitation Committee Report
R. Latorre (University of New Orleans, USA)

Cavitation noise, erosion, and

related performance reduction are major
cavitation research areas. The commis-
sioning of the La.rge Cavitation Channel
(LCC) of the David Taylor Model Basin
CDNSWC in Memphis, Tennessee [1],
now enables these problems to be
investigated in greater detail.
The small number of papers
reflects not only reduced cavitation
funding levels, but also schedule overlap
with two international meetings held in
China and Germany.
During the planning of the ATTC,
Mr. Sun ()in submitted a short commu-
nication, "An Expression of Inception
Cavitation Numbers," which summarizes a
methodology for predicting propeller
cavitation noise.
This completes our committee
report.

REFERENCES

[1] Etter, R. and M.B. Wilson, "The Large

Cavitation Channel," Proceedings 23rd
ATTC June, 1992, p. 243.

205
 An Expression of Inception Cavitation Number
S. Qin (Shanghai Ship and Shipping Research Institute, China)

ABSTRACT

The inception cavitation number a

and INTRODUCTION
corresponding propeller
hvolutions RPM, are essential
parameters for thé prediction of the Nowadays the noise control is an
interesting subject not only for the
propeller noise. While RPM and a
depend on many factors,
naval ships and submarines, but also
the thrust for commercial ships.
loading coefficient Te and pitch ratio Increased
P/D propulsive power makes the situation
are important. In this complex. More and more people in the
contribution a simple expression, marine inudstry are lay emphasizing
which comes from classical Burrill noise reduction and improved living
ship trial cavitation chart and modern
model environment.
noise experimental analysis is
presented. Based on the expression
the predicted results for the noise Apart from the vibration the
spectra noise comes from high frequency
(SPL, vs. 1/3 octave
frequency) are acceptable. pressure impulses from the machinery
and propellers. The inception tip
NOMENCLATURE
vortex cavitation number aT and
corresponding propeller revolution per
a' Inception noise cavitation
minute RPM, play an important role in
propeller arise. It has pointed out
number
that at the intermediate test
conditions between no visible tip
RPMI Inception noise revolution
per minute of propeller vortex cavitation and the formation of
the tip vortex cavitation there
Te thrust loading coefficient appeared bursts or spikes in the noise
signal and abrupt rise in the noise
level [1]. Obviously, the cavitation
c A cavitation number number of the tip vortex inception is
a little larger than the value
KT thrust coefficient corresponding to when the tip vortex
J observed attached to blade. How to
Advanced coefficient express this in a tip vortex
cavitation number c is an important
P/D pitch ratio problem. I
AE/AD blade area ratio

207
AN EXPRESSION FOR INCEPTION, CAVITATION

NUMBER

Many factors, such as propeller

0.7 r )2]
geometry, tip vortex circulation, a I- El+(
boundary layer thickness and air
content influence on inception tip
vortex cavitation number. The loading
and P - P
or thrust loading coefficient Te v
(4)
pitch ratio P/D are more important. It aI - p V2
is well known that the classical
Burrill chart for cavitation
criterion, which came from ship trial
results offered clear lines of the This expression should be in
cavitation formation and development results
From left higher part to right agreement with experimental
[2].
for model propellers tested in
lower part on the chart there are 30%,
2.5% back cavitation, cavitation tunnel for the noise
20%, 10%, 5%,
measurement of the inception tip
lines and a suggested lower limit line
It is vortex cavitation [I]. The model
for tugs, as well as trawlers.
propellers tested in [I] were Japanese
clear that just on or below the
noise six bladed SRI-a series for high speed
suggested limit cavitation
container ships. The propeller series
occurs. cavitation characteristics were found
to be superior to those of MAU
We can write an approximate
propellers. The systematic noise
expression for the limiting line as: measurements were made using a
hydrophone, 1/3 octave filter, level
recorder oscilloscope. These
Te =CJaA c= const = 0.208 (1)
measurements
and
showed there is the
propeller tip vortex cavitation noise
envelope. The upper limit aT of the
envelope is determined either by
where observing the oscilloscope trace of
hydrophone signal or from the 1/3
octave frequency analysis. It occurs
when the characteristic bursts first
Te = T/AP (i p [V2+(0.7 n nD)2]) Based on
appear in the noise signal.
measurement data, some of which
obtained from their open water test
curves there are three expressions for
= 2.55 x KT/J2 [AE/AD (1.067-0.229 x model
three different pitch ratio
propellers.

0.7 ir
Te = 6 0.130 for propeller
x P/D)] - [1+( )2 ]
(2) c1 A c1 =
P/D = 0.75 (5)

- P
Po v Te j a 0.160 for propeller
- (3) =c2 A c2- =
a A = p [V2 + (0.7 W nD)2J
P/D = 1.00 (6)

Te = j 6 c = 0.208 for propeller

c3 A 3
124, P/D - 1.50 (7)

208
 It goes without saying that the
third line is concordant with Burrill ACKNOWLEDGMENTS
lower limit line for tugs. The second
one is lower and the first is lowest. The author wishes to express his
The later two lines located in right sincere gratitude to Mr. P. A.
lower part of the Burrill chart. These Fitzsimmons, Dr. Kuochang Jane and Dr.
results are quite reasonable. Vladimir Ankudinov for their friendly
help when working at BMT Int. He also
Then the general expression has wishes to thank Dr. J. W. English for
been derived: his glancing over the article.

REFERENCES
Te = [0.02 + 0.166 x P/D - 0.027 x
Latorre, R., "Propeller Tip
x (P/D)2] x T.-J-7-A (8)
Vortex Cavitation Noise Inception,"
The Propellers '81 Symposium, 1981.

Burrill, L. C., "The Optimum

APPLICATION OF THE EXPRESSION Diameter of Marine Propellers; a New
Design Method," Trans. NECI Vol. 72,
The predicting noise spectrum, 1955.
especially in cavitation case is a
challenge target. The tentative
method is basic. A semi-empirical
method developed by BMT COTEC was
used. It was mainly based on ship
trial data. An improvement was
obtained when expression (8) was
applied in the BMT predicting noise
program with consideration given to
scale effect. The other improvement
to the prediction involves corrections
for the influence of inclined
propeller shafts. The predicting
results of sound pressure level SPL
vs. 1/3 octave frequency for four
different propeller cases are
reasonable and acceptable compared
with ship trial noise. The new BMT
program is now available for design
work.

209
 Cavitation Scaling Experiments
with Axisymmetric Bodies
Y.P. Kuhn de Chizellel, S.L. Ceccio', C.E. Brennan', Y. Shen'
('California Institute of Technology, USA, 'University of Michigan, USA,
3Carderock Division, Naval Surface Warfare Center, USA)

bubble radius at the base

ABSTRACT Re Reynolds number, U0D/v
time
Several experiments performed by Ceccio and Uo free-stream velocity
Brennen (1991, 1989) and Kumar and Brennen kinematic viscosity
(1992, 1991) have closely examined the density
interaction between individual cavitation
a cavitation number, p, )/o. 5 pUO2
bubbles and the boundary layer, as well as
statistical properties of the acoustical a. inception cavitation number
signals produced by the bubble collapse.
All of these experiments were, however, 1. INTRODUCTION
conducted in the same facility with the same
headform size (5.08cm in diameter) and over Questions on the scaling of cavitation
a fairly narrow range of flow velocities have been asked for many years. The purpose
(around 9m/s). Clearly this raises the of the experiments described herein is to
issue of how the phenomena identified change investigate the effects of scale in the
with speed, scale and facility. The present cavitation occurring on a simple
paper describes experiments conducted in axisymmetric headform. We will here focus
order to try to answer some of these on traveling bubble cavitation, and the
important questions regarding the scaling of interaction between the flow and the
the cavitation phenomena. The experiments dynamics and acoustics of individual
were conducted in the Large Cavitation bubbles. Experiments by Ceccio and Brennen
Channel of the David Taylor Research Center (1991) on 5.08cm diameter axisymmetric
in Memphis Tennessee, on geometrically headforms had revealed a surprising
similar Schiebe headforms which are 5.08, complexity in the flow around single
25.4 and 50.8cm in diameter for speeds cavitation bubbles. Photographic
ranging up to 15m/s and for a range of observations showed that the bubbles have an
cavitation numbers. approximately hemispherical shape and are
separated from the solid surface by a thin
film of liquid, as had been observed by
Blake et al. (1977). The collapse phase
NOMENCLATURE appeared quite complex and consisted of at
least three processes occurring
simultaneously, namely collapse, shearing
pressure coefficient, (P P)/0.5pUO2 due to the velocity gradient near the
headform diameter surface and the rolling up of the bubbles
I* dimensionless acoustic impulse into vortices as a natural consequence of
static local pressure the first two processes. It was noted that
Po static free-stream pressure the collapse phase was dependent on the
Pv water vapor pressure shape of the headform and the details

211
differed between the ITTC headform (Lindgren of solid blocks of clear lucite and were
and Johnsson, 1966) which possessed a mounted in the working section as shown in
laminar separation and the Schiebe body figure 1. The Schiebe headform has a
(Schiebe, 1972; Meyer, Billet and Holl, minimum pressure coefficient of Cpmin.-0.78
1989) which did not. Bubble fission during and is known to be quite robust to laminar
the collapse phase was sometimes observed flow separation. Lucite was chosen for its
and could produce several acoustic pulses. good acoustical match with water in addition
One set of scaling issues arises because to its electrically insulating properties.
the ratio of the cavitation nuclei size to 102 cm

the headform size changes as the headform

diameter changes. The other set of scaling
STFIUT
issues derives from the complex interactions THICKNESS
0.168
between the bubbles and the flow close to CHORO

the headform with which the bubbles

interact. Since the flow is Reynolds number 152 cm
dependent, scaling effects will also be
caused by the changes in both body size and
tunnel velocity. The question of stability LUCITE
HEADFORM
of the bubble surface also arises and will
FIN
be influenced by the Weber number. 50.8
In other papers, Kuhn de Chizelle et al. cm

(1992a) have presented data on the size,

51 cm 126cm
sphericity, population density and collapse
303 cm
location of traveling
the cavitation
bubbles. Data the onsurface streak
25.4 cm-
attachment location for those bubbles which
trigger patches was also included. The
present paper will focus on the scaling of 48 cr.1 27 cm
311 cm
the noise produced. The signals recorded
LUCITE
from theelectrodes will help in the HEADFORMS

understanding of the dynamics of the

bubble's collapse mechanism and the 50 3 cm
relationship with the produced noise.
45 cm83 crn 5.05cm

188 cm

2. EXPERIMENTAL SETUP Fig. 1 diagram

Schematic the three
headforms with afterbodies and supporting
strut.
Large Cavitation Channel
The authors were fortunate to have the
opportunity to examine some cavitation
scaling effects by conducting experiments in Electrode bubble detection
a new facility called the Large Cavitation Silver epoxy electrodes were installed
Channel, which has just been constructed for flush inheadform.
the A pattern of
alternating voltages is applied to the
the David Taylor Research Center (Morgan,
this facility is a very electrodes and the electric current from
1990). Briefly
large water tunnel with a working section each is monitored. When a bubble passes
which is 3.05m x 3.05m in cross-section. It over one of the electrodes, the impedance of
is capable of tunnel speeds above 15m/s and the flow is altered, causing a drop in
the pressure control allows operation at current (Ceccio 1989). Thirteen of these
sufficiently low pressures in the working electrodes take the form of small patches
section to permit cavitation investigations. (about lmm in diameter) at different axial
locations. A peak detector connected to one
Headforms of the electrodes allowed detection of the
Three Schiebe headforms of diameter
presence of a traveling bubble.
5.08cm, 25.4cm and 50.8cm were machined out

212
 0.54x106 to 9.41x106. For each velocity
High speed photography and flash
Two about five cavitation numbers were
cameras triggered simultaneously,
were set up in order to take flash pictures investigated, ranging from bubble inception
of individual to fully attached cavitation.
cavitation bubbles at
different angles and different enlargements.
Four powerful EGG SS166 flash heads with 3. CAVITATION APPEARANCE
SS162-165 energy storage racks were used.
The film exposure time was of the flash Inception
duration and was measured to be about 30gs. Cavitation inception data was described
Triggering could be done either manually or previously in Kuhn de Chizelle et al.
through a computer controlled lock-out (1992a,b) and will only be summarized here.
system in series with the electrode peak Inception was defined as the occurrence of
detector signal. A delay unit was employed some minimum cavitation event rate over the
in order to take photographs of bubbles at entire body. The inception cavitation
various times after passing an electrode. number was found to increase with increasing
air content and headforiit size. For the
Hydrophones
The larger headforms and 80% dissolved air
interiors of the headforms were
hollowed out and filled with water at content the inception cavitation number was
atmospheric pressure in order to place extremely close to the -Cpmin=0.78 value.
hydrophones as close as possible to the Naturally, higher cavitation numbers did not
cavitation and provide a fairly reflection result in any events. For later reference
it is important to observe that when the
and reverberation-free acoustic path between
the cavitation and the hydrophone. ITC-1042 pressure is sufficiently low for the
hydrophones with a relatively
smallest headform to reach inception, the
flat,
isotropic response out to 80kHz were used. larger headforms already exhibit extensive
The center of the hydrophone was placed on cavitation patterns. The interesting range
of cavitation numbers therefore varies
the axisymmetric axis, one headform radius
from the front stagnation point. 64 kbyte considerably from one headform to another.
signal acquisitions with 16 bit resolution Bubble shape
were made at a 1MHz sampling rate. For cavitation numbers close to the
Reciprocity type calibration was performed minimum pressure coefficient, -Cpm,=0.78,
by comparing the internal hydrophone signals the bubble life-time is extremely short.
with those from two STI hydrophones (with a Under these conditions they all assume a
flat frequency response up to 100kHz), very thin disk-like geometry and there is
installed upstream and downstream of the little or no growth normal to the headform
headform in a water filled recess in the surface due to the high normal pressure
upper wall of the working section. using gradients.
the manufacturer's response curves for the As the cavitation number is decreased
hydrophones in the receiving and below inception, the bubbles grow in volume
transmitting modes and the theoretical (in diameter and in height) and adopt the
transfer function of the water medium, the hemispherical cap shape seen in
mean error on the reciprocity tests over the photograph 2a. One entirely new observation
whole range of measured frequencies (from from the current experiments was of the
lkHz to 100kHz) proved to be less than 3dB presence of wave-like circular dimples on
for all three headforms. the top of the bubble (figures 2a, 2b).
Test conditions These were not seen on the 5.08cm headform
The test matrix included three
and seem to become more pronounced as the
saturation air contents of 80%, 50% and 30% volume of the bubble increases. This effect
for each of the three Schiebe headforms. may be Weber number dependent. The dimples
The measurements were taken at three seem quite stable, and remain on the bubble
velocities of 9m/s, 11.5m/s and 15m/s,
until the very last stage of collapse.
allowing a Reynolds number range from Observations of bubbles on all three
headforms show that the maximum radius of

213
the base of the hemispherical cap, R, scales Bubble streaks and patches
linearly with the headform diameter, D (Kuhn At lower cavitation numbers and on the
de Chizelle et al., 1992a) and appears to be
larger headforms many of the individual
mostly cavitation number dependent. bubbles produce streaks of cavitation which
are "attached" to the headform at their
upstream end and are stretched and grow as
the bubble moves downstream (see figure 2b).
In the present experiments the frequency of
occurrence of attached streaks increased
with velocity and headform size (increased
Reynolds number) and with decreasing
cavitation number. For small enough
cavitation numbers the patch can out-grow
the bubble, leaving behind large transient
patch-like cavities as shown in photograph
2

Bubble-patch interactions
When the cavitation number is
sufficiently small, the transient patches
become fairly stable and remain on the
headform, thus creating attached cavities
for periods of up to a few seconds. As
their number increases the patches will
merge to create larger attached structures.
For the larger headforms at cavitation
numbers below 0.7 the two different kinds of
cavitation patterns, namely traveling
bubbles and transient attached patches would
co-exist. Quite remarkably, even for the
conditions at which we observe extensive,
fixed patch-type cavitation, some very
smooth hemispherical traveling bubbles are
still present. Bubbles riding over attached
cavities will eventually collapse and merge
completely with the latter upstream of its
closure region. As will be shown further in
this paper, these interactions greatly
affect the noise generation.

4. CAVITATION NOISE

Averaged bubble noise

For a range of cavitation numbers
between inception and a value at which the
cavitation patches persisted, it was
possible to identify in the hydrophone
output the signal produced by an individual
Fig. 2 High speed photography of cavitation bubble collapse. It was found necessary to
events taken on the 50.8cm diameter headform digitally high pass filter the signals using
at 30% saturation dissolved air content a cut-off frequency of 5kHz in order to
(distance between the two pairs of patch reduce the effect of vibration and noise
electrodes: 2.54cm). Figures a,b: U0=15m/s, caused by cavitation at the top of the
0=0.60; Figure c: U0=15m/s, 0=0.55

214
 supporting strut. This filtering did not, frequencies of 10kHz to 100kHz. The
however, substantially effect the results. measured decay between lkHz and 80kHz in the
Secondly the processing amplifier gain present data appears roughly constant, with
response was calibrated and applied to the a value of about -22dB/dec for all
results. conditions. This value is similar to the
FFT analyses of the
signals from value of -24dB/dec obtained earlier by
individual events were
performed for Ceccio and Brennen (1990) and by Kumar and
different cavitation conditions for Nyquist Brennen (1991) in the Caltech Low Turbulence
frequencies up to 500kHz. In order to Water Tunnel.
compare the shape of the power spectral Measurements of the frequency decay as a
density for different cavitating conditions function of the cavitation number for
the values have been non-dimensionalized by different cavitating conditions are shown in
the number of sampled points, N, times the figure 4. We observe that this slope seems
mean squared power amplitude, PSD, where to decrease as the cavitation number value
is reduced below 0.6. For some cavitation
conditions the slope can be as low as
PSD = -=-[C2(f0) + C2(f,v)+ 2
N2
EC2(f)]. The
-35dB/dec.
dimensionless PSD curves shown in figure 3
are data averaged over several cavitation
events. D [cm] tio[m/s]
-15
50.8 } 15
2.5.4 0 11.5
-20 5.08j 9
1000 w
--- -25
Q
,§*
-30
a.
T2
10 -35
-g

-40
,
04 5 0.6 0.7 0.8
O

0., t
1000 10000 100000 Fig 4. Average power spectral density slope
[Hz]
decay between lkHz and 80kHz [dB/decj.
Measurements for different headform
Fig. 3 Averaged dimensionless power diameters and velocities as a function of
spectral density signals for the 50.8cm the cavitation number.

headform: U0=9m/s 0=0.66;

Uo =11 5m/s a=0.64; The amplitudes of the acoustic pressure
U0-15m/s 0=0.61 pulses were measured by defining the
impulse, I, as the integral under the
instantaneous pressure time history from the
First we notice that for all headforms beginning of the collapse pulse to the
and tests conditions the measured spectral moment when the pressure returns to its mean
shape varies little with operating value. Since the impulse will vary
cavitation conditions close to inception. inversely with the distance of the
The influence of the hydrophone cutoff hydrophone from the noise source, we divide
frequency above 80kHz can be observed in all I by the appropriate headform radius and
signals. Asymptotic analysis of the form dimensionless
a impulse, I*, by
Rayleigh-Plesset equation predicts a power dividing by the free stream velocity and the
spectral decay of -16dB/decade for fluid density as indicated by the Rayleigh-

215
 Plesset analysis, so that I* = 87cI/pDU0. correspond to circumstances in which the
of the
cavities cover about 20% of the surface area
The hydrophone output for each
was examined in
of the headform in the neighborhood of the
experimental conditions
minimum pressure point. Higher void
order to identify at least 40 of the larger
fractions increase the interactions between
pulses associated with a bubble collapse.
the bubbles and the patches and considerably
The average values of the non-dimensional
reduce the acoustic impulse. Such an effect
impulses obtained in this way are plotted
was also previously reported by Arakeri and
against cavitation number in figure E.
Shanmuganathan (1985). It is important to
note that the results presented here
correspond to averaged values, and that the
80
D [cm] 110[mis) standard deviation of these impulses is
70 50.8 15 Therefore
C2

0 11.5
about 40% of the average value.
25.4
60
--- 5.08 z' 9 for identical cavitation conditions the
to 50 + cavitation noise may vary considerably from
X 40 one event to another.
M In summary, we find that the acoustic
20 impulse produced by a single bubble
while exhibiting considerable
10 t collapse,
with
variability, nevertheless scales
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 headform size and tunnel velocity in the way
O
which is expected on the basis of the
Rayleigh-Plesset analysis. Moreover, when
the bubble concentration exceeeds a certain
Fig. 5 Average dimensionless maximum
value the noise from individual events
acoustic impulse, I*, for all three
becomes attenuated.
headforms as a function of the cavitation
number.
5. CONCLUSIONS

The present experiments first revealed

The non-dimensional impulse is of the lower cavitation inception
for all three substantially
same order of magnitude
numbers for the larger headforms. At bubble
headforms. It initially increases as the smallest
inception conditions on the
cavitation number is decreased below
headform we observed fully developed
inception. However most of the data also The
attached cavitation on the largest.
indicates that the maximum impulse ceases to
most noticeable effect of scale on the
increase and, in fact, decreases when o is
appearance of cavitation was the increase in
decreased below a certain value (about 0.43, streaks and
bubble-generated attached
0.50 and 0.62 for the 5.08cm 25.4cm and On the
patches for the larger headforms.
50.8cm diameter headform). The decrease at
5.08cm headform a traveling bubble would
low cavitation numbers could be explained by occasionally generate two attached streaks
the increasing presence of attached
or tails at the lateral extremes of the
cavitation patches, damping the bubble
bubble. These would disappear almost
collapse mechanism. The location of the
immediately after the bubble collapsed. On
peaks appears to be somewhat influenced by
the larger headforms at higher speeds
the velocity: they are shifted towards
(larger Reynolds numbers) and low cavitation
higher cavitation numbers for lower
numbers the streaks began to occur more
velocities. This trend is consistent with
frequently and extend behind the entire
previous observations (Kuhn de Chizelle et
width of the bubble. The streaks would tend
1992a) the average void fraction
of
al., to produce a transient patch of attached
over the headform at constant cavitation
cavitation which would disappear as the
numbers, which exhibited an increase with a We saw that the pressure
bubble collapsed.
decrease in velocity. The conditions at
impulse could be adequately scaled with the
which the impulses, I*, are maximum seem to free stream
headform diameter and the

216
 velocity, in agreement with the Rayleigh- dimensional bubbly and cavitating flows
Plesset analysis. As expected, lower over slender surfaces. J. Fluid Mech.,
cavitation numbers lead to higher impulses Vol. 192, pp. 485-509.
as long as the interference with larger Fitzpatrick, H.M. and Strasberg, M. 1956.
patch type cavities remained limited. The Hydrodynamic sources of sound. First
shearing mechanism of individual bubbles, Symp. on Naval Hydrodynamics, Washington
producing trailing streaks and patch D.C., pp. 241-280.
cavities greatly decreased the emitted Kuhn de Chizelle, Y., Ceccio, S.L., Brennen,
collapse noise. Finally there appears to be C.E. and Shen, Y. 1992 (a). Cavitation
a bubble interaction effect when the cavity scaling experiments with headforms,
coverage on the surface of the headform Bubble dynamics. Proc. Second
exceeds about 20% which tends to decrease International Symposium on Propeller and
the cavitation noise. Cavitation ISPC, Hangzhou, China.
Kuhn de Chizelle, Y., Ceccio, S.L., Brennen,
ACKNOWLEDGMENTS C.E. and Shen, Y. 1992 (b). Scaling
experiments on the dynamics and
Large scale experiments like these acoustics of traveling bubble
require help of many people and the authors cavitation. Proc. Institution of
are very grateful to all of those who helped Mechanical Engineers, Cambridge, UK.
in this enterprise. We are very grateful to Kumar, S. and Brennen, C.E. 1991.
the ONR for their support under contracts Statistics of noise generated by
N00014-91-J-1426 (SLC) and N00014-91-J-1295 traveling bubble cavitation. ASME
(CEE, YKdC). We are also extremely grateful Cavitation and Multiphase Flow Forum,
to the David Taylor Research Center (DTRC) Portland OR, June 1991, FED Vol. 109,
and to their staff including W.B. Morgan for pp. 55-62.
making the use of the LCC possible for us Kumar, S. and Brennen, C.E. 1992. An
and to both Scott Gowing and James Blanton. acoustical study of traveling bubble
Po-Wen Yu from the U. of Michigan also cavitation. Submitted to J. of Fluid
provided important help with the Mach.
photography. Lindgren, H. and Johnsson, C.A. 1966.
Cavitation inception on headforms. ITTC
REFERENCES comparative experiments. llth Int.
Towing Tank Conf., pp. 219-232.
Arakeri, V.H. and Shanmuganathan, V. 1985. Meyer, R.S., Billet, M.L. and Boll, J.W.
On the evidence for the effect of bubble 1989. Free-stream nuclei and
interference on cavitation noise. J. cavitation. Proc. ASME Third Int. Symp.
Fluid Mech., Vol. 159, pp. 131-150. on Cavitation Noise and Erosion in Fluid
Brennen, C.E. and Ceccio, S.L. 1989. Recent Systems, San Francisco, FED-Vol. 88, pp.
Observations on cavitation and 52-62.
cavitation noise. Proc. ASME Third Int. Morgan, W.B. 1990.
David Taylor Research
Symp. on Cavitation Noise and Erosion in Center's Large Cavitation Channel. Proc.
Fluid Systems, San Francisco, FED-Vol. Int. Towing Tank Conference,
88, pp. 67-78. Madrid,
Spain, pp. 1-9.
Blake, W.K., wolpert, M.J. and Geib, F.E.
Schiebe, F.R. 1972. Measurements of the
1977. Cavitation noise and inception as cavitation susceptibility of water using
influenced by boundary layer development standard bodies. St. Anthony Falls
on a hydrofoil. J. Fluid Mech., Vol. Hydraulic Lab., Univ. of Minnesota, Rep.
80, pp. 617-640. No. 118.
Ceccio, S.L. and Brennen, C.E. 1991. The Vogel, A., Lauterborn, W. and Timm, R. 1989.
dynamics and acoustics of traveling Optical and acoustic investigations of
bubble cavitation. J. Fluid Mech., Vol. dynamics oF the
Laser-produced
233, pp. 633-660.
cavitation bubbles near a solid boundary
D'Agostino, L., Brennen, C.E. and Acosta, layer. J. Fluid Mech., vol. 206, PP.
A.J. 1988. Linearized dynamics of two- 299-338.

217
 Propeller Erosion Reduction
with an Asymmetric Preswirl Stator
T.B. Smith, K.D. Remmers
(Carderock Division, Naval Surface Warfare Center, USA)

tests on a model of the propeller shaft

ABSTRACT and strut duplicated this cavitation
pattern, beginning at approximately 25
This paper presents the results of knots on the leading edge of the root.
a series of model-scale tests of the After leading edge modifications were
erosion-reduction capabilities of a shown to be ineffective, cavitation
partial-span asymmetric preswirl stator relief holes were proposed, tested and
on the propeller of a large United found to solve the problem in the
States Coast Guard (USCG) patrol boat. tunnel. On the basis of this study, the
The propeller and stator were tested at entire stock of propellers was modified
one-third scale (X = 3.2) in the DTRC to include cavitation relief holes[1].
36-inch Variable Pressure Water Tunnel
(VPWT). A single shaft strut upstream Unfortunately, the installation of
of the propeller and a shaft inclination these holes did not provide a complete
of 9.25° simulated the full-scale solution to the propeller root erosion
propeller flow field. Cavitation problem. As shown in Fig. 1, the point
erosion tendency was investigated using of maximum damage had moved to a point
an erodable coating painted on the approximately halfway between the
propeller blades. Stator performance cavitation relief holes. The cavitation
was evaluated in a series of model tests d light erosion
at 30 knots, both with and without the
currently-installed cavitation relief ¡Mill heavy erosion
holes, using three different series of
stator blade angle of attack settings
These high Reynolds number experiments
showed that the blade root erosion
tendency was almost completely
eliminated by the preswirl stator with
the cavitation relief holes at the
propeller blade root closed. A lesser
degree of erosion relief was provided by
the preswirl stator with the cavitation
relief holes open.

BACKGROUND

During the routine drydocking and

inspection following builder's trials,
severe cavitation damage was noticed on
the propellers of a large, high-speed o
U.S. Coast Guard patrol boat. This
erosion was at the suction-side root of Fig. 1. Sketch of erosion pattern on suction side of
each propeller, about 75 percent of the full-scale, right-hand propeller.
chord length aft. Cavitation tunnel

219
erosion had become extensive enough that
some boats had to be brought into
drydock for replacement and repair of
their propellers after as little as 40
hours of high-speed operation.

Faced with the prospect of having

to replace over 100 propellers (worth
about three million dollars) with a new
design, the Coast Guard decided to look
for more economical solutions.

INTRODUCTION

At the reauest of the U.S. Coast

Guard (USCG), the David Taylor Research
Center (DTRC) began investigating this
problem. Preliminary studies indicated
that the leading edge suction side
cavitation responsible for the blade
root erosion was caused by the high
angles of attack characteristic of
inclined shaft propeller blades. One
proposed remedy was to improve the
propeller inflow velocity field with a
partial-span asymmetric preswirl stator.
By turning the flow near the blade root
closer to parallel to the propeller
shaft, the effective shaft inclination
angle could be reduced, resulting in
less severe angle of attack variations
at the inner propeller radii.

On the basis of these preliminary

studies, DTRC proposed an experimental
program to evaluate the effectiveness of
an asymmetric preswirl stator in
reducing blade root erosion on the USCG
patrol boat propeller. Reference [2]
discusses the DTRC design of this
stator. This paper presents the results
of a series of experiments performed in
the DTRC 36-inch Variable Pressure Water Fig. 2. Sketch of model configuration in DTRC 36-inch
Tunnel (VPWT) on DTRC Propeller 5138, a VPWT.
one-third scale model of the patrol boat
propeller. Cavitation patterns and number scale effects.
blade erosion tendencies were evaluated,
both with and without the asymmetric Two major factors were considered
preswirl stator. in modeling the upstream flow field. Of
primary importance is the hydrodynamic
MODEL DESCRIPTION flow angle, the angle of the propeller
inflow with respect to the axis of
To avoid flow blockage in the 36- propeller rotation. Secondary, but
inch VPWT, a model propeller diameter of still important, are the wake deficit
15.5 inches was chosen for this produced by the propeller shaft, strut
experiment. This set the model scale and barrel.
ratio at 3.2. Full-scale speeds were
consistently used to minimize Reynolds

220
 Table 1. Cavitation conditions in the 36-inch VPWT at
20, 25, 28 and 30 knots.

Vs 20 Ids 251<ts 281<1s 30 kts

Js 0.792 0.879 0.903 0.907

KQ .0465 .0436 .0408 .0393
KT 0.204 0.191 0.175 0.171
wF 1.000 1.011 1.020 1.033
0-p 2.123 1.329 1.041 0.884

No wake survey exists for this

class of patrol boat. From the
similarity of the ship's lines in the
area of the propeller and shafting to
previous/y-surveyed ships, the mean
hydrodynamic angle was estimated to be
9.5 degrees. Due to the constraints
imposed by existing hardware in the 36-
inch VPWT, a shaft angle of 9.25 degrees
relative to the flow was chosen for this
test.

Existing hardware for the 36-inch

VPWT was modified so that the single
strut and barrel for the USCG design
could be mounted securely in the open
jet test section of the 36-inch VPWT. A
sketch of the model configuration is
shown in Fig. 2.

EXPERIMENTAL PROCEDURES Fig. 3. Erosion pattern on model-scale propeller with

sealed cavitation relief holes. PSD 18444-03-90,
Table 1 lists the advance ratio, Frame 4.
torque and thrust coefficients, Taylor
wake fraction and cavitation number used photographs and videotape of the
in this experiment. These values were cavitation patterns were taken.
calculated from USCG full-scale trial Finally, the propeller was removed, anda
and DTRC open water data. Since full- photographic and/or videotape record was
scale thrust data were not available, made of the size, location and severity
the torque identity was used to set the of the eroded patches.
tunnel test section velocity.
Philly-Bond Epoxy (TA-30) was used
Cavitation erosion surveys were to seal off the cavitation relief holes
performed by running the propeller at 30 during some of the experiments. This
knots for 10 minutes. Before each allowed their effect on the propeller
erosion survey, the propeller was cavitation patterns and erosion
removed from the tunnel and coated with tendencies to be determined, both with
a mixture (half-and-half, by volume) of and without the stator.
white stencil marking dye and mineral
spirits. After drying for a minimum of Flow Field Verification
thirty minutes, the propeller was
reinstalled in the tunnel. The The first experiment was intended
propeller was then run for ten minutes to verify the simulation of the existing
at the 30 knot condition, while
cavitation conditions on the full-scale

221
Fig. 4. Cavitation on model-scale propeller with open Fig. 5. Erosion pattern on model-scale propeller with
cavitation relief holes. PSD 18524-03-90, Frame 1. open cavitation relief holes. PSD 18442-03-90, Frame
7.
propeller. For the first run, the
propeller was installed in the tunnel underneath.
with no erodable coating and with the
cavitation relief holes sealed off. For the third run, the cavitation
Test conditions were set at 20, 25, 28, relief holes were unplugged, and the
and 30 knots, while the cavitation propeller was prepared for another
patterns were recorded on videotape. erosion survey. This run simulated the
existing situation on the full-scale
vessel at 30 knots. An example of the
For the second run, the propeller
was painted with the erodable coating resulting cavitation pattern is shown in
Fig. 4. Especially notable is the cloud
and reinstalled for an erosion survey.
The cavitation relief holes remained cavitation at the suction side blade
The resulting erosion pattern root, which can be seen crossing over to
closed.
can be seen in Fig. 3. Even though the the pressure side through the cavitation
blade root erosion covers a fairly small relief hole on blade 5.
part of the blade, its intensity is
shown by the bright spot in the middle As shown in Fig. 5, the blade root
of the blade root erosion patch. This erosion occurs approximately halfway
marks where the erosion removed both the between the cavitation relief holes.
erodable coating and the underlying The erosion patches are slightly wider
anodizing to reveal the bare aluminum than those shown in Fig. 3; however,

222
 asymmetric preswirl stator was installed
I Pitchangle in the test section. This stator was
designed to replace the rope guard,
which normally covers the gap between
11' the propeller and the propeller strut
5 4 3 2 10 9 8 7 6 barrel.
1
Ten variable-pitch blades, five
expanded view of stator right-handed and five left-handed, were
set in a central hub of the same
diameter as the rope guard. To decrease
manufacturing complexity, all blades
were given an identical geometry. A
sketch of the stator is given in Fig. 6.

The cavitation relief holes were

left open for the first erosion survey.
The stator blades were set at the design
pitch settings, which were based on a
mean hydrodynamic inflow angle of 11
degrees. This setting had little effect
on the erosion pattern.

The next design iteration (rev A in

Table 2) rotated each blade by 2
degrees. This increased the lift
produced by each blade, turning the flow
into the propeller disk. The stator was
reinstalled, and another erosion survey
was made with open cavitation relief
holes. Unfortunately, stator blade #8
was insufficiently tightened down, and
rotated during the test to a high angle
of attack. Nevertheless, the
improvement in the erosion patterns, as
shown in Fig. 7, is obvious.

A more detailed analysis produced a

third pitch setting (rev B in Table 2)
starboard shaft,
that increased the blade loading,
lkg upstream especially on blades 2,3,7 and 8. The

Fig. 6. Sketch of asymmetric preswirl stator (DTRC Table 2. Stator blade root pitch settings (DTRC 5165).
5165) and propeller (DTRC 5128).
Stator design pitch pitch
blade pitch rev A rev B
only the erodable coating was removed
number (deg.) (deg.) (deg.)
from the blade, which suggests that the
cavitation relief holes do reduce the
erosion tendency somewhat. The
similarity between the model-scale and 1 97.3 95.3 94.3
2 99.6 97.6 94.6
full-scale erosion patterns is quite
good, 3 99.4 97.4 94.4
and shows that the flow field 4
around the propeller was well simulated 96.3 94.3 93.3
in this experiment. 5 88.3 86.3 88.3
6 82.7 84.7 85.7
7 80.2 82.2 85.2
Stator performance assessment
8 80.5 82.5 85.5
9 83.4 85.4 86.4
Once the fidelity of the model flow 10 92.2
environment had been established, 94.2 94.2
the

223
Fig. 7. Erosion pattern on model-scale propeller, Fig. 8. Cavitation on model-scale propeller with closed
cavitation relief holes open and stator blades at rev A cavitation relief holes and stator blades at rev B setting.
setting. PSD 18442-03-90, Frame 18. PSD 18439-03-90, Frame 14.

next erosion survey with the new blade DISCUSSION OF RESULTS

setting was carried out with the
cavitation relief holes sealed. Figure Once properly adjusted, the stator
8 shows the resulting reduction of sheet was highly effective at reducing the
and root cavitation on the propeller severity of blade root erosion. Small
blade surfaces directly downstream of changes in the stator blade angle of
the stator blades The erosion survey attack were found to produce large
changes in stator effectiveness. The
(see Fig. 9) was very encouraging, with
best results were found with the
almost no erosion anywhere on the
propeller, and none at the blade root. cavitation relief holes closed and the
stator blade angles set as shown in the
A final erosion survey was made at third column of Table 2.
the same stator pitch settings with the
cavitation relief holes open. Fig. 10 Since closing the cavitation relief
shows the resulting erosion pattern, holes in all full-scale propellers would
which was far worse than any noted involve considerable effort and expense,
before. Extensive erosion downstream of
erosion surveys were made with open
the cavitation relief holes suggests holes. Setting the stator blades at the
that, in this instance, they caused more best closed-hole angle caused by far the
worst erosion seen in the test. Still,
cavitation than they relieved.

224
Fig. 9. Erosion pattern of model-scale propeller, Fig. 10. Erosion pattern of model-scale propeller,
cavitation relief holes closed and stator blades at rev B cavitation relief holes open and stator blades at rev B
setting. PSD 18452-03-90, Frame 4A. setting. PSD 18454-03-90, Frame 17A.
the second stator test, with the rev A DTRC propeller 5128 at speeds up to 30
stator blade pitch settings, showed a knots. Excellent results were obtained
measurable improvement in erosion over with the cavitation-relief holes closed,
the existing conditions. while acceptable results were obtained
with the cavitation-relief holes open
CONCLUSIONS
REFERENCES
The flow-prediction codes used to
design the stator did not (and, indeed, Latas, R.M., et al. "U.S. Coast
could not) model the cavitation relief Guard 110 ft. Island Class Patrol Boats-
holes in the propeller. Clearly, there -Concept to Completion," Patrol Boats
were significant interactions between '86, A Technical Sy posium. Arlington,
the stator-induced flow field and these VA, Am. Soc. Nav. Eng.
holes. Future analyses should address
this problem.
Neely, Stephen K. and Chen, Benjamin
Y.-H., "Asymmetric Preswirl Stator
The asymmetric preswirl stator
Design for U.S. Coast Guard Island Class
(OTRO propeller 5165) proved to be very
effective at reducing the blade root Patrol Boats," DTRC Report 5HD-1335-02,
March 1991.
cavitation erosion tendency exhibited by

225
 Propeller Tip Cavitation Suppression
Using Selective Polymer Injections
G.L. Chahine, G.F. Frederick, R.D. Baseman, (Dynaflow, Inc., USA)

ABSTRACT EXPERIMENTAL SETUP AND FACILITIES

Selective injection of a drag-reduc- The tests were carried out in a
ing polymer solution into the tip vortex cavitation tunnel whose test section is
region of the blades of an 11.5 inch di- 12 feet long and 2 feet wide. Large win-
ameter propeller is seen to be effective dows on the two sides and bottom of the
in significantly delaying tip vortex test section provide excellent accessi-
cavitation. For well positioned in- bility for observation and flow visuali-
jection ports, the cavitation number at zation studies. Test section velocities
inception was decreased by as much as 35 up to 30 feet per second can be
percent. Injections of water and a mix- achieved, and the ambient pressure can
ture of water and glycerin for the same be reduced to about 2 feet of water ab-
conditions did not affect the inception solute.
characteristics of the blades. Propeller The facility is equipped with a re-
thrust and lift were not affected at the movable lid on which the strut pod and
selective polymer injection rate needed propeller were mounted to a shaft dyna-
to affect the cavitation characteristics mometer to allow measurement of thrust
of the propeller and torque. A magnetic pick up tachome-
ter measured the rotation speed of the
INTRODUCTION propeller. Pressure taps and pitot tubes
allowed measurement of pressures and ve-
The first attempts towards the re- locities in the test section. A data ac-
duction of tip vortex cavitation on quisition system enabled timely record-
lifting surfaces involved the modifica- ing of the various test parameters.
tion of the blade shape in order to un- A stroboscopic system was used for
load the blade tip. Despite the success the visualization study. The strobe ele-
of these attempts, demand for increased ment was fed signals through the exter-
speeds and performance lead to the re- nal triggering input line from the mag-
search of other means to further de- netic pick up on the propeller shaft.
crease cavitation inception (1). Various This was extremely useful since it al-
schemes have been devised and tested on lowed a quick matching of the propeller
lifting surfaces or three-dimensional rotational speed with the strobe flash-
single blades. These include: installing ing. This effectively ufroze" the pro-
an end plate, sail or bulb at the foil peller blade motion for observation and
tip, roughening the foil surface or considerably simplified cavitation in-
edge, and injecting a mass of fluid from ception detection. In addition a delay
the tip. box enabled control of the phasing of
Mass injection schemes appear to be the flashes. A manual control in the
the most attractive (1). The study of form of a knob allowed the experimenter
reference (2) showed that pure water in- to select at will the particular blade
jection at a significantly high rate was of the five in the propeller to be
effective in delaying tip vortex cavita- observed and to freeze it where needed
tion inception. Less positive results for easy observation. Photographic re-
were, however, obtained recently with cording of the visual observation was
pure water injection at lower rates done using the stroboscopic system de-
(3,4,9), while very encouraging results scribed above and both still and video
were seen with dilute polymer solutions photography.
(3,4,9).

227
Propeller and shaft modifications serve effects of water and water plus
A five-bladed propeller with a di- glycerin solution injection on the tip
ameter of 11.5 inches was selected for vortex cavitation inception character-
the tests (figure 1) because it was de- istics.
signed to have low tip vortex cavitation Inception tests were conducted on
inception characteristics and had been the propeller by fixing the water speed
previously extensively tested (8) - in the test section and the ambient
Important practical reasons also guided pressure in the channel and then chang-
its choice; mainly its shape, size and ing the propeller rotation speed (rpm)
configuration type enable one with rea- until cavitation occurred. The rpm was
sonable efforts to fit it with drilled then reduced very carefully until cavi-
injection ports without modifying the tation disappeared and the characteris-
smoothness of its surface. After photo- tics of the desinence point (rpm,
graphic and visual observations of this were recorded. The
Pambient, Vchannel)
propeller preliminary positions of in- ambient pressure in the channel was then
jection ports were selected. Only two of changed and the tests repeated until the
the five blades were modified. The two full curve was obtained. Propeller
blades were modified differently in thrust and torque were also simultane-
order to assess two injection location ously measured using the channel dyna-
possibilities. They were both fed si- mometer.
multaneously through the propeller
shaft. This had the advantage of allow- Results of the injection tests
ing simultaneous observation of the be- A series of injection port positions
havior of the flow about the unchanged on the blade tip were investigated.
propeller blades and about the two Using flow visualizations with dye
blades fitted with injection ports in injection, photography, and trial and
exactly the same water flow conditions. error tests, two sets of positions from
which Polyox injections were seen to be
Injection system effective, were selected. As sketched in
The shaft was then modified, by par- Figures 2 and 3, two injection ports
tial hollowing and adaptation, to re- using 1/16 in. tubes were embedded in
ceive the modified propeller. the blade No 1, and a single injection
Appropriate seals were positioned in the port again of the same diameter was em-
hub to create a feed chamber to receive bedded in blade No. 3. The three other
the injection fluid and prevent its blades were left unmodified and used as
leakage. This chamber was directly fed reference.
using adequate piping from a feed tank A series of injection tests was then
located outside the channel and through conducted using both modified blades.
the cover of the channel. The feed tank The importance of polymer solution con-
was a sealed chamber whose pressure centration and polymer solution injec-
could be controlled by the channel vac- tion rate was investigated. The impor-
uum pump or a compressed air source and tance of the precise injection location
by a bleed valve. A metering device was was inferred from a comparison of the
located in the injection line and al- results of blades No. 1 and No. 3 which
lowed monitoring of the flow injection were differently modified. In the fig-
rate. The propeller was also fitted with ures presented below Blade No. 5 is one
a plenum that received the fluid arriv- of the three unmodified blades and
ing through the hollow shaft and then serves as a baseline to evaluate the two
distributed it to the two blades fitted modified blades.
with injection ports. Leaks through the Figure 4 shows a summary of tests
two end faces of the propeller hub were for three different Polyox WSR301 con-
prevented by using face seals. centrations (3000, 5000 and 7000 ppm;.
An intensive series of tests were As seen on the figure, the unmodified
then conducted to: a) determine base- Blade No. 5 at a given cavitation number
lines for tip vortex cavitation incep- G, repeatability gave higher advance
tion on the propeller; b) observe dye coefficients, J,than those obtained
injection from the injection ports into with Blades No. 1 and No. 3. The
the flow; c) observe effects of polymer cavitation inception number, a, and the
injection on the tip vortex cavitation
inception characteristics; and d) ob-

228
 advance coefficient, J, are defined as the order of 35 % in the conditions that
follows: cause cavitation inception to occur can
a = 2(Pambient -Pv)/(PV2) ;
J = V/ (nD), be obtained, i.e. at the same depth the
propeller rpm and the ship advance ve-
where Pbient is the far field pressure locity can be increased by more than 17
in the channel, P. is the vapor pres- % before cavitation occurs.
sure, p is the liquid density and V is It is very apparent from the above
the inflow velocity. D is the propeller figures that the location of the injec-
diameter and n its rotation speed. tion port is very important for the
Similarly, Blade No. 3 equipped with elimination of steady tip vortex cavita-
a single injector almost always gave tion. Indeed even with twice the injec-
better results than Blade No. 1 equipped tion flow rate of Blade No. 3, Blade No.
with two injectors, even though Blade 1 achieves only about two thirds of the
No. 3 ejected half the amount of polymer improvement observed with Blade No. 3.
solution as Blade No.l. This highlights However, this conclusion seems to depend
the fact that Blade No. 3 has a better on the cavitation number. At the lower
injector location It is apparent from
. values of the cavitation number, c, this
the figure that significant effects due limited test series (especially in terms
to the polymer injection are observed, of the flow speed) appear to indicate
especially at the higher values of a. We that the injection configuration of
presently do not have enough elements to Blade No. 3 becomes much less good than
conclude about any real dependence on a, Blade No. 1 which continues to be effec-
since the experimental conditions at the tive at delaying cavitation inception.
Combined with our earlier comment about
lower values of a became polluted with the presence of a large number of rela-
air bubbles entrained in the test sec- tively large bubbles at the low cavita-
tion of the channel. Note in addition tion numbers and the inefficiency of
that increasing rpm (reducing J) in fact polymer injection in preventing the
amounts to a reduction of the actual growth of these bubbles, it may be that
cavitation number. This is not apparent the scheme on Blade No. 1, by injecting
only because of the particular defini- the polymers not exactly in the core re-
tion of a which does not reflect the lo- gion but on its periphery, may be ef-
cal conditions within the tip vortex of ficient in delaying the capture of the
the propeller. In order to minimize the large bubbles by the vortex. These com-
effects of gaseous bubble entrainment on ments are, however, only of a specula-
the tip vortex dynamics, most tests were tive nature at this point and need to be
conducted at a low channel test section confirmed in future work.
velocity, V=1° ft/s.
Figure 5 shows the influence of the Influence of the injection rate
concentrations of Polyox WSR301 injected The influence of the injection rate
on the cavitation number at inception .
on the effectiveness of polymer injec-
A range of concentration between 1000 tion in delaying tip vortex cavitation
and 7000 ppm is covered. This figure is is illustrated in Figure 7. In order to
a 'cut through the curves (cavitation reduce testing time, a channel speed and
number, a, versus advance coefficient, an ambient pressure were first selected.
J) at a constant value of J, here se- Then for each injection flow rate, the
lected to be 0.9 .
Also shown on the rpm of the propeller -- and therefore
same figureare the results obtained the advance coefficient, J -- at which
with water injection from the same cavitation inception would occur was
blades. It appears from the curves that noted. When the injection rate is zero
3000 ppm is close to an optimum injec- the J value is that of the unmodified
tion rate for the injection positions propeller For non zero values of the
.

and the injection port sizes studied. injection rate a reduction in the value
Figure 6 presents percentages of im- of J (an increase in rpm) is needed for
provement -- i.e. decrease -- in the cavitation inception to occur. Figure 7
values of a needed to obtain cavitation shows the results obtained for two val-
for the 3000 ppm Polyox solution case. ues of the cavitation number, a. Both
This figure shows that improvement of values of a -- 12.4 and 17.6 -- were

229
selected in a range where gaseous bubble fect on the cavitation inception of the
influence was minimal. blades. In fact a small effect in the
The observed differences between the opposite direction (enhanced or earlier
values of J at zero injection rate indi- cavitation) can be observed on blade No.
cate the amount of data uncertainty 1. This is probably due to the injection
(e.g. 0.95 and 0.955 for a=12.4, indi- with the water of some bubble nuclei.
cate an uncertainty of the order of 1%).
This could also be interpreted as due to
A significant amount of drop in the cavitation of the injected water jet.
During injection of polymer solutions
value of J can already be observed at an this secondary effect is overshadowed by
injection rate for all ports of $60 opposing cavitation inception
the
cm3/min (about $20 cm3/min/port$) . At
retardation effect (pressure rise in the
that injection rate J decreased by 3%. vortex viscous core) due to the presence
This performance improvement continues of the viscoelastic fluid.
to increase with the injection rate. At Also shown on the figure is the in-
the relatively low cavitation number, a jection of a highly viscous mixture of
12.4, Blade No. 3 appears to exhibit a 50 % water and 50 % glycerin -- viscos-
minimum value of the advance coefficient ity of the order of 10 centipoises,
at inception for the total flow rate in close to that of a 3000 ppm concentra-
the range of 230-240 cm3/min (75-80 tion Polyox solution. Here again little
cm3/min per port). At this minimum, the effect of the viscous solution injection
improvement in J is about 8%. On Blade is seen. If any, the effect seems again
No. 1 and for both modified blades at a to be negative on Blade No. 1 showing
= 17.6 there is no distinct maximum for a given advance coefficient, J, an
cavitation reduction point. However, the increase in the cavitation number at tip
amount of cavitation retardation appears vortex cavitation inception.
to attain a saturation range when the The results presented above strongly
injection rate exceed about $310 cm3/min indicate that the viscoelastic proper-
for all ports. These results indicate ties of the polymer solutions are re-
the existence of an optimal injection sponsible for the observed significant
rate .
delay in the cavitation inception of the
tip vortex cavitation on the tested pro-
Injection of water and alvcerin peller.
In order to check if the positive
results observed earlier were due to Influence on thrust and torque
mass injection only, or because the in- The positive results observed in
jected mass was very viscous, or if di- these tests would not be of interest if
rectly related to the viscoelastic char- they were accompanied by any detrimental
acter of the drag reducing polymers, we effects such as loss of performance, due
tested the influence on the propeller to increased drag or reduced thrust or
tip vortex cavitation of the injection torque. This is the case for a homoge-
of pure water and of a 50 % mixture of neous, uniform solution of polymers
water and glycerin. which induces improvement in cavitation
Figure 8 shows the results of two inception characteristics but is also
series of tests. In the first series a accompanied by an undesirable loss of
solution of 3000 ppm of Polyox WSR301 is thrust. To investigate this, thrust and
injected from Blade No.1 equipped with torque of the propeller were measured
two injection ports, and from Blade No.3 simultaneously during the tests in the
equipped with a single injection port. presence and absence of injection. The
In the second series of tests shown in results of these measurements are shown
the figure pure water is injected from in Figures 9 .

Blade No.1 and Blade No.3 at the same

injection rate (about 70 The solid lines were obtained in ab-
cmi/min/injector) and the new charac- sence of any fluid injection, while the
teristics of these blades (cavitation unconnected symbols indicate measure-
versus advance coefficient, ments made when a solution of 2875 ppm
number, a,
of Polyox WSR301 was injected from both
J) are compared with those of Blade No.
Blade No.1 equipped with two injection
5. The injection of the same flow rate equipped with a
ports and Blade No.3
of pure water is seen to have little ef-

230
 single injection port. The injection ACKNOWLEDGMENT
rate from the three ports totaled 250
cm3/min. Figure 9 indicates clearly that The work described in this paper was
no measurable influence of the polymer conducted in execution of an SBIR con-
injection is obtained on either thrust tract sponsored by the Naval Sea Systems
or torque. Details of the observations Command, Washington DC.
(9) show that variations in thrust and
torque due to changes in the ambient
pressure were easily measurable while REFERENCES
the variations due to mass injection of
the polymer solution could not be de- Platzer, G. P. and Souders, W.
tected (Figure 10). G., "Tip Vortex Cavitation Delay with
These results allow us to conclude Application to Marine Lifting Surfaces:
that the tip vortex cavitation inception A Literature Survey," DTNSRDC-79/051,
improvements observed with the injection 1979.
of polymer solutions are not accompanied Souders, W. G. and Platzer, G.
by any measurable decrease in propeller P., "Tip Vortex Cavitation
performance. Characteristics and Delay of Inception
on a Three-Dimensional Hydrofoil,"
CONCLUSIONS DTNSRDC-81/007, 1981.
Fruman, D.H. and Aflalo, S.S.
This study addressed the feasibility "Tip Cavitation Inhibition by Drag-
of the concept of tip vortex cavitation Reducing Polymer Solutions," ASME
suppression by selective polymer injec- Journal of fluids Engineering, 1990.
tions. The following conclusions can be Inge, C. and Bark, G., "Tip
drawn from the results: Vortex Cavitation in Water and in Dilute
Injection of solutions of a drag Polymer Solutions," Royal Institute of
reducing polymer into the tip vortex Technology Technical Report TRITA-Mek
area of the blade was effective in de- 83-12, 1983.
laying tip vortex cavitation inception. Hoyt, J., "Vortex Cavitation in
Some combinations of polymer con- Dilute Polymer Solutions, "Cavitation
centration / injection rate (in the and Multiphase Flow Forum, Fort Collins,
range 3000 ppm and 80 cm3/min/port) CO, June 1978.
showed a decrease in the cavitation num- Chahine, G. L. and Fruman, D. F.,
ber at inception as large as 35%. "Dilute Polymer Solution Effects on
Injections of pure water and a 50% Bubble Growth and Collapse," Physics of
water-glycerin viscous mixture did not Fluids 22, 7, pp. 1406-1407, 1979.
show any improvement in the tip vortex Chahine, G. L., "Experimental and
cavitation characteristics of the pro- Asymptotic Study of Nonspherical Bubble
peller. Collapse," Proceedings IUTAM Symposium
The location of the injection port on Mechanics and Physics of Bubbles in
was seen to be extremely important. Out Liquids, Pasadena, California, June
of the two configurations extensively 1981, also in "Applied Scientific
studied, the better one achieved im- Research," 38, pp. 187-197, 1982.
provements in the cavitation inception Barr, R.A., Walker, K.W., and
characteristics that were at least 50% Lindenmuth, W.T., "The design and
better than the second one, even though Testing of a Series of Six
the less performing configuration had Propellers...," Hydronautics, Inc.
injection rates twice as large. Technical report 455-2, April 1966
The selective local injection of (Confidential)
the polymer solutions did not adversely Chahine, G.L., Frederick, G.S.,
affect either propeller torque or and Bateman, R.D., "Propeller Tip Vortex
thrust. Cavitation Suppression Using Selective
Selective injection makes the con- Polymer Injections," DYNAFLOW, INC.
cept practical by reducing the weight of Technical Report 91001-1, August 1991.
the polymers needed for injection and by
not effecting the propeller performance.

231
 0.80 0.85 0.90 0.95 1.00 1.05 1.10
Advance Coefficient, J

F. 4: Influence of Various Polyox Solution Injection on Tip Cavitatio

Inception. V=10ft/s
Fig. 1 Picture of Propeller used in the injection tests.

17

8
11

Rg. 2: Sketch of tube embedded

in the propeller blade. 1.10
"000 3600 5600 7000
Polymer Solution Concentratkon, ppm

Fig. 5: Influence of the concentration of injected Polyox solueons co tip vortex

cavitatbn inception. V=10 fVs. Composite picaire of all runs - Comparison
with water and Water/Glycerin solution injection.

BLADE 11 BLADE 43
PRESSURE Siti£ PRESSURE SICE

Cm pressure Blade #1, 2 Injection Ports Blade #3,1 Injection Port

Occotored on olds at edge
loacttog edge
=El 0.048 0333710}1 0.048.
0.9 1 1.1 12
Advance Coefficient, J

Fig. 3: SKETCH OF THE PObrriONS OF

INJECTION PORTS ON THE TWO BLADES. Fig. 6: Improvement in the Cavitation Number at Inception - Correspon
to Figure 8. V 10 ft/s, Injection Rate =70 cc/min/injector.
232
 0.96 20
.,
ll \\ \
co
0.94- \\\\
..
-..--,.-

. 0.926,
8
c o 's
' Sigma .-. 12.39
s it-,....
L Blade #1
50% glycerin solutio
and Water I niection
-171 0.9-
Sigma = 17.6
\ . / Blades # 1 and #3
.--/-
C.) !Blade t3
--- ...... Blade #31
co

-::,
o
0.86- 1Blade #1 4
Polyox 2875 ppl ---- -
0.85 0.9 05 1.05
0.84 Advance Coefficient J at Inception
o 50 160 150 260 250 360 3,.40 400
Injection rate to all three oorts,ccfmn
Fig. 8: Comparison of Cavitation Inception Curves at blade #5 with
Fig. 7: Influence of Itte rate of injection of Polyox WSR 301 on the no injection and blades #3 (1 injection port) and #1(2 injection porta)
aotvance coefficient J at tip vortex cavitation inception. V=10 ft/s. with Pure Water, Polyox and Glycerine solutions.
Injection rate = 100 cc/mn/port.

60- 140
O Polymer Injeceon. Pa=0.18 aun

50, A Pdyrner Injectuyi. P,-0.79 aum .120

No Injecton, atrn
No Injecnon, Pa--0.79 atrn

60 0
10-
ao

o
450 500 20
5 600 550 660 650
PROPELLER ROTATION SPEED, rpm PROPELLER ROTATION SPEED, rpm

Fig. 9: Comparison of thrust and torque on propeller with and without Fig. 10: Comparison of thrust and torque on pcpeiter with and witout
polymer injection at sever-al ambient pressure (0.18atm<Parnb<0.79atr1). pdyrner injeztion at high ambient pressure in the channel. Potyoz WSR 301
Polyox WSR 301-2875 ppm. Injection Rate 250 cc/min, 3 injectors, V=10ft/s. 2675 ppm. Injection Rate = 250 crJmin, 3 injectors, V=10 ft/s.

233
 SESSION III-A

NEVV FACILITIES, TECHNIQUES,

ICE TESTING, AND VALIDA'TION-I
 The New Facilities, Techniques, Ice Testing,
and Validation Committee Report
J. Zseleczky (The Naval Academy, USA)
Chairman: J. Zseleczky
Committee Members: Contributors:
P. Johnson (Texas A&M) P. Johnson (Texas A&M)
R. Etter (DTRC) A. Goodman (Hydronautics)
M. Murray (NRC, IMD) G. Stensgaard (B.C. Research)
R. Barr (Hydronautics) S. Jones (NRC, IMD)
R. Ettema (U. of Iowa) C. Hsiung (U. of Nova Scotia)
D. Spencer (NRC, IMD) J. Duncan (U. of Maryland)
J. Tatinclaux (CRREL) J. Zseleczky (USNA)
B. Johnson (USNA)

The "'Mew Facilities and Techniques" Offshore Technology Research Center

Committee was expanded for this year's Wave Basin (College Station, Texas) -
conference to include Ice Testing and 150x100x19 ft (45.7x30.5x5.8 m); details
Validation. In past conferences, "Ice in "Contributions" section and in session
Testing" work has been represented in a paper.
seperate session. This year however, the
schedule of the ATTC overlapped with two BC Research Wave Basin (Vancouver,
ice-related conferences that were expected B.C.) - 100x87x8 ft (30.5x26.5x2.4 m);
to keep most ice-tankery experts from details in "Contributions" section.
participating. In spite of the
competition, we are fortunate to have DTRC Large Cavitation Channel (Memphis,
three papers on ice testing from IMD of TN) - 239x65 ft (73x20 m); details in
Newfoundland. The subject "Validation" session paper.
has also been added to the committee to
ensure that all subjects covered by the Tech. U. of Nova Scotia Towing Tank
ITTC are represented by the ATTC. (Halifax, N.S.) - 3.2x3.2x92 ft (1x1x30
"Validation" refers to the validation of m); details in "Contributions" section.
theoretical and numerical models of marine
hydrodynamics. U. of Maryland Towing Tank (College
Park, MD) - 3.5x4x48 ft (1.1x1.2x15 m);
A brief summary of new facilities details in "Contributions" section.
and major equipment put into service since
the last ATTC is listed below. Detailed U. of Sao Paulo Towing Tank (Sao Paulo,
descriptions of these facilites can be Brazil) - 3.2x3.2x46 ft (1x1x14 m);
found in the "Written Contributions" details in session paper.
section of this report, and/or in the
published papers for the session. Some of
the written contributions have been edited
to fit within the available space. Maior New Equipment:

IMD Sting Support for Dynamic Testing -

Capabilities: Model length - 6 m,
SUMMARY OF FACILITIES COMMISSIONED Amplitude - 1 m, Incidence - 30 deg.
SINCE THE 22nd ATTC
DTRC Sting Support for Rotating Arm
Capabilities: Model length - 5 m,
New Tanks: Incidence - 35 deg.
Institute for Marine Dynamics Offshore Davidson Lab Coning Motion Apparatus
Engineering Basin (St. John's, Capabilities: Incidence - 20 deg.
Newfoundland) - 245x105x12 ft (75x32x3.5
m); commissioned in 1990, details in DTRC Right Angle Dynomometer Drive
session paper. System - Capabilities: Rpm - 3000, Thrust
674 lb, Torque - 1328 in-lb.

237
 DTRC Wind Turbine - Two blade Aircraft Model Testing In Water
propeller, 24' diameter, produces 20 fps
wind in test area of MASK facility. A comprehensive series of model
experiments were conducted in the Tracor
U.S. Naval Academy Wind System - Bank Hydronautics Ship Model Basin (HSMB) to
of squirrel cage fans, produce 30 fps wind determine the individual subsonic static
15' downstream of fans. longitudinal as well as static and dynamic
lateral stability derivatives of a 0.19 -
scale model of the X-31A aircraft with an
F-18 canopy operating at a Reynolds Number
WRITTEN CONTRIBUTIONS FROM FACILITIES of 1.35 x 10 and angles of attack up to
70 degrees.

Offshore Engineering, Deep Water Multi- The tests were conducted using the
Directional Wave Basin, OTRC Large Amplitude Horizontal Planar Motion
P. Johnson Mechanism (LAHPMM) and a 0.19 scale model
of the X-31A aricraft. Special model
The Offshore Technology Research support systems and data acquisition and
Center (OTRC) deep water, multi- processing systems were developed for the
directional wave basin, in College test program. Stability derivitaves were
Station, Texas, was commissioned in March, determined for pure rolling and yawing.
1991. The basin is 150 feet (45.7 m) x Static longitudinal and lateral
100 feet (30.5 m) wide x 19 feet (5.8 m) coefficients determined from the tests
deep. A 15 foot (4.6 m) x 30 foot (9.1 m) were compared with preliminary results
pit, located in the center of the basin, obtained from tests of the X-31A in the
has a variable depth from 19 feet (5.8 m) NASA Langley Full-Scale Wind Tunnel and
to 55 feet (16.7 m). Waves are generated earlier tests in the HSMB. In addition,
by 48 hydraulically driven, independently the dynamic lateral-stability derivatives
controlled, partial depth, hinged flaps. were presented as a function of angle of
Regular and random, long-crested and attack for several canard angles and
short-crested seas can be made. The reduced frequencies. Presented for the
maximum regular wave height is 34 inches first time, for a complete aircraft
(0.80 m). Data acquisition capability is configuration, were the pure rolling, pure
presently a maximum of 64 analog channels. yawing and pure sideslipping stability
Localized current generating capability is derivatives for a range of angles of
presently being added. Localized wind attack up to 70 degrees.
generating capability is planned.
Model and Sting-Support System for DTRC
The wave basin was constructed to Rotating Arm Facility
support the academic research program of
the ORTC. This program is being carried Hydronautics Research, Inc.
out by Principal Investigators and designed, built and tested a sting-support
graduate students at Texas A&M University system and 17-foot model of the SSN688
in College Station, the University of submarine for use with the DTRC Rotating
Tesas at Austin and selected outreach Arm facility. The model and support
institutions. The OTRC research program system were evaluated by performing static
is directed towards the engineering stability tests in both the vertical and
challanges associated with platforms in horizontal planes. The results were
"deep" water, greater than 3000 feet (915 compared by DTRC with unpublished DTRC
In addition, the wave basin is
horizontal plane results for the SSN688
m).
available to OTRC industry affiliates and with propeller off. The comparisons
others for industrial research on a showed excellent agreement between the
commercial basis. sting-supported and strut-supported model
data. In addition, the comparison of
results for rudder deflections of +/- 35
degrees indicated no "vortex" bursting
Hydronautics Research, Inc. phenomenon alluded to by some experts.
A. Goodman
Acoustic Measurements at the Hydronautics
Hydronautics Research, Inc. has been Ship Model Basin
involved in several areas covered by this
Summaries of three projects DTRC Code 1922 has been interested
committee.
have been submitted: in finding an appropriate facility for
making future acoustic measurements.

238
 Measurements documented in the 21st ATTC 2). Accurate simulations of shallow,
proceedings (1986 showed that the HSMB had intermediate and deep water waves are
20 to 30 db less noise than the old DTMB generated using the WM15 wavemaker, which
carriage. In 1989, further investigations is on long-term loan from the Hydraulics
were conducted to study the acoustical Laboratory of the National Research
signatures of the facility. Two types of Council of Canada. The dimensions of the
measurements were made: a "fly-by" waveboard are 50x6 ft (15x1.8 m). The
measurement, where the hydrophone was board may be operated in piston, flapper
stationary and the carriage moved toward or combination mode. Wave generation,
and past the hydrophone, and a "towed" experiment control, data acquisition,
measurement, where the hydrophone was analysis and presentation of results are
towed from a strut in the water. performed using a dedicated VAXstation
3100 with the GEDAP software package
The fly-by measurements were found (developed at the NRCC Hydraulics
to be fully within the DTRC facility noise Laboratory). GEDAP offers such useful
criteria at a carriage speed of 5 knots features as nonlinear wave generation,
and partially within the criteria at a convenient sensor calibration routines,
carriage speed of 10 knots. From the real-time monitoring of sensor outputs,
towed measurements, it was discovered that and an extensive capability for the
the towed hydrophone had unanticipated frequency-domain and time-domain analysis
high frequency noise cympyaents. Splash and presentation of wave records and wave-
noise and electrical noise were determined related time series data.
to be significant factors in the noise
data. It was postulated that with a The Ocean Engineering Centre is
proper strut, the measurement of the 10 slowly developing a numerical
knot signature may have been within the hydrodynamics capability. Working in
DTRC facility noise criteria. Time cooperation with the NRC's Institute for
constraints did not allow further Marine Dynamics, analytical methods are
investigations. being applied to diverse planing and
displacement hulls. The techinque being
References: used is a panel method which features a
non-linear free-surface boundary.
Chrisler, V. and Lasky, M., "Levels and Although the analytical methods are
Frequency Distribution of Noise in the currently in the developmental stage,
Test Basins of the David W. Taylor Model correlation between model and experimental
Basin," DTMB Report 643, June 1948. results is gratifying. Several papers are
slated to be written as the numerical
Goodman, A., Silverstein, B. and hydrodynamics project progresses.
Gottwald, J., "Acoustic Noise Levels in
the Hydronautics Ship Model Basin," 21st
ATTC, Washington D.C., August 1986. National Research Council, Institute for
Marine Dynamics
S. Jones
British Columbia Research Corporation
G. Stensgaard Our ice tank research over the last three
years has concentrated on:-
Although still under development,
the B.C. Research Ocean Engineering Development of Controllable Density
Centre's wave basin may now be considered (CD) model ice.
a serious research facility for ocean and
coastal engineering studies. Within the Propeller-ice interaction
past year we have conducted major
breakwater design studies for the ports of Full-scale measurement of ship
Seward, Alaska and Seattle, Washington. performance in ice.
Thesis research has also been carried out
by coastal engineering graduate students Full-scale/model-scale correlation.
from the University of British Columbia.
The basin also holds promise for some The new controllable density (CD)
specialized ship research in the areas of model ice was introduced at IMD in order
seakeeping and manoeuvering. to produce model ice over a wide range of
density, and hence give the correct
The wave basin measures 100x87 ft buoyancy forces on a ship model.
(30.5x26.5 m) and has a maximum operating It has
been very successful, and in addition to
depth of 8 ft (2.4 m) (see Figures 1 and correct buoyancy, the fracture and elastic

239
properties of the ice are improved. Also, conical structure in ice, particularly in
the whitish appearance of the ice gives multi-year ice ridges, was completed for
better contrast in video records of the Conoco Inc. This necessitated the
tests. Tests have been run in level, and development of techniques for
ridged, ice and results are given in the manufacturing the ridges, and for
paper by D.Spencer. measuring their mechanical properties.

A major project was initiated in In collaboration with Memorial

1991 on propeller-iceinteraction in University, a small but interesting test
collaboration with the Canadian Coast series was conducted using an air cushion
Guard and the Finnish Board of Navigation. vehicle to break ice. For these tests the
The NRC is conducting the research on model was free of the carriage except for
behalf of the CCG, and VTT is conducting a tow line and supported on its own air
the Finnish part of the project. An cushion.
extensive set of model tests are in
progress at IMD to measure the non-contact Fleet Technology completed a large
load on a ducted and non-ducted propeller. test program for the Canadian Coast Guard
This is the load due to ice being in close on a model of the "CCGS Bernier", which
proximity, but not actually touching, the was possibly to have a new bow fitted as
propeller. In addition, some high speed part of a mid-life refit. The comparative
mechanical tests on ice will be conducted test program involved the original bow and
to determine the fracture and crushing three alternatives, including a Thyssen-
properties of ice at strain-rates Waas bow. These tests also involved a
comparable to those found in ice-propeller considerable amount of open-water work.
interaction.
Related research on ice friction and
In February 1991, a set of full- ice mechanics was also continued.
scale trials of the Canadian Coast Guard
ship "Sir John Franklin" was conducted off
the coast of Newfoundland in ice and open Centre for Marine Vessel Design and
water. The trials were very successful, Research Technical University of
giving consistent results for the ship's Nova Scotia
performance in different ice and snow C. Hsiung
conditions. The ice conditions were well
documented, over 30 cantilever beams were The Centre for Marine Vessel Design
broken to measure strength, and many and Research and the Naval Architecture
thickness measurements were made. The Program at the Technical University of
deflection of the ice sheet as the ship Nova Scotia share the operation of a small
passed was also measured for the first model towing tank for teaching as well as
time. Three turning circles and a pre- research purposes. The tank dimensions
sawn test were successfully made, as well are 1x1x30 m.
as over 20 speed/power and bollard pull
tests. A paper will be presented at the A computer controlled wave-making
SNAME Annual Meeting in November 1992. system (as shown in Figure 3) has been
installed in the tank. It has two wave-
A set of tests designed to measure makers, one at each end. Either one can
the repeatability of resistance act as a wave-maker or a wave-absorber.
measurements of ships in ice was carried The system can make progressive waves or
out in the IMD ice tank. The object was standing waves, as well as regular waves
determine how accurate, and how or irregular waves. The maximum wave
to
reproducible, are the values obtained from height is about 0.3 m.
a test program. The results of this work
are given in a paper by S.Newbury. Recently, the drive-control system
of the towing carriage has been completely
Model-scale/full-scale correlation redesigned and upgraded with a fully-
automated carriage control system.
for ships in ice was continued, with tests
being conducted on several ships for which Currently, speeds up to 4.2 m/s are being
we have reliable full-scale data. In the achieved, but the final objective is to
course of this work, a standard method of attain a top speed of 5.0 m/s. The tank
ice resistance tests was has a useable rail length of 25 m.
conducting
developed, now in use at IMD, and is
described in a paper by D.Spencer. This small towing tank has been
utilized for various model tests, for
An extensive series of tests of a example: seakeeping test for a SWATH ship;

240
 determining wave resistance of a half- Underway inclining experiments of a
model with the longitudinal wave-cut planing boat model (details presented in
method; fishing-boat powering test; rigid report of High Speed Craft committee)
hull inflatable performance test; and more
recently a planing hull series test. Static and dynamic planar motion tests
Meaningful results have been obtained. with different water depths and model
scale ratios, to determine hydrodynamic
coefficients for moored ships (details
University of Maryland Wave Tank presented in report of Steering and
J. Duncan Manoeuvering committee)
A new wave tank is under Dynamic testing of racing oar blade
construction at the University of propulsive characteristics (a
Maryland. The dimensions of the tank are competitive oarsman actually rowed an
48x4x3.5 ft. The tank is raised above the instrumented rowing box that was suspended
floor, has glass walls and has a stainless from the towing carriage; different sweep
steel floor. Regular and random waves can oar shapes were tested and evaluated)
be generated using a vertical plunger
wavemaker located at one end of the tank.
The tank is equipped with a towing
carriage that rides on a film of oil
using custom built bearings and ways. The
carriage is capable of speeds of up to 4
fps. The tank is also equipped with a
system for towing a submerged hydrofoil
that spans the width of the tank. Tracks
for supporting and towing the foil are
built into the walls of the tank. The
wavemaker, towing carriage and towed foil
are all driven by computer controlled
servo motors.

U.S. Naval Academy Hydromechanics

Laboratory
J. Zseleczky
The U.S. Naval Academy
Hydromechanics Laboratory is currently
developing a wind generation system for
its 380' towing tank. The system will
consist of four centrifugal "squirrel
cage" fan units, each of which is 8 feet
wide. The fans can be positioned in-line
to span the tank (3 fans x 8' = 26'), or
arranged in pairs to produce a higher wind
density over a narrower width. The system
has been designed to produce wind speeds
of 30 fps, 15 feet downstream of the fans.
The fans will be used in the summer of
1992 to investigate the intact and damaged
stability of Navy ships is a seaway, with
the effects of wind accounted for.

Since the last ATTC, the

Hydromechanics Laboratory has developed
several new techniques for model testing.
Some of these new techniques are:

Side-by-side testing of models in

regular and random head seas (details
presented in report of High Speed Craft
committee)

241
 Removable (End) Removable (End)
wave Absorber Wave Absorber
Modules Modules

Ramp

I
IL
,-,

ó 1

Model
1 I..

tm
i
'

e.,, i 11

0 ..
rl

>e lio
I wave crest
e
Z
'11

WM15 Wavemaker

L. 15m
26.5m

15mWide Unidirectional
Wave Basin
(depth up to 2.4m)

Fig. 1 B.C. Research Ocean Engineering Wave Basin

Fig. 2 Wavemaker/absorber system at Technical U. of Nova Scotia

242
 The Large Cavitation Channel
R.J. Etter, M.B. Wilson
(Carderock Division, Naval Surface Warfare Center, USA)

ABS'TRACT hydrodynamic models. The need for model tests of thp

complete hull-propulsor-appendage system has become more
The Large Cavitation Channel (LCC) of the David Taylor important as modern ship designs have evolved. New ship
Model Basin, CDNSWC, is a large variable pressure water designs have resulted in improved speed, efficiency, and
tunnel recently put into operation in Memphis, Tennessee. quieting and require proportionately improved model test
This facility incorporates many of the latest technological facilities and techniques to verify the important features
concepts in the design of hydrodynamic and hydroacoustic test before full scale construction. This also applies to ship
facilities. This paper discusses a number of the key technical upgrades and design alterations for existing ship classes. Of
and design issues that had to be addressed first in the basic course, the same facility can be a great asset for a variety of
specifications and then in the final features as constructed. basic hydromechanics research and engine,ering development
Issues include: size, shape, speed, pressure range, turbulence studies as well.
management, turning vanes, contraction and diffuser, main
pump impeller, hydroacoustic considerations of noise The LCC test section size, shape, and flow speed were chosen
sources, quieting, and noise measurement; and operational such that models large enough to minimize adverse viscous
features involving model size, powering, and measurement of scaling effects due to Reynolds number mismatch could be
model propulsor thrust and torque. tested with small wall effects (due to flow blockage).
Hydrodynamic goals included uniform velocity profile,
INTRODUCTION minimal background turbulence level and well- controlled test
section speed and pressure. From the start, the LCC was also
The Large Cavitation Channel (LCC) is the newest and one intended to be a hydroacoustic test facility. Therefore, low
of the most technologically-advanced large cavitation tunnels background ambient noise levels and accommodation of
in the world. Plans for this facility began about ten years extensive acomstic instrumentation were required.
ago. Motivation for the construction of the tunnel was the
need for the U.S. Navy to have a variable-pressure, closed An earlier outline of t.he LCC hydrodynamic/hydroacoustic
circuit experimental facility large enough to conduct model features, process control system, and data acquisition system
tests with scaled, full length, completely appended has been presented by Morgan and Etter (1).

239 ft
Flow
Main Test Straightener
Elbow Test Top Section
Diffuser Contraction Elbow No. 4
No. 1

r. "
Acoustic 65 ft
Trough
V Honeycomb
,r7
J.

Elbow No. 2 Main Pump Bottom Leg Diffuser Elbow No 3

Figure 1 - Sketch of the Large Cavitation Channel (LCC)

243
INITIAL SIZING AND FEATURES portion of the hull in the flow, would be much less than for a
fully submerged body of the same scale ratio. A test section
Size, Speed, Pressure length of 40.8 ft (12.4 m) was chosen to accommodate the
From the point of view of cavitation and noise performance anticipated large hull models.
the propulsor is the most critical component of a ship or
submarine design, and this provided the starting point for Cascade Solidity
facility design requirements. Extensive correlation experience spacing/chord = 0.5
has led the David Taylor Model Basin to target a standard
Reynolds number goal for propeller cavitation testing of
around 5 million based on the propulsor rotor chord and
resultant velocity at the 0.7 radius

Rn0.7 = (VAc().7/v)[1.0 + (370)2'1/2

where VA = inflow velocity, co3 = chord length at 0.7R, J =

advance coefficient (VAinD), n = propeller rotational speed
(rps), D = propeller diameter. It is also necessary to keep in
mind the minimum propeller size for which models of
adequate accuracy for cavitation testing can be built
economically. Propeller model material strength is another
consideration in the choice of test section speed. Full scale
speeds result in full scale stresses in a geometrically scaled
model. In most cases this means tests much above full scale
speed are not advisable. In light of these considerations, it
was decided that typical propeller rotors of diameter 12 to 18 FLOW
in. (30 to 46 cm) and a test section maximum flow velocity
of 30 to 35 knots (50 to 60 ft/s, 15 to 18 m/s) would suffice.

The propeller diameters selected above result in ship and

submarine models of about 20 to 40 ft (6 to 12 m) in length, Figure 2- Typical LCC Elbow Turning Vane Cascade
or scale ratios of approximately 25:1 to 12:1. These scale
ratios are consistent with models which can be used for
resistance and powering tests in the large basins at the David In view of the desired and anticipate,d range of cavitation
Taylor Model Basin. However, model hulls designed for numbers, together with the selected range of test section flow
towing tank tests are usually not structurally adequate to be velocities, the range of test section static pressure was chosen
tested in 30 to 35 knot flows because they are intended for to be 0.5 to 60 psia (3.5 to 415 kPa, 0.03 to 4 atmospheres),
much lower Froude-scaled speeds. High Reynolds number specified to apply at a point at the top of the downstream end
and high test section flow speed are important for propulsor of the test section.
performance, cavitation inception, and hydroacoustic noise
measurements. In light of the scaling issues with all these To minimize most potential problems with material
factors, it was decided that the LCC be designed to operate in corrosion and to simplify maintenance, the complete shell
the closed jet mode (with no free surface). The option of a and all internal wetted parts of the water tunnel circuit have
second LCC test section, with a free surface, was dropped been fabricated of stainless steel. External stiffeners and
because of cost and in order to concentrate on the design of a supports welded to the shell are carbon steel. The structural
top grade closed jet section. design of the LCC was made in accordance with the American
Society of Mechanical Engineers (ASME) Boiler and Pressure
Because of the interest in testing submerged bodies as well as Vessel Code. All the shell butt welds received 100%
surface ship models, the basic LCC test section was chosen radiographic examination. Load requirements included
to be square. If only surface ship models are considered, the Seismic Zone III earthquake loads for the Memphis area.
test section area could have been significantly reduced by Extensive finite element analyses of the structural design
choosing a height less than the width. Surface ship hulls were conducted by the prime contractor, CBI Na-Con, Inc. of
would normally be mounted on the ceiling of the test section. Oakbrook, Illinois for many of the major components of the
The basic LCC cross section dimensions of 10 x 10 ft (3.05 channel. Fabrication of the entire LCC structure was carried
x 3.05 m)were chosen to accept a submarine model or body out by CBI Na-Con.
of revolution model of about 3.5 ft (1.07 m) in diameter, in
order to keep the cross sectional area blockage less than 10%. Contraction and Main Diffuser
Typically, the blockage for a surface ship model mounted on Details of the hydrodynamic design of the contraction and
the ceiling of the test section, with only the below-waterline diffuser and other aspects of the LCC circuit are discussed by

244
 Pressure Side liters) of water. The upper leg of the tunnel circuit has square
Anechoic Treamaent
T.E. C.enterline section segments, while most of the lower leg has circular
of Elbow sections upstream and downstream of the axial flow main
pump. Transition from square to round section shape
upstream of the pump occurs in the down-flow vertical leg.
Downstream of the pump, there is transition from round to
square just before the elbow.

Flow Quality
Design goals for the test section flow included a velocity
profile which is uniform outside the boundary layer within ±
1% of the mean velocity and a low RMS turbulence intensity
level of 0.1% of the mean velocity. These goals required
careful design of the shape of the non-symmetric contraction
L.E. and diffuser and selection of the flow straighteners and
honeycomb ahead of the contraction. These components and
Figure 3 - Typical Turning Vane Foil Section with NACA the use of radii in all the corners of the flow passages helped
Four-Digit Section Shape assure the elimination of secondary flow cells and
concentrated patches of vorticity in the test section.

Wetzel and Arndt (2). The tunnel test section contraction

VERTICAL SURVEY HORIZONTAL SURVEY
ratio is an important feature affecting the overall facility size. 100
Typical contraction ratios for existing water tunnels range ob 00
from about 2:1 to 16:1. Hydrodynamically, a large
contraction ratio is desirable because it results in lower
velocities in all parts of the tunnel circuit other than the test 80
section. This provides lower extraneous noise generation and
lower facility head loss. The main advantage of a large
contraction ratio is the rapid acceleration of the flow into the 60
test section, resulting in reduced boundary layer thicicness at
the walls and a uniform velocity profile in the test section. ± 1% -1
Large contraction ratios do have some disadvantages. They 40
result in physically larger elements of the tunnel structure o Measured
away from the test section. This means larger overall Calculated
dimensions, more structural weight, and higher cost. All
factors taken into account, a contraction ratio 6:1 was chosen 20 11

for the LCC.

o
Historically, water tunnel circuits have used symmetric O
contraction and diffuser designs, generally modelled after wind 0.8 0.9 1.0 08 0.9 1.0
tunnels. However, the special requirements of water tunnels UlUc
have led to a choice of a flat top, nonsymmetric arrangement U/Uc
for the LCC. The top of the contraction, the test section, and Figure 4- Measured and Calculated Results for Longitudinal
the diffuser are all in the same horizontal plane. This allows Component Velocity Distributions in LCC Test
operation over a large range of velocities with lower pressure Section Model. Data from SAFHL Wind Tunnel
in the test section without encountering vapor pressure in the Model
adjacent parts of the circuit. It also minimizes the volume of
water that must be drained from the tunnel upper leg when it The turning vane cascades in each of the four 90-degree
comes time to work on model installation or removal. These elbows of the circuit are made up of segments of airfoil-
are factors which become significant for a very large facility. shaped vanes. Figure 2 shows a typical turning vane cascade,
with a solidity ratio = spacing/chord length = 0.5. A sketch
The LCC main diffuser in the upper leg has the flat top upper of the geometry of a single turning vane is given in Figure 3.
boundary, noted above, with the lower boundary sloped The section shape is a NACA four digit thickness
downward at 5 degrees and each of the two side walls angle distribution, with high camber. Each vane segment is made
outward at 2.6 degrees. of cast stainless steel. Hydrodynamic details on the cascades
can be found in Wetzel and Arndt (2), and in the references
Figure 1 is a sketch of the overall geometry of the LCC. cited. To help with the acoustic isolation of the test section
The water tunnel circuit shell is 65 ft (19.8 m) high and 239 region and to inhibit noise from passing up through the flow
ft (72.8 m) long, containing 1.4 million gallons (5.3 million circuit from the lower leg, acoustic treatments have been

245
 VERTICAL SURVEY During the design process, concerns with the hydrodynamic
100 performance of the non-symmetric contraction and main
diffuser led to a separate experimental program in which the
upper leg components were built as a wind tunnel model at
80 1/10 scale. These tests were conducted at the St. Anthony
0 HOT WIRE °C1
Falls Hydraulic Laboratory (SAFHL) of the University of
PITOT CYLINDER Minnesota. Velocity profile measurements were made at the
60
ocoo 1 entrances and exits of the contraction, test section, and the
diffuser. Flow visualization surveys were conducted and
many surface static pressures were obtained. Tests were
40 043
conducted in this model to determine the effects of various
o fillets in the corners of the flow passages. A mathematical
op model of the contraction geometry was developed, based on a
20 o4151
three dimensional, time-dependent Euler numerical flow code
Q0 from the U.S. Air Force. Comparisons of measured and
computed mean test section velocity distributions are in good
o P agreement, as indicated in Figure 4. The final 6:1 contraction
o 0.2 0.4 0.6 0.8 1.0 12 ratio geometry was found to be tolerant of inflow distortions
and resulted in a very uniform test section velocity profile,
UfUrnax.
within the goal of ± 1%.
Figure 5- Measured Longitudinal Component Velocity The main diffuser geometry was also developed and verified
Profile at Vertical Centerline of the Main Diffuser
Exit. Data from SAFHL Wind Tunnel Model so that it would perform with acceptable efficiency while
avoiding flow separation and instabilities. Figure 5 shows a
typical measured velocity profile at the vertical centerline of
applied to the pressure sides of the turning vane foils in the diffuser exit. This nonuniform profile influenced the
Elbows number 1, 2, and 3. Figure 3 shows the typical turning vane geometry in elbow number 1, immediately
acoustic tile coverage of the pressure side (face) of a turning downstream of the diffuser. Wetzel and Arndt (2) have
vane. discussed some of the problems of dealing with diffuser exit
flows interacting with turning vane cascade flows.
HORIZONTAL SURVEY
OF LONGITUDINAL VELOCITY Thus far, very little systematic velocity profile data has been
200 obtained with the full scale LCC. Figure 6 presents the
results of a horizontal velocity survey of the longitudinal
velocity component obtained using the LCC Laser
175
Velocimetry (LV) equipment. It was taken just downstream
- Test Section Centerline entrance to the test section and it confirms the predicted
150 flatness of the velocity profile. Eventually, there will be
extensive flow velocity characterizations carried out in the
LCC test section and elsewhere in the circuit.
125
o
CG
In addition to the 1/10 scale partial model discussed
100 previously, another wind tunnel model at a scale of 1/6 was
C.) constructed at Bolt Beranek and Newman, Inc. (BBN) of
Cambridge, Massachusetts to evaluate the steady and unsteady
75 flow in a partial model consisting of the main diffuser +
elbow number 1 + vertical leg + elbow number 2 + approach
50 to the main pump. The objectives of the wind tunnel tests
were first to measure the detailed velocity field at the impeller
location for impeller design and noise predictions, and second
25 to measure flow excitation and unsteady forces on the two
turning vane cascades. As a result of these experiments, there
o
0 ,0 were recommendations for improvement of the proposed
0 1 2 3 4 5 5 7 8 9 10 turning vane shape and for improvement of the quality of the
VELOCITY, U (m/s) impeller inflow velocity pattern.

The LCC flow management section consists of the flow

Figure 6- LDV-Measured Transverse Profile of Longitudinal straightener and honeycomb located upstream of the
Component Velocity, Taken Near the Entrance of
contraction and was designed with the goals of development
the Full Scale LCC Test Section

246
 of a uniform velocity profile, removal of large scale turbulent allow excellent flow and cavitation observations,
eddies, and reduction of the background RMS turbulence level photography, laser velocimetry, and provide for various laser-
to 0.1% in the test section. A combination of theoretical optical measurement schemes for determining bubble size
results, small scale model measurements, and consideration of distribution and for particle displacement tracing and
existing water tunnel honeycomb installations led to the final velocimetry. To accommodate the extremes of test section
configuration. Some pertinent information on honeycomb pressures, the windows are 3.75 in. (95 mm) thick.
attenuation of turbulence is discussed in Wetzel and Amdt (2).
The LCC flow straightener is 24 in. (610 mm) long and is The test top is a large inverted box structure that functions as
made up of 4 in. (102 mm) square cells. The honeycomb a removable cap to cover the entire top side of the test
just downstre,am of the flow straightener is 19 in. (483 mm) section. It encloses a utility space lying above the ceiling
long and consists of hexagonal cells with the dimension plane of the test section designed to accommodate extensive
between parallel cell walls of 0.22 in. (5.6 mm). model/strut installation equipment; access ports and
collection systems for cabling and piping needed for model
A highly capable deaeration system has been installed with powering and instrumentation; and the above-waterline
the LCC. Two large deaeration tanks, 34 ft (10.4 m) in portion of a surface ship model hull. The internal
height and 12 ft (3.66 m) in diameter, can be operated in dimensions of the test top measure 5.33 ft (1.63 m) wide by
parallel. Each one has a flow rate capacity of 3500 gal/min 2.17 ft (0.66 m) high by 42.67 ft (13 m) long. Also shown
(13,249 liters/min) using a 350 hp (261 kW) positive in Figure 7 are the sixteen motor-actuated latches or clamping
disp/acernent pamp. With these two tanks working, it is mechanisms which hold the test top in place. Use of the
possible to reduce the air content of the LCC circuit flow latches instead of bolts results in a large saving of time
from 100% saturation to 30% saturation in four hours, and to needed to remove the test top. The general scheme for
10% saturation in 10 hours. mounting models in the LCC is to install the hull or
mounting strut onto the removable test top only. This
allows easy installation and removal of the model and
complete outfitting of the model outside of the test section
on a separate test stand. Eventually having multiple test tops
will allow rapid exchange of test models. The test top is also
reversible; that is, it can be turned 180 degees. This feature
will permit the collection of four-quadrant propeller data
without having to remount the ship hull model. The water
tight boundary which must be maintained between the test
top and the lip of the hole over the test section is provided by
an inflatable pneumatic seal.

PUMP AND POWERING PERFORMANCE

Based on the results of the wind tunnel experiments and

estimates of head loss coefficients for components around the
LCC circuit, overall empty tunnel (no model) head loss at 30
knots (15.4 m/s) test section velocity was estimated to be
about 10.5 ft (3.2 m) of water. Of this total, about 15
percent was estimated to be in the test section and about 44
Figure 7- Isometric View of LCC Test Section percent in the main diffuser downstream of the test section.
With about 60 percent of the overall head loss involved with
the test section and diffuser, it was decided to specify that the
Test Section and Test Top main pump be designed to handle up to a 50 percent increase
Immediately downstream of the main contraction, the facility of head loss in the test section due to the presence of a large
test section is a constant cross-section area segment of the model. This led to an initial recommended design head loss
flow circuit, 40.8 ft (12.4 m) long, where models are placed of 13.6 ft (4.15 m) for the main pump.
for testing. The basic 10 x 10 ft square cross section shape
of the test section was modified with flat comer fillets, fitted The low head loss around the circuit and the high flow rate
with viewing windows. The net cross section area is 96.15 required a pump with unusual characteristics not covered by
ft2 (8.93 m2). Figure 7 is an isometric view of the test normal commercial pump design. Therefore
section showing the deep girder external stiffening frames and DTMB/CDNSWC embarked on a program to develop a
the eight bays of viewing windows along both side walls. custom pump design through Voith Hydro, Inc of York,
All the viewing windows along the vertical sides, in the four Pennsylvania. To help with the hydrodynamic and cavitation
corner diagonal panels, as well as along the top of the performance evaluation of the pump, a 1/14 scale water
removable test top are made of polymethyl methacrylate tunnel model of the LCC was built by Voith Hydro. Model
plastic manufactured with an optical quality finish. These measurements of velocities, pressures, visual and acoustic

247
cavitation inception speeds, and flow visualization were all Acceptance testing results for the LCC with empty test
used to evaluate the pump performance. Five different blade section (no model) indicate that a test section velocity of 35
tip geometries were evaluated to settle on a design which knots (59.1 ft/s, 18 m/s) can be a.chieved with a main pump
would suppress tip vortex and gap cavitation. The pump was speed of 56 RPM.
required to be free of face, back, hub, and root cavitation. It
was specified to have minimal tip vortex cavitation over the A summary of the principal particulars and performance
range of speeds, pressures, and velocity distortions features of the LCC is given in Table 1.
encountered in the LCC operating matrix. The final pump
configuration is a single stage, fixed pitch, seven-bladed,
axial flow rotor 18.083 ft (5.512 m) in diameter with a hub TABLE 1- Summary of Main LCC Particulars
diameter ratio of 0.55 and a downstream stator with nine
blades. At the design condition head of 14 ft (4.27 m) of Test Section 10 x 10 ft (3.05 x 3.05 m)
water and flow rate of 4900 ft3/s (138.8 m3/s) the design Square with flat comer fillets
rotation speed is 55 RPM. Final impeller blade tip clearance Net Test Section Area 96.15 ft2 (8.93 m2)
Max Velocity (Test Section) 35 lcnots (59.1 ft/s, 18 m/s)
is 0.25 in. (0.64 cm) or approximately 1/1000 of the rotor Pressure Range (Top of Test Section) 0.5 to 60 psia (3.5 to 415 kPa)
diameter. Figure 8 shows a projected view of the pump Velocity Nonuniformity Less than ± 1.0%
rotor. Turbulence Level (Goal) 0.1%
Deaeration Capability
Time to reduce air content from
LEADING EDGE full saturation to 30% saturation 4 hours
Contraction Ratio 6:1
Main Diffuser Ratio (Before Elbow) 1:3.14
Main Pump Power 14,000 hp (10,440 kW)
Impeller Diameter 18.083 ft (5.512 m)
Main Pump Rotation Speed (Design) 55 RPM

SOME HYDROACOUSTIC CONSIDERATIONS

The characteristic hydroacoustic noise of the LCC is a very

important aspect of the design and for facility performance.
For a number of potential hydrodynamic noise sources,
Figure 9 from Gre-eley, et al (3), shows a collection of
theoreticaVempirical predictions of facility baseline one-third
octave band sound pressure levels in the test section of the
LCC at 30 knots flow speed. These sources include pump
blade leading edge and tip gap cavitation, noncavitating
turning vane noise, test section turbulent boundary layer
(TBL) noise, and impeller blade trailing edge noise. Many
other potential noise sources are not shown in this plot.
Some of the other possible origins of noise are: two phase
flow noise due to free bubbles, flow noise in and out of
access piping of the pressure regulation and deaeration
systems, external airborne environmental noise transferred
Figure 8 - Sketch of Seven-Bladed Main Pump Impeller through the tunnel shell, and machinery noise form the main
drive motor and from the several subsystem auxiliary pumps.
The LCC main pump is driven by a 14,000 hp (10,440 kW) When de,aling with just the items included in Figure 9, it is
24-pole synchronous electric motor with a cycloconverter apparent that the elimination of pump blade leading edge
control system for providing the variable drive frequency. cavitation is a very important aspect of test section noise
Designed and manufactured by GEC Large Machines of reduction. As noted earlier, the LCC main pump impeller
Rugby England, the motor is a smaller version of the QE-2 has been shown to be cavitation free.
main propulsion drive motors. Maximum motor speed is 65
In order to reduce noise and vibrations generated by rotating
RPM. machinery in the LCC, all such equipments are mounted
Extensive testing in the 1/14 scale water tunnel model with typically on double stage vibration isolation bases. In
the final impeller design indicated no cavitation present addition to the installations of the smaller auxiliary
anywhere in the operating range. So far, observations of the equipment on this type of mounting, the main 14,000 hp
impeller tip flow region of the full scale LCC using (10,440 kW) pump motor is mounted on a two-stage inertia
stroboscopic lighting through viewports in the top of the block system. The two concrete inertia blocks each measure in.
main pump housing confirm cavitation-free operation. about 35 ft (10.7 m) square by 9 ft (2.7 m) thick. Al
248
 (25.4 mm) thick elastomeric layer separates the blocks from TEST 20P
each other as well as the lower block from the main motor UPPER
foundation. Also, the motor foundation is separate from the DIAGONAL
VIEWING
foundation of the main LCC structure. Potential reduction WINDOW
gear noise was eliminated by the selection of the low RPM
main drive motor described earlier. Added isolation is
provided by the use of two elastomeric drive couplings along CEILING PLATE

the drive shaft leading from the motor to inside the tunnel.

140 BASELINE LCC TEST SECTION SPL AT 30 lcnots

SIDE
L.E. VIEWING
'c3
a. Cavitation kb._ bki Ps, psia
267
1
WINDOW

130 . csi =28 --.444%.7mmo 15

30
I
120 45
tr.]
Gap
Cavitation
Li4
110 r csi = ZOG
Z)cz
> C4
<
gz.1
100
ILI= T13L Noise 1

ACOUSTIC
TROUGH
90 Impeller Trailing
Q.0 Turning Vanes IF' .<-: Edge Noise
. (INSTRUMENTATION CHAMBER)

80 . .

0.25 05 1 2 4 8 16 32 64 125 g g
1/3 OCTAVE BAND CENTER FREQUENCIES, IcIlz
Figure 10 - Sketch of LCC Test Section, Showing the Test
Figure 9 - Predicted Contributions from Various Noise Top (Equipment Mounting Chamber) and
Sources In the LCC Test Section at 30 knots Acoustic Trough (Hydrophone Instrumentation
(from Greeley, et al (1987)) Chamber)

As a result of the predictions collected in Figure 9, and

The main hydroacoustic measurement system of the LCC accounting for the various quieting features which
consists of a 95-element hydrophone array which is made up actually contracted to be incorporated into the LCCwere
of four nested sub-arrays, each designed for a portion of the as
constructed, the target contractual background noise goal was
frequency range from 1 kHz to 20 kHz. This system will given as sound pressure level equal to 110 dB
allow beam forming analysis in order to determine locations at 1 kHz,
decreasing to 100 dB at 10 kHz, and then constant at 100 dB
of noncavitating sound sources along the hull-propulsor
model. The acoustic beam steering capability will also out
to higher frequencies. This contractual noise goal is
shown in Figure 11. As of this writing, the acoustic
provide a noise signal discrimination effect against unwanted characterization of the LCC in its initial noise testing
noise sources other than those coming from the model located configuration is incomplete. Observers standing
in the center region of the test section. The array system is to the test section with flow in the tunnel, have notedon or next
that
mounted on a movable platform placed inside a flooded the perceived noise and vibration levels
are 'very low', even at
instrumented chamber ("acoustic trough") located below the 30 knots test section velocity.
test section. This acoustic trough houses the array away
from direct contact with the test section stream flow. It has If needed, the lower half of the tunnel circuit
its own deaeration system and the pressure in the trough is could be
surrounded by a pool of quiescent water as a background noise
balanced with the static pressure in the test section. The two silencing technique aimed at low frequency noise
are separated by an "acoustic window", which also serves as the lower leg of the LCC. The large trench generated in
the floor of the test section. Figure 10 is a typical cross in which the LCC sits could be lined with or concrete box
section tfu-ough the LCC test section, showing the acoustic materials such as saturated wood and sand sound absorbing
trough, as well as many of the other features described earlier. or gravel mixed
with water. Such a flooded trench would reduce the acoustic
The acoustic trough measures 7 ft (2.1 ni) wide by 3.75 ft impedance mismatch at the tunnel wall and would
(1.1 m) high by 45 ft (13.7 m) long. Anechoic treatments allow
sound to pass more easily out of the tunnel to be absorbed by
being applied to the inside of the acoustic trough consist of
the treatments. To date, the lower leg trench has not been
de,coupler material tiles and a random pattern of two sizes of
filled with water and sound absorbing materials have not been
de,coupler material wedge shapes. installed.

249
 around 8000 RPM, and a typical required operating speed
130
3010aots
130 range of 200 to 6500 RPM. Physical size of the motors is
60 psia
120
an important factor in matching the motor to the
requirements of the experiment. The high range motor
dimensions (diameter and length) are such that, for single and
110
twin screw arrangements, the motor(s) can be accommodated
100
only by a larger size hull model, with hull length of 30 ft (9
100
m) or longer; and with propulsor characteristics corresponding
90 to Range III of Table 2. The high range motor is too big for
typical moderate size hull models. However, the medium
80 so range motor is well suited for the moderate size category of
models, in terms of both its physical dimensions and its
Goal Based on Calculations Excluding 70
70 Government-Furnished Acoustic Treatments torque capability.
60 60
as
E TABLE 2- Basic Particulars of LCC Transmission Dynamometers
§ as
o

113ocrAVE BAND CENTER FREQUENCY (Hz) Typical Model Prop Max Max Max
Range Model Prop Diam Weight RPM Thrust Torque
(in.) (lb) (lb) (fi-lb)
Figure 11 - Contractual Background Noise Goal for LCC [mm] [kg] [N] [N-ml
Test Section 20 6000 ±400 ±80
I 5_ 10
[254] [9A] [±1780] [±108]

II 10 25 6000 ±1200 ±320

DYNAMOMETRY, MOTORS, AND MODELS [254] [11.3] [±5340] [±430]

A facility with the size and speed capability of the LCC ECI 15 75 3000 ±3600 ±1280
[380] [34] [± 16,015] [± 1735]
makes special demands on the instrumentation, model hull
size and strength, and model propulsor drive motors. Included
here is an outline of the basic dynamometry and model
equipment needed for conducting powered experiments in the The first powered-model cavitation evaluation experiment
LCC. carried out in the LCC was performed with a 36.2 ft (11 m)
long hull model of a twin screw combatant ship. Some
There are three ranges of transmission dynamometers that particulars of the water tunnel model are given in Table 3.
cover the requirements of the basic categories of LCC
hydrodynamic and cavitation evaluation testing. The
properties of these thrust-and-torque sensors (TQS) are TABLE 3- Particulars of Twin Screw Combatant Ship
outlined in Table 2. These TQS elements have been designed Water Tunnel Model
so that they can be used in either single propeller or
contrarotating propeller arrangements. The dynamometers of
Range I are for experiments with moderate size surface ship Length, LwL 36.2 ft (11. m)
and submerged body models of length around 20 ft (6 m) or Beam, 13, 4.59 ft (1.4 m)
multiple screw (three or more propellers) ship models around Draft, Tx 1.61 ft (0.49 in)
30 ft (9 m) long, for which the high Reynolds number goal Max Section Area, Ax 16.12 ft2 (0.568 m2)
Prop Diameter, D 15.86 in. (402.7 mm)
of RØ7 5x106 is not deemed necessary (e.g., for unsteady
pressure pulse evaluation). Ranges II and III dynamometers Model Hull and Mounting
are intended for the main line propulsor cavitation evaluation Material Glass Reinforced Plastic (GRP)
experiments where high model propulsor blade Reynolds Nominal hull shell thickness 0.625 in. (1.6 cm)
numbers are advisable in order to minimize scale effects. Total Weight on Lift Post 9000 lb (4082 kg)
Tunnel Blockage 6.4%
There are two ranges of model propulsion system drive
motors available for experiments in the LCC. Both are Essentially all future surface ship model testing in the LCC
special high power density, three phase, six pole, constant
torque, variable speed induction motors that require water will resemble this experiment. Constructed of thick
cooling. Both are built by the Able Corporation of Yorba fiberglass reinforced plastic, the large hull model was
designed to withstand a peak test section flow velocity of 35
Linda, California. (i) The high range motor has a torque lcnots. The empty weight of the hull shell alone is around
rating of 525 ft-lb (712 N-m) with a maximum speed of 7200
RPM, and a typical required operating speed range of 200 to 2000 lb (900 kg). Total weight of the model hull plus
internal equipment and miscellaneous support structure comes
4000 RPM. (ii) The medium range motor has a torque
rating of 110 ft-lb (150 N-m) with a maximum speed of to about 9000 lb (4080 kg). The entire model package was

250
 -

",..
1--.1Err ,o,
...............................................
TEST TOP OPENING

LIFT POST
L1FT BEAM

Tff
ThTff,, , ,,, rhril
LJiLJLILJ : =J1=-1!_[51/41[7=i

j LL___
Figure 12 - Twin Screw Surface Ship Model Mounted in the LCC Test
Section, Including Possible Yawed Model Arrangement

attached to a robust lift beam mounted to the bouom end of a photography were also possible through the upper diagonal
vertical lift post located near the hull midships. The lift post windows, fitted with viewing wedges.
can be raised and lowered vertically using a motorized drive
located outside and on top of the test top. The single lift Figure 12 shows two sketches of the model positioned in the
post allows the hull model to be rotated about the post's LCC test section. This first major powered experiment
vertical axis,so that the hull may be positioned with a non- consisted of cavitation observations, cavitation inception, and
zero yaw angle as well as aligned with the test section stre,am cavitation noise evaluation conducted with the tunnel test
flow (see Figure 12). velocities equal to full scale speeds. Chord length Reynolds
numbers ranged from 3.71 x 106 to 5.84 x 106 ,
With the test top removed from the tunnel and sitting on the corresponding to the ship speeds of 20 to 30 knots.
test stand, most of the model drive system and
instrumentation equipments were installe,d with the model on DISCUSSION
the lift post in the lowered position. When ready, the lift
post plus the loaded model were raised to the final position, The LCC facility has advanced technology features important
with the design waterline located at the ceiling-plane of the to superior hydrodynamic and hydroacoustic performance.
test section. Two pairs of rugged cross support bearns held Much of the planned flow and background noise
the forward and aft ends of the model firmly fastened to the characterization testing has not yet been carried out in a
test top. After the test top has been lifted up and clamped complete and systematic way. The main reason for this is
into place over the test section, the ceiling plates were cut to that there is still finish installation work underway. For
fill in the gap between the hull and the test top opening. example, the anechoic treatment of the acoustic trough has
With the model in the test section, final connections were
not been completely finished as this paper is being prepared.
made of cooling water lines, cooling air supply lines, The only check-out experiments completed in the LCC are of
electrical power cables, etc.
the basic acceptance testing type. In all cases where
performance has been measured, the LCC specifications have
This model was fitted with a total of eight 3 in. (7.6 cm) been met or exceeded.
diameter viewing ports cut into the bottom of the hull in the
vicinity of the two propellers. These offered close-up views
As of this writing, three successful experiments have been
of the pressure and suction sides of the propeller disc, on both
executed in the LCC: a performance evaluation of a full scale
sides of both shaft lines. These ports were used for various
combinations of strobe lighting and photo-observations with ship appendage; an investigation of speed and size scale
small, high resolution video cameras. Excellent viewing and effects on cavitation bubbles and cavitation inception in the
flow over headforms (Kuhn de Chizelle, et al (4)); and the
251
propeller cavitation performance evaluation of a twin screw ASME International Symposium on Cavitation Rese,arch
combatant ship describe,d earlier. Facilities and Techniques-1987, FED-Vol. 57, Boston,
Dec 1987, pp. 91-97.
ACKNOWLEDGMENTS
Kuhn de Chizelle, Y., S.L. Ceccio, C.E. Brennan, and
The acquisition of a major test facility such as the LCC Y. Shen, "Cavitation Scaling Experiments with
involves the dedicated efforts of hundreds of individuals and Headforms", Second International Symposium on
many organizations. While space does not permit a complete Propeller and Cavitation, Hangzhou China, September
recognition of all contributors, the authors would like to 1992.
acknowledge the continued support of the LCC Project by
Naval Sea Systems Command, the Space and Naval Warfare
Systems Command, and the David Taylor Model
Basin/Carderock Division, Naval Surface Warfare Center
management over the past 12 years through three name
changes and four Commanding Officers. Throughout this
period, the LCC Project Office has managed the technical and
administrative aspects of the LCC design and acquisition for
the Ship Hydromechanics Department.

Initially, studies by the David Taylor organization, Naval Sea

Systems Command, Kempf and Remmers GmbH, and
Hydronautics, Inc led to a design contract between Naval
Facilities Engineering Command, Chesapeake Division and
the Joint Venture (JV) of Seeley, Stevenson, Value and
Knecht (SSV&K) and Daniel, Mann, Johnson and
Mendenhall (DMJM). The JV team included Bolt, Beranek
and Newman, Inc., St. Anthony Falls Hydraulics Laboratory
(University of Minnesota), Davidson Laboratory (Stevens
Institute of Technology), and Kempf and Remmers, GmbH.
The David Taylor organization provided additional design
support through Allis-Chalmers Hydro-Turbine Division and
Atlantic Applied Research Corporation. This effort resulted
in the preliminary design and specifications for the LCC.
The prime contractor for the detailed design and construction
of the LCC was CBI Na-Con, Inc. with major support from
CBI Services, Inc., Voith Hydro, Inc., GEC Large Machines,
Ltd., and Ralph M. Parsons Company. The LCC was
dedicated in April 1991. Support and data acquisition
systems for the LCC were developed with the contract
support of Oak Ridge National Laboratories, Wyle
Laboratories, Inc., Coggins Systems, Inc., and Advanced
Marine Enterprises, Inc. The authors are grateful to Ms.
Melanie Joy for typing the manuscript.

REFERENCES

Morgan, W.B., and R.J. Etter, "Initial Experience with

the Large Cavitation Channel", International
Symposium on Hydro-and Aerodynamics in Marine
Engineering, Varna, Bulgaria, Oct 1991.

Wetzel, J.M. and E.A. Arndt, "Hydrodynamic Design

Considerations for Hydroacoustic Facilities", ASME
Hydroacoustic Facilities, Instrumentation, and
Experimental Techniques, NCA-Vol. 10, Atlanta, Dec
1991, pp. 1-8.

Gre,eley, D.S.> P.A. Abbot, N.A. Brown, and R.S.

Rothblum, "Water Tunnel Background Noise Models,"
252
 Flow Field Measurements at a Free Surface
Using Digital Particle Image Velocimetry
J. Hamilton, S. Fish, D. Anthony
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT using DPIV compared to hours for

conventional PIV techniques, which require
A new tool for flow field the time-consuming processing of
measurement is in use at the David Taylor photographic film, followed by
Model Basin (DTMB). The Digital Particle interrogation of the film images to obtain
Image Velocimetry (DPIV) technique the particle displacements. At present,
de-veloped by Willert and Gharib (1) has conventional PIV flow field measurements
been used to measure the flow fields of a have higher spatial resolution and are
2D standing wave and of a laminar vortex applicable to higher speed flows. It is
pair interacting with 2D trav'eling waves expected that DPIV will become a
at a free surface. Application of the competitive technique for flow field
DPIV technique to free surface flows measurement with the development of higher
introduces new measurement challenges as resolution imaging arrays and faster video
the position of the non-stationary surface frame rates.
is not known a priori, but must be DPIV flow field measurements of a
determined from the video images 2D standing wave were made to test the
themselves. DPIV system using a well-known flow field
The initial application of the DPIV and to develop techniques for measuring
technique at DTMB has shown that it is a velocities near a free surface. The
useful tool for flow field measurement and authors were interested in applying the
visualization with a high throughput rate. technique to the study of more complicated
It is possible to obtain flow field flows, in particular to the interaction of
measurements in almost real time, and the laminar vortex pairs with traveling waves
results are immediately available for use at a free surface, investigated recently
in directing experimental efforts. It is by Fish (3).
envisioned that a myriad of applications This paper will describe the DPIV
will be found for the DPIV technique at system as it is implemented at DTMB.
DTMB, including the mapping of flow fields Representative data from the initial 2D
around hull forms, appendages, and standing wave experiments will be
propulsors. presented, and the approach used by the
authors to analyze the data, especially
INTRODUCTION that obtained just beneath the free-
surface waves is discussed. In spite of
The DPIV technique has evolved from the initial difficulties encountered in
more familiar, related techniques such as applying the DPIV technique to flows with
laser speckle velocimetry (LSV) and a moving free surface, it became evident
particle image velocimetry (PIV) (2). In during this work that DPIV is a valuable
DPIV, fluorescent or light-scattering tool for obtaining quantitative flow field
particles are used to seed a flow measurements with many potential uses at
illuminated by a laser light sheet. DTMB.
Images are collected in digital form using
a video camera and a computer equipped THE DPIV METHOD
with a video frame-grabber board.
Particle displacement fields are The DPIV analysis method of Willert
calculated from two successive video and Gharib (1) will be described briefly.
frames, and velocity fields are derived The method uses a statistical cross-
from the displacement fields by dividing correlation technique to determine the
by the time interval between frames. The average displacement of a group of
pair of video images is sampled using an particles within a sampling window. The
interrogation window, and the average Fast Fourier Transform (FFT) algorithm is
displacement for particles in a window is used to evaluate the cross-correlation
calculated using a spatial cross- function efficiently.
correlation technique. A vector plot of Two video images captured at times
a flow field can be obtained in minutes At apart are sampled using interrogation

253
windows of equal areas, N rows high by M glass walled tank was 48 in. long, 18 in.
columns wide, at the same location in each wide, and 20 in. deep, and was filled to
image. The cross-correlation function of a depth of 14 in. The 2D standing wave
the two samples is given by had a period of 0.51 s and a wavelength of
about 16 in., since three full cycles were
+N2 M
2
observed over the length of the tank.
The water in the tank was seeded
4)fg(j,k) =E E f(n,m)g (n +j,m Al) with fluorescent particles. The particles
were neutrally buoyant, 30-40 micron
N M diameter polystyrene spheres with a
2 2
two fluorescent dyes,
combination of
fluorescein and rhodamine, encapsulated in
where f(n,m) and g(n,m) are the pixel gray them. The particles emitted a yellow-
levels in the imaging array at row n and orange fluorescence when illuminated by
column m at the two times. The gray level the green (514.5 nm) line from an argon-
is a number pr000rtional to the integrated ion laser.
light intensity absorbed by the pixel over The laser light sheet was produced
the period of one video frame. by a 6 watt argon-ion laser. The 1.5 mm
The spatial cross-correlation diameter beam from the laser was expanded
function will have a maximum value at by two, using a beam expander, then made
offset j,k when the pattern of particles into a sheet using a 6.3 mm focal length
in the first window best matches the diverging cylindrical lens. The resulting
pattern of particles in the second window, light sheet was approximately 3 mm in
after shifting by j pixels horizontally thickness. The laser was operated in a
and k pixels vertically. The average single-line mode at 514.5 nm and 3 watts
displacement for all the particles of power.
contained within the interrogation window Images are acquired using a CIDTEC
is obtained by searching for this peak 2250 Charge-Injection-Device (CID) camera.
value of the cross-correlation function. This camera has a 512 by 512 pixel
Direct computation of the cross- element solid state imaging array. The
correlation function using the above array is square, measuring roughly 0.5 in.
equation is computationally intensive; on each side. The pixel elements of the
Willert and Gharib achieved a significant array are also square with a width of 15
reduction in the computer time required to microns. The video output from the camera
evaluate the correlation by using the Fast is a single, non-interlaced field
Fourier Transform (FFT) algorithm to consisting of 506 rows by 512 columns. The
transform the two functions f(n,m) and video framing rate is 30 Hz.
The nonstandard video output of the
g(n,m) to the spatial frequency domain. Poynting
The (complex) transforms are then CID2250 is converted by a
the convolution TN224MEG camera interface card to
multiplied to obtain
function in the spatial frequency domain. standard RS-170 video format which is
cross-correlation function is compatible with an EPIX 4MEG frame grabber
The
recovered by taking the inverse FFT of the board. The 4MEG board was obtained from
convolution, and the average particle Poynting and had modifications made to its
displacement is found by searching for the DC restoration circuits and pixel clock
maximum value of this cross correlation circuit. The normal 14.3 MHz pixel clock
function. Once the maximum of the array module of the 4MEG board was changed to a
of cross-correlation values is determined, 12 MHz clock module by Poynting. The
the estimate of the value of the average modified 4MEG frame grabber and the
particle position is refined by fitting a Poynting TN224MEG camera interface are
2D Gaussian curve to the maximum and its available as a package from Poynting which
four neighboring elements. The position also includes rudimentary camera control
of the interpolated peak is determined to software.
subpixel accuracy. The 4MEG frame grabber is capable of
capturing, in its 4 megabytes of on-board
EXPERIMENTAL SETUP video memory, up to sixteen video frames
at the 30 Hz framing rate. Each pixel
DPIV System Hardware integrated intensity is digitized to 256
gray levels by an 8-bit analog to digital
A simple flow was selected for converter. One 506 line by 512 column
initial tests of the DPIV system hardware video frame requires 259 kilobytes of
and software. A 2D standing wave was disk storage.
generated in a 70 gallon tank using a Image capture and storage were
plunger type wavemaker (Figure 1). The controlled by an IBM-compatible Gateway

254
2000 486-33MHz personal computer, which Willert and Gharib to obtain velocity
was also used to analyze the images. At measurements in free surface flows.
the end of each run, sixteen images Figure 2 is a pair of images taken
grouped as eight imaae pairs were 0.20 s apart showing one half wavelength
downloaded from the image memory of the of the 2 Hz standing wave. The field of
frame grabber and written to disk. A 90- view of the camera was roughly a square
megabyte Bernoulli disk drive was used to 8.5 in. wide. Artifacts in the raw images
archive the images. which were responsible for some of the
The image acquisition was timed stray velocity vectors obtained in the
using a Data Translation DT2819 vicinity of free surface can be seen in
counter/timer board. The counter/timer Fig. 2. The free surface is easily
board was used to measure the frequency observed in these images, because of
and then totalize counts from the 15.7 kHz fluorescent particles floating on the
horizontal line scan clock signal from the surface. These particles form a broken
camera. Time interval measurements were curve which delimits the intersection of
obtained by dividing the difference in the laser light sheet and the free
counts at two times by the frequency. surface. Above the line of bright
Using this clock signal was convenient particles can be seen a hazy gray band
since 526 horizontal scan clock pulses containing bright blurred particle images;
were generated during each video frame (a these particle images are actually
frame consisted of 506 lines plus a 20- reflections from the underside of the free
clock-cycle vertical blankina period). surface. There is also an accumulation of
The end of frame signal from the particles on the glass wall of the tank
video camera was used ro strobe the laser above the wetted area of the wall lapped
light sheet. An NM Laser products L5200 by the waves. Fig. 3. shows the results
mechanical shutter was used to interrupt from analyzing the raw video frames using
the laser beam. A shutter control circuit the DPIV analysis program of Willert and
consisting of two one-shots (monostable Gharib. Numerous stray vectors can be
multivibrators) was constructed to control seen in the region of the free surface.
the exposure. The input for this circuit Stray vectors are also present in the
was the end of frame (EOF) signal from the region devoid of particles near the top of
camera. The output of this circuit was a the images. Stray vectors result when the
pulse delayed by at least 20 horizontal search for the maximum of the cross-
line scan periods (in order to avoid the correlation function yields a noise peak
vertical blanking period after the end of instead of the desired average
frame pulse) with an adjustable width. displacement of the particles in the
The width of the pulse controlled the sampling window.
exposure time for the images. An exposure It was evident that some form of
of 4 ms was used to freeze the particle image processing would be necessary to
images and prevent them from smearing. extract the desired velocities near the
free surface and to eliminate these stray
DPIV System Software vectors. The authors' approach was to
process the images to remove the artifacts
The image acquisition and analysis by cropping off the area of the image
software was obtained from Willert and above and including the line of
Gharib. Modifications to the image fluorescent particles at the free surface.
acquisition software were necessary to Initially, the image cropping was done
adapt it for our purposes. Changes were interactively using the 4MIP image
made in the timing routines for image processing program by Epix; later a
capture. These routines were converted to program was written to automate the
use the Data Translation DT2819 finding of the free surface and cropping
counter/timer board to obtain better of the images. Using the 4MIP software,
timing accuracy than the PC system clock a Bezier polynomial was fitted to eight
(which only updates at a rate of 18.2 points chosen from the line of particles
times per sec). at the free surface and all pixels in the
area above the curve were blanked (set
FLOW FIELD MEASUREMENTS OF A STANDING equal to zero). The cropped images were
WAVE analyzed using the DPIV software. Vector
plots obtained using cropped images are
The DPIV systemwas tested by making shown in Fig. 4. In addition, a program
measurements of the flow field of a 2 Hz for locating and eliminating stray vectors
standing wave in a tank. This allowed us was included in the DPIV software.
both to optimize seed concentrations and This
program would search the grid of vectors
to adapt the image analysis routines of for candidate vectors that had magnitude

255
greater by some tolerance level than its be used to crop the images obtained with
nearest neighbors. If the tolerance level the DPIV camera.
was exceeded, the vector was replaced by
an average vector obtained by CONCLUSIONS
interpolating the nearest neighbors in the
grid. The vector plots shown in Fig. 4 Our initial application of the DPIV
were obtained from cropped images and have technique has shown it to be a powerful
had stray vectors with a magnitude of 5 tool with many potential uses at DTMB,
pixels greater than neighboring vectors especially in situations where
removed and replaced with interpolated instantaneous, multi-point data are
vectors. needed, such as in transient flows like
The vectors in Figs. 3, and 4 were those discussed above. We expect that
calculated using a 32 by 32 pixel DPIV will be used initially to obtain
interrogation window and are plotted on a quantitative flow visualizations,
square grid of 8 by 8 pixels. The revealing interesting features such as
interrogation windows were overlapped by vortices, flow separations, and flow
75 percent. The computer time required to reversals. The DPIV results, which can be
compute a displacement field of 61 by 51 obtained in minutes, will then be used as
nodes (3111 vectors) was seven minutes. a guide for later measurements using
conventional single-point techniques, such
DPIV MEASUREMENTS OF A VORTEX PAIR as laser Doppler velocimetry and hot-film
INTERACTING WITH TRAVELING WAVES anemometry.
The limitations of the DPIV
The image processing techniques technique, at present, are its resolution
which allowed us to eliminate many of the and its dynamic range. The resolution is
stray vectors in the standing wave images limited by the pixel densities of the
did not prove always to be applicable to currently available solid state arrays,
The while the range of velocities that can be
measurements in traveling waves. measured is limited by the 30 Hz video
particles lying on the free surface, which
framing rate. Improvements in resolution
allowed us to determine the location of
in the standing wave will come with the development of higher
the free surface
Lmages, were not always present in resolution solid state imaging arrays.
traveling wave images. These traveling High speed video cameras, and cameras with
waves actually cleaned the free surface by programmable electronic shutters will
transporting the particles to a beach permit much shorter times between frames,
opposite the wavemaker, where many of them allowing measurements of higher flow
were deposited. Higher frequency (shorter velocities.
wavelength) large amplitude waves were
more effective at cleaning the surface ACKNOWLEDGaMENTS
than longer wavelength, small amplitude authors owe much to Mr.
traveling waves. Frequencies near 3 Hz The
were chosen for the vortex free surface Christian Willert and Prof. Morteza Gharib
of the University of California at San
wave interaction studies, because it was Diego for the use of their software and
desired that the wavelengths be comparable
to the vortex separation. At this for guidance in implementing our DPIV
frequency, the surface was swept clean of system.
This work was funded by the Office
particles. of Naval Research under the Free Surface
Knowing the location of the free
surface in the image is essential for the Vorticity Program, administered by Dr.
cropping technique that worked well for Edwin Rood, ONR Code 1132F.
the standing waves. An alternative
technique for locating the free surface is REFERENCES
currently being explored by the authors.
This technique uses a second video camera Willert, C. E., and M. Gharib 1991
synchronized with the DPIV camera. The "Digital particle image velocimetry."
second camera is used to view the waves by Exp. Fluids 10, 181-193.
looking down at the free surface from Adrian, R. J. 1991 "Particle-imaging
above with the axis of the lens at an techniques for experimental fluid
20 degrees to the mechanics." Ann. Rev. Fluid Mech.
angle of about
undisturbed free surface. From this 23, 261-304.
vantage point, the wave shape outlined by Fish, S. 1991 "Vortex dynamics in the
the laser light sheet is clearlyvisible. presence of free surface waves."
The waveforms determined by digitizing the Phys. Fluids A, Vol. 3, No. 4, 504-
images collected by the second camera will 506.

256
 TANK -

CYLINDRICAL LENS
EAM EXPANDER
NM LASER SHUTTER

E WATT ARGON ION LASER

NM LASER
SHUTTER RGB VIDEO
CONTROLLER MONITOR

SO H. END OF FRAME PULSES

OTMEI
SHUTTER
CONTROL
CIRCUIT
VGA MONTTOR.

VIDEO AND CONTROL TN224MEC

SIGNALS 4MEG
INTERFACE RAME 48S-33 MHZ
CARD
GRABBER PERSONAL COMPUTER

p It.. .10
15.7 KM= LINE SCAN CLOCK SIGNAL
072819
CC/UNTE,
TIMER

Fig. 1. Schematic diagram of DPIV system

Fig. 2a. Half wavelength of a 2 Hz standing Fig. 2b. Half wavelength of 2 Hz standing
wave at t=0 s wave at t=0.20 s

257
 VECTOR PLOT OF THE DISPLACEMENT FIELD OF A 2 HZ STANDING WAVE
COMPUTED FROM VIDEO FRAMES AT T=0 AND T=0.033 SEC
512

,, , , ,

, ,, ... , --
, , -----.--
N.\ .\ ,\... ,

384 ,,,,////// /4 ,M\\\,.Z' ',:Z...-

\'\\\\\\,,,,,,,....---..------
l\\\\\\\.,.......,-...
1\ \\\\\\\
\
---------
........... %%"'"
%;: % '/' .5
1\W\\
, s. s s. s. ..

..
I
.. f.

256
r

128

. I I t e I I 1 . . 1 w 1 1
0
128 256 384 512
0
X (PIXEL UNITS)
Fig. 3a. Vector plots from raw video frames at t=0 and 0.033 s

VECTOR PLOT OF THE DISPLACEMENT FIELD OF A 2 HZ STANDING WAVE

COMPUTED FROM VIDEO FRAMES AT T=0.200 AND T=0.233 SEC
512

384

256
X

>-

128

e ,11-111.1. 128 256

,
384
1

512
X (PIXEL UNITS)

Fig. 3b. Vector plots from raw video frames at t=0.200 and 0.233 s
258
 VECTOR PLOT OF THE DISPLACEMENT FIELD OF A 2 HZ STANDING WAVE
COMPUTED FROM VIDEO FRAMES AT T=0 AND T=0.033 SEC
512 r-

- ----------
-.....---- ---//''''''';''i/J Ili
,,,,,,,/ihi -______ __
-

384 _ -----./;:iiiii-l!
--
I/ 1 ,

11111111=7.
........,-,/,
__,r,// 1,
.
,,,, ,,I,1,
....,,,,
,,
I

I%,,X\
I
,

256 .. .

I.

128
o

o 1- 1 . I I i V a a I I

o 128 256 384 512

X (PIXEL UNITS)
Fig. 4a. Vector plots from cropped video frames at t=0 and 0.033 s

VECTOR PLOT OF THE DISPLACEMENT FIELD OF A 2 HZ STANDING WAVE

COMPUTED FROM VIDEO FRAMES AT T=2.00 AND T=0.233 SEC
512 r

384

256

0,
>-

II
128

I I p

o 128 256 384 512

X (PIXEL UNITS)
Fig. 4b. Vector plots from cropped video frames at t=0.200 and 0.233 s

259
 Calibration of the Wavemaker System in the
Offshore Engineering Basin at the
Institute for Marine Dynamics
J.J. Murray, L.M. Mak, M.A. Sullivan
(Institute for Marine Dynamics, Canada)

ABSTRACT

The wavemaker system at the Institute investigations into wave absorption using
for Marine Dynamics (IMD) in St. John's, wavemakers were carried out.
Newfoundland was designed to simulate At present, the facility has been
oblique regular and irregular waves as calibrated in regular, and irregular uni-
well as short crested waves, sometimes directional and multi-directional waves
referred to as multi-directional waves. over a range of water depths- The
At present, the wavemaker system consists following paper presents and discusses the
of 192 segments arranged in a J results of these initial calibrations.
configuration. Mesh-type wave absorbers
are installed on the remaining sides of FACILITY DESCRIPTION
the basin. Each segmented wavemaker can
be operated in five modes of articulation The Offshore Engineering Basin, shown
to optimize its motion with respect to the in Figure 1, is a reinforced concrete
type of waves being generated. The structure 75 m long, 32 m wide with a wall
concept of this J design is to surround a height of 4 m. Water levels may be varied
model on three sides with controllable up to a depth of 3.5 m. Waves are
boundaries, thus providing a wider range generated by a 192 segmented wavemaker, of
of spreading angles in short crested
waves.
wet back design, and can be mounted in a
Consequently, these wavemakers fixed J or L configuration. Each segment
must be capable of simultaneous wave is 2 m high by 0.5 m wide. The wavemakers
generation and absorption. Initial can be operated in five modes of
calibrations of the wave generation system articulation: flapper, piston and 3
are now completed for regular waves and different combinations of flapper and
irregular uni-directional and multi- piston. These modes enable optimization
directional waves. Typical wave of segment motion for wave generation in
calibration results are presented in this deep, shallow and intermediate water
paper. Results of preliminary depths respectively. The maximum wave
investigations into wave absorption are height is 1.0 m in regular waves and the
also presented and discussed. maximum significant wave height in
INTRODUCTION
irregular waves is 0.60 m. Depending on
the water depth and wave height,
wavelengths up to 23 m can be generated.
The Offshore Engineering Basin at the Passive absorption is accomplished using
Institute for Marine Dynamics was expanded metal sheets of varying
commissioned in June 1990. Since that porosities and spacings.
time, much effort was focused on advancing The acquisition system, illustrated in
techniques of simulating environmental Figure 2 ,
uses Neff Instruments 620-500
conditions of waves, wind and current, for buffered and unbuffered analog subsystems.
use in both research and commercial test The digital converters have a 15 bit
programs. Equipment used to simulate each conversion accuracy and
of these a 50 kHz
environments is undergoing throughput with front
calibration to quantitatively determine end signal
conditioners. A VAX 3200 is used as the
their performance. acquisition
One of the main computer. The data
features of the acquisition and real time control is
facility is its capability to generate
oblique and directional waves.
supported through a software package,
When referred to as GEDAP,
generating directional waves,
(Generalized
the Experimental Control, Data Acquisition and
wavemakers are required to absorb waves Analysis Package) [1].
where they impinge on the wavemakers. As It has a
a prelude to
comprehensive set of programs for wave
this requirement,

261
synthesis and experimental data analysis deterministic and probabilistic methods.
with graphic output capability. Depending on the test requirements, one
method may have certain advantages over
REGULAR WAVE CALIBRATION the other. A sample of a measured uni-
directional spectrum is compared to the
The basin was calibrated in regular target JONSWAP wave energy spectrum in
and irregular waves. In regular waves, Figure 4. The spectrum was synthesized
the performance envelopes were found to using the Random Fourier Coefficient (RFC)
compare well with those predicted using technique and consists of 1024 wave
linear wave theory. Figure 3 shows an components. This RFC frequency domain
example of the theoretical maximum technique is analogous to white noise
attainable regular wave heights along with filtering in the time domain.
the test matrix run in a 2 m water depth. The square of the wave envelope is
Depending on the mode of important when considering the low
articulation and water depth, the frequency behaviour of floating moored
theoretical envelopes are bounded by the vessels. This is because the slowly-
following parameters: varying second-order response of the
Breaking wave steepness limit. In vessel is significantly affected by
linear waves, this is defined by: grouping patterns of waves in the time
domain. The variance spectral density of
It = 0.1*TANH(kh) (1) the square of envelope of wave groups
found in a continuous spectrum, is found
where, H is the wave height, L is the from,
wavelength, k is the wave number and CO

h is the water depth. (2)

Wave machine stroke limit. E(L) = 8 f S(f) S( f+R) df
This is the maximum mechanical
displacement of the wavemaker units.
A number of authors such as [2] and The modelled waves have a finite duration
[3], have presented transfer functions and represent a sample of the full
relating wavemaker stroke to wave spectrum. Grouping activity in the
amplitude. The limiting factor is the modelled uni-directional irregular waves
size of the generator which in turn is is described in terms of the Smoothed
normally limited by capital cost. Instantaneous Wave Energy History (SIWEH)
Wave machine velocity limit. [4]. The variance spectral density of the
This is related to the maximum power SIWEH should match the grouping spectrum
available to the wavemakers. described by equation (2) for an infinite
Wave machine over-topping limit. number of wave components.
This is calculated from the total wave Figure 5 compares the spectral density
elevation at the mean wavemaker of the square of the envelope of the waves
position due to the progressive wave found in the measured spectrum (Figure 4)
The
component plus the local evanescent to those predicted by equation (2).
wave component. measured envelope spectrum is found by
low-pass filtering the square of the
The measured reflection coefficients surface elevation using a Hilbert
within the performance envelope for the transform technique. As indicated in the
test matrix shown in Figure 3 are figure, there was good agreement.
documented in Table 1. In general,
reflection coefficients are less than 5 %. MULTI-DIRECTIONAL IRREGULAR WAVES
Waves generated in the basin were also be
Multi-directional waves can
found to be highly repeatable.
modelled using the single summation or
IRREGULAR WAVES double summation technique. The single
summation technique differs from the
double summation method in that it
There are two types of irregular wave
spectra, uni-directional and multi- produces a single frequency per direction.
In contrast, all frequencies are found in
directional. In the uni-directional case,
each direction with the double summation
all frequencies propagate in the same
direction while in the multi-directional technique. The double summation technique
case, the waves propagate over a range of also produces a wave field that is neither
ergodic nor spatially homogenous due to
directions.
Uni-directional irregular spectra can phase-locking.
be synthesized using a number of
techniques, but are generally grouped into

262
 Multi-directional wave spectra, S(o),0) units, were computed based on linear wave
are modelled by pairing an energy group velocity, and the location of the
spectrum, S(w), with a frequency wavemakers. The drive signals for the
spreading function, D(6),0), such that, wavemakers were computed using linear wave
S(w,e) = S(63)D(6),(3)
(3) theory.
Using the above technique as a first
estimate of the control signal, the
The spreading function can be either a signals were phase shifted in 5 degree
function of frequency and direction or a increments. By repeating this process, an
function of direction only. Generally a optimum phase angle with minimum
cosine square formulation is used, such as reflection was determined. This phase
zs angle was then maintained and the
(4)
D(u), 0) = C(s)[cos( (6)Q-60" )1 amplitude of the absorbing wavemaker drive
2 umax signals were decreased by 5 % each time.
A typical set of results for a 2.5 second
where, 00 is the principal direction of period wave with a steepness of 0.017,
wave propagation, C(s) is a normalizing illustrating the effects of phase and
coefficient and, s is the spreading index. amplitude adjustment of the absorbing
As in the case of synthesizing uni- wavemakers are shown in Figure 8 and 9
directional wave spectra, multi- respectively.
directional waves can be synthesized using
deterministic and probabilistic methods. CONCLUSIONS
There are also a number of techniques for
selecting the directions associated with Preliminary calibration of the wave
each wave frequency. A detailed study of generation system at the Offshore
these techniques can be found in [5]. Engineering Basin of the Institute for
Figures 6 and 7 show comparison of Marine Dynamics has been completed.
measured energy spectrum and spreading Typical results in regular, irregular uni-
with computed target quantities. Good directional and multi-directional waves
agreement was observed in both cases. are presented in this paper.
The energy spectrum was modelled using a In regular waves, the measured results
JONSWAP spectrum and the spreading from the calibration matrix were compared
function was a frequency independent, to theoretical performance envelopes. The
cosine square, version of equation (4). reflection coefficients for a number of
wave heights and periods, at various water
PRE-CALIBRATED WAVE ABSORPTION depths, using different modes of
articulation were also determined. The
Some experiments were conducted in the measured performance envelopes were found
basin using wavemakers to absorb waves. to compare well with the predicted ones
These tests were carried out at a water and typical reflection coefficients were
depth of 1.5 m using the piston mode of below 5%.
articulation. Regular waves were In uni-directional irregular wave
generated by the wavemakers on the south runs, the measured wave energy spectra
and west walls, propagated in a direction were compared to the corresponding target
which makes a 60 degree angle with the spectra. In addition, the spectral
south wall, and were absorbed by density square of the wave group envelope
wavemakers on the north wall. The wave was compared to the target SIWEH spectra.
time series were measured by an array of Both the first order waves and the SIWEH
5 wave probes and the reflection spectra compared well with the target
coefficients were computed by separating spectra. This demonstrates the wavemaker
the incident and reflected waves using a system's ability to accurately generate
least square method documented in high quality irregular waves with pre-
reference [6]. defined wave groups and wave energy.
The stroke and phase for each For multi-directional waves, the
wavemaker which is required to absorb measured wave spectra and spreading were
waves were computed in relation to the compared to the target spectra and
impinging waves. The absorbing wavemakers spreading functions. Again, good agreement
basically reverse the process of the between measured and theoretical results
generating ones. A delay time was was observed. This
implemented to prohibit the absorbing demonstrates the
capability of the wavemakers to generate
wavemakers from moving until the incoming multi-directional waves which model the
waves arrive. On the absorbing wall, the target sea state and associated spreading
delay time for start up and the phase of function.
the signals relative to the generating

263
 Initial results of pre-calibrated wave REFERENCES
absorption using wavemakers were also
presented. These results indicate that the Miles, M.D., The Gedap Data Analysis
wavemakers can be reasonably controlled to Software Package, NRC Technical Report
absorb waves using drive signals computed TR-HY-030, 1990.
from linear wave theory, as long as the
waves are not too non-linear. Results Ursell, F. Dean, R.G., and Yu, Y.S.,
showed that reflections from absorbing Forced Small Wave Amplitude Water
wavemakers are highly sensitive to both Waves: A comparison of Theory and
phase and stroke amplitude of the Experiment. J. Fluid Mech., Vol. & (1)
absorbing units. Minor phase and amplitude pp 33-35, 1960.
adjustment of the motions of the absorbing
wavemakers were found to have significant Gilbert, G., and Thompson, D.M. and
effect on the reflection coefficient Brewer, A. J. 1971 Design Curves for
measured. This suggests that the optimum Regular and Random Wave Generators,
phase and amplitude of the wavemaker Journal of Hydraulic Research, Vol. 9,
motions necessary to absorb waves are well No. 3 pp. 163-196, 1971.
defined for a particular wave frequency
and wave height. It is believed that the Funke, E.R., and Mansard E.P.D., On
effectiveness of pre-calibrated wave the Synthesis of Realistic Sea States,
absorption are influenced by evanescent NRC Technical Report LTR-HY-66, 1979.
waves and non-linear wave phenomena
related to interactions between first- Miles, M. D., Numerical Models for
order components. These effects will be Synthesis of Directional Seas, NRC
further investigated in future study. The Technical Report TR-HY-016, NRC NO.
wavemakers will now be arranged in a L 31610.
configuration to enable further
into directional wave Mansard, E.P.D., Sand, S.E., Funke,
investigation
generation which does not require active E.R., Reflection Analysis of Non-
wave absorption. linear Regular Waves, NRC Technical
Report TR-HY-011, 1985.

4.0m MAX.DEPTH 3 5

411111
5.0n
6.3n
A
SECTION A-A 1.75fi
12r1 2.9r,
3.5n BLANKING PLATES 46m

IllLÌDhlib
IlIIUlt
RI
1111 I
MI 3.0rn

11111 I.
1110111

"MI 1
IllIllillIllI
Eiliiiiiinillili111111111111111111111111111111111111111111111111111111111--
58rn

75.1 ts.urr-

A
OFFSHORE ENGINEERING BASIN
sc.," 1400

Figure 1 Offshore Engineering Basin

264
 LAN Ethernet
MCU MCU
VAX 3200
64 Mb !HMI
VAX 3200 Wave Generators
Analysis Console

Wind
D/A
NEFF 500
Converter
Plotter Current
co
NEFF AID
03
Printer Transducers
100/300 Acquisition
Terminal
Qualisys Model Tracking

Optopos Basin Tracking

Institute
VAX Cluster Telemetry Transducers

Figure 2 Acquisition System

1.0
-PERFORMANCE ENVELOPE I 0.006

NAVE MATRIX
0.875
0.005

0.75
-TARGET
N
= 0.001
- MEASURED
. 0.625

_C

o
0.5 , 0.003

o
o 0.375
3 0.002

0.25

0.001
0.125

0.0 0.0
0 0 1.0 2.0 3.0 9.0 00 0.5 10 15 20 25 30
Wove Period (seconds) Frequency Hz)
Figure 3 Regular Waves Figure 4 Irregular JONSWAP Spectrum

265
 0.00004 0.003

-TARGET -TARGET
0.00003
--MEASURED -
2 0.002
MEASURED
CV

* 0.00002 E
E

0.001
rn
0.00001

0.0 0.0
00 05 10 15 20 00 0.5 10 15 20 25 30
Frequency (Hz) Frequency (Hz)

Figure 5 SIWEH Variance Spectral Figure 6 Multidirection JONSWAP

Density spectrum

-TARGET
MEASURED

0.0 0.0
45.0 90.0 -20.0-10.0 0 0 19.0 20.0 30.0
-90.0 -45.0 0 0
Theta (degrees) Phase Adj (degree)

Figure 7 Spreading Function Figure 8 Phase Adjustment

266
Figure 9 Amplitude Adjustment

Period (sec)

Wave
Height 0.90 1.27 1.55 1.80 2.04 2.26 2.50 2.74 2.97 3.23 3.48
(m)

0.750 -- -- -- 2.3 4.5 4.3 -- --

0.675 - -- -- --- 55 7.6 3.4 2.5 3.6 -
0.600 - _ -- -- - 5.3 2.7 4.1 2.1 2.4 5.1
0.525 -- - -- -- 2.6 5.5 3.0 3.3 2.2 2.1 4.9
0.450 - - - 1.4 0.9 4.3 2.5 3.7 3.1 1.9 4.8
0.375 - -- -- 0.4 2.8 4.4 1.9 4.0 3.5 --- --
0.300 - -- 2.7 0.6 2.3 3.4 1.7 -- - -- -
0./75 --- 1.6 0.8 ---
0.150 --- 3.6 --
0.080 3.0 - -- --- - -- -- --- --- --- ---

Table 1. Reflection Coefficients Corresponding to Figure 3.

267
Two Novel Free Surface Measurement Techniques INTRODUCTION
Developed at DTMB Capacitance wire wave height probes have
P.A. Chang, III, T.J. Ratcliffe, J. Rice, S. McGuigan
(Carderock Division, Naval Surface Warfare Center, USA) been used at DTMB since the late 1970's
to measure the wave heights in the
Kelvin wake of towed ship models. A
longitudinal wave cut method, developed
by Sharmal, calculates a wave spectrum
and integrated wave pattern resistance
from a time history of measured wave
heights at a fixed transverse distance
from the towed model, with the assump-
tion that the waves are in the far
field. The longitudinal wave cut method
NOMENCLATURE has been used in various ways to opti-
mize ship hull forms, from determining
waveheight (distance from the volume and position of bow domes and
trough to crest) bulbs to affecting the characteristics
longitudinal coordinate of the transom stern. It has also pro-
time vided experimental data for the valida-
X wavelength tion of numerical free surface predic-
o Lookdown angle for laser height tion tools.
probe Recently, interest in the details
free surface displacement of the free surface in the near field
and in the centerline wake, have re-
ABSTRACT quired that measurements be made of
highly sloped nonlinear waves and, of
Capacitance wave probes have been tradi- small amplitude, high frequency waves.
tionally used at DTMB for measuring the Previous experience (see e.g.
wave heights in the Kelvin wake of towed Ratcliffe2) has indicated that capaci-
surface ship models. Two new wave height tance probes are limited to measuring
measurement systems have recently been
frequencies less than about 3 Hz, and
developed at DTMB: a laser waveheight
that contamination problems exacerbate
probe, using a linear array of charged this frequency limitation. The mechan-
couple devices, and a surface sensing ical probe and the laser height probe
mechanical probe. These devices repre-
are being developed to satisfy these new
sent improvements to the capacitance
requirements as well as to improve the
probes in the form of reduced and/or
quality and repeatability of the far-
eliminated susceptibility to contamina- field free surface measurements.
tion and increased frequency response.
In this paper we will discuss the
In this paper, the physics of op-
physical principles underlying the vari-
eration of the measurement systems and
ous free surface measurement techniques,
design considerations for their opti- and in so doing, will discuss design
mization will be discussed. Measurements
considerations. We will then show some
of the free surface in a regular wave
results that have been obtained using
field, and for longitudinal wavecuts
the new (and old) measurement techniques
with a ship-model, over a wide Froude
and discuss the various strengths and
number range, have been made. Based on
weaknesses of the devices.
experimental results, evaluations of the
The term "latching" will be used
various attributes and shortcomings of
throughout this paper and refers to the
each type of probe will be made.
collection of a data point by the
electronics (e.g. it is put into a

269
sample and hold circuit). The "data and wave pattern resistance. However,
collection rate" is the rate at which problems such as zero drift and signal
the computer collects the latched data attenuation are endemic to their usage,
making them very time consuming to use.
IllaTEIMEELITIDE It is believed that surfactants
(primarily carriage motor oil) stick to
the wires causing them to lose their hy-
Capacitance Probe
drophobic properties. This means that
The operational principle of this probe the water may slide off the probe more
is that the wire acts as a capacitor slowly, or remain on the probe. This
which changes its capacitance with im- manifests itself in lowered frequency
mersion in the fluid. The capacitance is response, low frequency bias errors, and
one component of a circuit, the compo- severely attenuated signals.
nents of which, include 1) a capacitor,
which consists of the immersed length of Mechanical Probe
the capacitance wire for which the wire
The mechanical probe, shown in Figure 1,
core and the water act as the two plates
is minimally intrusive and therefore,
of a capacitor, while the insulation
should not suffer from the contamination
acts as the dielectric 2) a resistor,
problems of capacitance wire wave
which consists of the fluid surrounding
probes. A strip chart pen servo-motor
the wire and 3) a voltage given by the drives an electrode up and down via a
difference between that at the top of
cantelivered arm. The electronics used
the capacitance wire and an immersed
to drive the servo-motor detect whether
ground wire. The waveheight is obtained
or not the electrode is in contact with
by measuring the time that it takes the
the grounded (i.e. water) surface. When
capacitor to reach a certain percentage
The time is
the probe is not in contact with the wa-
of its fully charged value.
ter surface, the electronics signals the
then related back to the capacitance and
servomotor to drive the probe downward.
indirectly (through use of calibration
When contact with the water surface is
constants) to its change in immersion.
made the servo-motor reverses direction
It is assumed in their use that
and travels upward until contact with
the capacitance wire is highly hydropho-
the grounded surface is broken. The mo-
bic so that when the waterlevel de-
tor again reverses direction and the
creases none of the exposed surface re- process begins again. The result of this
mains wetted. process is an oscillation of the probe
The capacitance wave probes used
about the water surface. The returned
at DTMB are TEFLON-coated stainless position signal is an AC signal with DC
steel cable with an overall probe diame- shifts corresponding to changes in the
ter of 0.017 in. (0.44 mm). Both the water surface position. The position of
probe and a grounding wire are suspended the probe is latched each time the probe
in the water from an electronic circuit
makes a downstroke. This reduces the
board card. The whole apparatus is held
error due to the fluid which accompanies
at a fixed point in basin, mounted on a
the probe on the upstroke. The position
wave boom which is cantilevered from the
signal is latched through a sample-and-
basin wall. hold circuit, optionally filtered to
The capacitance probe has gener-
remove voltage steps, and then collected
ally worked well at DTMB, providing the
by computer.
experimental wave height data for a mul-
The variables in optimizing the
titude of wave resistance calculations,
design of a mechanical probe are:
bow designs etc. Their strong point is 1) Acceleration. To increase the data
that they work well for the range of
collection rate, the search acceleration
wave amplitudes (0.25" < H < 6") and
of the servo-motor can be increased.
frequencies (0.5-2.5 Hz) which are im-
However, this has the effect of increas-
portant for computing free wave spectra

270
ing the overshoot. Overshoot is caused beginning of every scan and counting the
by the finite response time of the pulses needed to scan through the 1024
servo-motor and the rotational inertia element array. Once an element position
of the actuator arm. For our experiments has been latched, it is converted to a
an overshoot of about 0.1 inch gave a voltage with the use of a 10-bit D/A
data collection rate of about 45 Hz. converter, filtered, and collected by
2) length of the arm: The pen motors the computer.
have a maximum half angle of travel of Our setup employs a 50 mm f1.8
20 degrees. Thus, the arm length deter- lens from a single lens reflex camera
mines the maximum amplitude of distur- for focusing the image of the target on
bance that can be measured. However, a the CCD array. A through-the-lens (TTL)
longer arm has a higher moment of iner- focusing device allows us to look
tia, which then limits the maximum angu- through the lens and manually adjust the
lar acceleration that the motor can un- focus. The focal plane of the lens is
dergo. Examples of such problems will be approximately 2 feet from the target
discussed in later sections of this pa- with a lookdown angle, 0, of 42 degrees.
per. The lens aperture is set at f4.
Some of the important design is-
Laser Height Probe (LHP) sues that are important for optimization
This device uses a linear array of of this device include:
charged couple device (CCD) elements to Alignment. The LHP it is very sensi-
detect the location of a bright fluo- tive to the relative orientation between
rescing spot on the free surface. As the CCD array and the laser beam. The
shown Figure 2, the fluid is seeded with electronics latch the first array ele-
florescene dye; a laser beam is shot ment that exceeds a preset threshold.
down through the free surface; the laser Thus, misalignment of the beam causes a
beam excites the dye, the image of which change in the latched value. For this
is focused on the CCD array. A similar reason, a rigid bracket keeping the
device was used by Duncan3 to measure laser beam emitting device and the cam-
breaking waves. era in alignment is necessary for an
As the free surface changes in easily deployed system.
height, the fluorescent spot at the Lookdown angle. The amount of radia-
laser beam/free surface interface tion reaching the CCD array is propor-
("target") moves vertically along the tional to sin20. As the lookdown angle
laser beam. By aligning the CCD array decreases (camera closer to horizontal),
vertically with the laser beam and fo- the image on the CCD array becomes an
cusing on the fluorescent spot, the ra- increasingly long oval. Thus, for the
diation from the target will form an im- brightest image, and greatest tolerance
age on the CCD array. The change in ra- to misalignment, a larger lookdown angle
diation intensity on each CCD element is desirable. However, as the vertical
creates an electrical charge. The array displacements of the free surface become
is swept continuously at approximately large, the change in distance between
470 Hz, giving as an output an analog the target and the lens becomes signifi-
signal proportional to the charges on cant. The resulting loss of sharpness of
each array element. The position of the the image on the CCD array may cause
first element in the array for which the errors in latching the correct array
charge exceeds a preset threshold will element.
be latched and placed in memory. This Radiation absorption of CCD. The ra-
insures that only the return from the diation absorbed by the CCD elements is
surface, and not the return from beneath a function of the absorptivity of the
the surface, will be latched. The posi- CCD elements (a function of X), the ab-
tion of the active element is kept track sorptivity and emissivity of the dye,
of by clearing binary counters at the the emissivity of the laser and the out-
put power of the laser. We used a 100 mW

271
Argon ion laser in conjunction with
Rhodamine-590 flourescene dye. We found Miniature Model Basin (MMB) Experiments
that for the light levels typical of our Further experiments were conducted in
towing basin a concentration of approxi- regular waves. We tested in a section of
mately 5 ppm of dye gave a sufficiently the MMB 15 feet long, 2 feet wide and 2
bright image of the free surface. feet deep. Waves were created by a
paddle-type wavemaker which was immersed
8 inches. The mechanical and capacitance
VALIDATION EXPERIMENTS
probes were placed 8.75 feet from the
wavemaker. Our calibration setup
Longitudinal Wavecut Experiment necessitated that the CCD array, be
A longitudinal wavecut experiment was located at the same longitudinal
conducted on a low block coefficient PG position as the the mechanical and
class hull (Model 5365). The experiment capacitance probes, with the laser beam
included a wide Froude number range (the actual measurement location)
(0.25-0.48) which enabled us to make located 6.875 feet from the wavemaker. A
measurement comparisons for small ampli- beach with a surface of soft foam cov-
tude, high frequency waves as well as ered with window screen was placed oppo-
higher amplitude, nonlinear waves. The site the wavemaker at a 24° angle.
mechanical and capacitance probes were Waveheights from 0.2 to 1.0 inch were
both placed in the same transverse loca- generated over a frequency range 1.5 to
tion, 3.583 feet (1.3 model beams) from 4.5 Hz.
the model centerline, but were 2 inches
apart longitudinally. RESULTS AND DISCUSSION
The data collection and reduction
for both probes was done by our usual
methods: the probes are statically cal- Probe Performance in Longitudinal Wavecut
ibrated by raising and lowering them set Experiments
amounts with a stepping motor. The sec- Waveheights. Figure 3 is a comparison of
ond order curve fit which is applied to the waveheights measured by the capaci-
the voltage-height data takes into ac- tance and mechanical probes. Several
count the nonlinear behavior caused by trends are apparent: when the free sur-
the rotation of the mechanical probe ac- face is undergoing rapid changes, the
tuator arm. The capacitance wire is capacitance probe is always about 0.1-
essentially linear. The longitudinal 0.2 inch above the mechanical probe.
wavecut records are reduced using the However, when the rate of change de-
technique developed by Sharma' and im- creases the height measurements become
plemented at DTMB by Reed4. In this very close. Since a quadratic calibra-
method the wave record is truncated at a tion curve fit was used for the mechani-
zero crossing, before the basin wall re- cal probe, we feel that this discrepancy
flections enter the wave record. The is not due to a calibration error. In
truncated wave record is then Fourier Figure 3 the lag in height between the
transformed in transverse wavenumber two probes is smaller when the free sur-
space. The information from one period face is rising and larger when the free
of the measured transverse wave is used surface is falling. Surface tension on
to correct the spectra to account for the capacitance wire is probably the
the truncation of the wave record. The cause of this delay. The exact cause of
corrected spectra is then integrated to the difference in peak heights is not
obtain the wave pattern resistance known, although it is suspected that it
coefficient, C. The frequencies which is related to surface tension.
contribute most to Cw are typically in Limitations of Mechanical Probe.
the 0.5-2.5 Hz range. While performing well for smaller wave-
heights, the mechanical probe is limited

272
 by both wave amplitude and by high val- each sample, each was Fourier trans-
ues of the temporal derivative of the formed, the four samples averaged, and
free surface, dC/dt. Figure 4 shows a smoothed. The PSD function for the ca-
wave record from an extremely steep pacitance probe was then divided by the
(dC/dt = 80 in/sec) and high amplitude PSD for the mechanical probe to obtain
divergent stern wave for Fn = 0.48. It the transfer function. The transfer
shows 1) the probe has reached its maxi- function shown in Figure 6 is the ratio
mum half angle of travel (20°) and 2) of the responses of the capacitance and
the latching frequency is proportional mechanical probes and thus is a measure
to dC/dt. Thus if dC/dt is large the of their relative responses over the
computer may sample the same latched given frequency range. Figure 6 indi-
value more than once resulting in a cates that up to about 3.5 Hz, the two
"stepped" appearance in the wave record. probes have the same response. Above
Note that the probes were at the same 3.5 Hz the response of the capacitance
transverse location but were separated probe appears to diminish with respect
by 2" longitudinally which may account to that of the mechanical probe.
for some of the differences in the two Figure 7 shows a representative
wave records. time history for the laser height probe
Free Wave Spectra. The amplitudes compared with the mechanical probe. It
of the free wave spectra, for the shows that within the variability of the
Fn = 0.38 case mentioned above, are the incident waves, the responses of the
shown in Figures 5. These spectra repre- two probes are the same. The difference
sent the frequency range 0.4-2.5 Hz. The in longitudinal locations between the
spectra calculated from the wavecuts mechanical and laser probes accounts for
measured by the two probes, compare very the lag between the two signals. Results
well, especially for the high wavenumber for the frequencies 1.6-4.25 Hz, with
components. The discrepancy at u=0, for heights of approximately 0.5 inch, show
the Fn=0.38 case, is probably due to the similar correspondence. A transfer
height discrepancy shown in Figure 4. function between the two probes'
responses was not computed due to the
Performance of Probes in Regular Waves spatial variation in the incident waves.
The purpose of these experiments was Further experiments in larger waves and
twofold 1) to provide a suitable envi- in longitudinal wavecuts is necessary
ronment for the development of the laser for a more complete validation.
height probe and 2) to examine the re-
sponses of the probes at higher fre-
CONCLUSIONS
quencies than could be obtained in wave-
cut experiments. The regular waves gen- Capacitance wave probes have suffi-
erated by the wavemaker had significant cient frequency resolution for collect-
energy at harmonics of the carrier fre- ing waveheight data for calculation of
quency which allowed us to examine the Cw and free wave spectra. That is, they
responses over a much larger frequency perform well in the frequency range
range than is indicated by the wavemaker (0.5-2.5 Hz), and waveheight range
frequencies. (0.25- 6.0 inches). However, they have
Data were collected for a rela- been shown to have contamination prob-
tively long time, 40 seconds, at 50 Hz, lems which result in unpredictable, low
which enabled us to obtain good fre- frequency bias errors and signal attenu-
quency resolution. Power spectral den- ation. These problems are omnipresent
sity (PSD) functions were obtained by and cause substantial delays in a test
the following: the 2000 point record was programs.
subdivided into four 512 point samples, Mechanical probes perform very well in
with the appropriate number of zeroes
all but the most extreme conditions and
added, a Hanning window was applied to
do not suffer from contamination prob-

273
lems. They have a much larger range of Resistance," 22nd American Towing Tank
frequency response than capacitance Conference, pp. 408-416 (1989).
probes. Duncan, J.H., "An Experimental
The laser height probe is a non intru- Investigation of Breaking Waves
sive method of measurement and thus, Generated by a Two-Dimensional
Hydrofoil," Quest Integrated, Inc./532
cannot suffer from contamination
problems. Also, it does not have the (Jul 1991).
acceleration limitations of the Reed, A.M., "Documentation for a
Series of Computer Programs for
mechanical probes. Of the three devices
Analyzing Longitudinal Wave Cuts and
it has the most potential for
measurement of steep waves and high Designing Bulbous Bows," DTNSRDC/SPD-
frequency waves. However, it requires 0820-01 (June 1979).
that the surface layer at the laser/free
surface intersection be dyed and is
relatively more complex than the other
two probes.

RECOMMENDATIONS
Since capacitance probes work acceptably
well when free of contamination it may
be worthwhile to pursue research into
their improved performance. Areas of
research may include:
Sampling of surfactants from basin
free surfaces and physical analysis of
their effects on probe performance.
Evaluation of methods for surfactant
elimination (skimming etc.)
Evaluation of probe surface geometry
to determine how the hydrophobic proper-
ties of the wires may be improved.

ACKNOWLEDGEMENTS
Donald C. Wyatt of Science Applications
International Corporation is acknowl-
edged for his very helpful discussions
regarding longitudinal wavecut theory
and spectral analysis.

REFERENCES
Sharma, S.D., "A Comparison of the
Calculated and Measured Free-Wave
Spectrum of an Inuid in Steady Motion,"
International Seminar on Theoretical
Wave Resistance, Ann Arbor, MI (Aug
1963).
Ratcliffe, T.J. and M.B. Wilson,
"Uncertainty in the Measurement of Ship
Model Wave Profiles and Wave Pattern

274
Strip Chart Pen Servo-
Motor

Actuator arm

Electrode

Pivot Point
Free Surface

Fig. 1. Schematic of mechanical waveheight probe

Mirror

Laser
TTL
Focusing Linear
Device CCD
Laser Beam Array

Lens
Lookdown
angle

Target on
Free
surface
Free
Sur face

or(3tehr.
ti. 1110,
Flourescene Dye

Fig. 2. Schematic of laser waveheight probe

275
Fig. 3. Waveheight record from Model 5365, Fn = 0.38, measured with mechanical
and capacitance probes.

0.6
Run 5 : Fn = 0.38 : y/B = 1.3

--- capacitance
--- mechanical
1., 0.3
o

m 0.2

0.1
0.0 I I I 1 1
1

0 2 4 6 8 10
Transverse Wavenumber

Fig. 4. Comparison of wave records measured Fig. 5. Free wave spectra from waveheight
by capacitance and mechanical waveheight measured with capacitance and mechanical
probes showing amplitude limitations of probes; Fn = 0.38.
mechanical probe.

276
 1.0-
.H0 0.8
IP

w
s 0.4

0.2

0.0
2 4 6 8 10
Hz

Fig. 6. Transfer function between

capacitance and mechanical probe
responses

Fig. 7. Representative plot of waveheights measured with laser height probe

and mechanical probe; input regular waves at 2.5 Hz.

277
 Developments at
University of New Orleans Towing Tank, 1989-1992
R. Latorre (University of New Orleans, USA)

ABSTRACT
RESEARCH PROJECTS IN SHALLOW WATER
This paper presents some of the
developments One of the unique features of
in the University of New the
tank is the capability to test
Orleans Towing Tank over the period May in
1989 to May 1992. shallow water. With the tank bottom
During these three
years a number of research projects were leveled to ± lmm tolerance, it has been
possible to complete a number of shallow
initiated and completed, as well as tank water projects which are summarized
equipment brought on-line. in
Table 1.
INTRODUCTION
PROJECT I BOW WAVE IN SHALLOW WATER
In the period of May 1989 to May
1992 there have been a number Research Project 2 was begun as a
of M.S.
projects completed in the towing tank thesis by S. Wongneryang in 1989.
[1-4]. There have been The project goal was to examine the flow
three major developments: around the ship bow in shallow water.
In a series of tests, graduate student
Greg Osborne tested 10 and 15
Approval as ITTC member tank. degree
wedges and allotted the wave
Madrid, 1990. [5] pattern
measured from photographs. It was shown
Installation of deep water that as the model reached critical speed
wave
maker. November, 1990. V =JT there was the generation of a
,

Host of 23rd American Towing soliton. The tests with

Tank wedges
Conference, June, 1992. clarified the wave pattern. A numerical
study based on these experiments is
now
being pursued by a third
During the period 1989-1992, the graduate
tank usage can be roughly assigned to: student as his Master's thesis.

Teaching: 25% PROJECT II ITTC COOPERATIVE SHALLOW

NAME 3150 Lab
Research: WAIER IESTS. EXPERIMENTAL PROGRAM, 5/91
25%Masters Course - 4/92
Thesis work
Equipment installation calibration:
20% At the invitation of Dr. E. Muller
VBD, the UNO tank participated in
Contract work: 30% the
cooperative experimental program for
This short paper discusses investigating the influence of water
four depth on ship resistance.
experimental investigations of ship
behavior in shallow water.

279
TABLE 1. Research Work in Shallow Water
In a parallel effort, Dr. Kirill
at UNO.
Rozhdestvensky is extending his earlier
work [6] to study this problem. Since
Project Objective we have generated a consistent
experimental data base, it is possible
I Study of Bow Wave in Shallow
to formulate the hydrodynamic problem on
Water
the basis of matched asymptotic
II Resistance Tests of Series 60
in Shallow Water (ITTC expansions.
Cooperative Project)
Forces and Moments on Segmented PROJECT IV MEASUREMENT OF TOWED VESSEL
III
Ship/Barge near Bank or IRAJELIOR7 IN SHALLOW WATER 12/91 -
5/92
in a Canal
IV Measurement of Towed Vessel
Trajectory in Shallow
It has been shown theoretically by
the authors [7] that towing a vessel in
Water shallow water will improve its course
stability. In order to verify this
of
experimentally, a series of 100 towing
A L = 10 ft (3.048 m) model
PP tests were made in shalTow water. Using
was
a specially designed video based system
Series 60 with C, = 0.6,
the trajectory of the towed ship fitted
manufactured followingythe ITTC supplied
with skegs was recorded. The skegs were
offsets. The model was tested at four
fitted to a cargo ship as shown in Fig.
conditions. 4. Fig. 5 shows the test underway.
I. Deep water (NAME 3150 Lab 1989)
Figure 6 demonstrates a very
h/T = 2.0 interesting phenomena. As the water
h/T = 1.5 depth decreases from deep water in Fig.
h/T = 1.2 4 (Trace A) to h/T = 2.0 (Trace B) and
h/T = 1.25 (Trace C) there is a delay
The test data has been sent to Dr. in the towed vessel response so the
Muller for analysis and presentation at towed vessel is very sluggish in
the 1993 ITTC. extremely shallow water. This is also
the case when the skegs are set at a
PROJECT III ST. PETERSBURG MARITIME larger angle.
UNIVERSITY-RUSSIA AND UNO COOPLRAIIVE
PROJECT 2/92 - 4/92 DISCUSSION
In order to study the influence of The results illustrate the
gaps on the side force and moment of a capability of the UNO tank to perform
barge tow moving at an inclination in
shallow water tests. A number of
shallow water, a series of 15
commercial projects are now under
experiments were made with a 1/50 scale discussion for developing design data
2 x 3 river barge train. The model was
for shallow water excursion and
first idealized as a slender plate and commercial vessels. These four studies
then as a flat box. Fig. 1 shows the
are the basis for a number of future
test geometry Fig. 2 shows a typical
projects.
test trace. The experimental results in
Fig. 3 illustrate the reduction in side
REFERENCES
force due to distance from the adjacent
wall. Tests with the barge train in a 1. Latorre, R., "Development of Ocean
simulated channel showed a more
Engineering Test Rig," Paper 181, OCEANS
pronounced influence on the side force
89, Seattle, Sept., 1989.
and moment from the separation distance.

280
 Latorre, R., "Development of Ocean
Engineering Projects at UNO," SEG-USN Fig. 5. Photo of Towed Model Tests.
Deep Ocean Technology Symposium,
Stennis, MS, Feb. 13-14, 1990. Fig. 6 Variation of Towed Vessel
Trajectory with Shallow Water.
Osborne, G., Latorre, R.,
"Investigation of Mooring Line Tension
and Barge Orientation on Single Point
Mooring in Presence of Wind, Waves, and
Strong Current," Proceedings OMAR,
Houston, Feb. 18-23, 1990. See also
ASME Journal of Energy Resources
Technology, Vol. 113, No. 2, June 1991,
pp. 75-79.

Latorre, R., Hatamura, Y., "U.S.-

Japan Cooperative Research: Realization
of a Six-Component Force Sensor ES-2,"
International Journal of Engineering
Education (In Press).

Executive Committee Report,

Proceedings 19th ITTC, Madrid, Spain,
Sept. 16-23, 1990, Vol. 2, p. 587.

Rozhdestvensky, K., "Ship Motions

Close to a Bank or in a Narrow Channel,"
Report No. 227, Dept. of Naval
Architecture and Marine Engineering,
University of Michigan, December, 1981.

Charters, S., Thomas, G., Latorre,

R., "Analysis of Towed Vessel Course
Stability in Shallow Water," Trans Royal
Institute of Naval Architects, Vol. 12,
1986, pp. 247-258.

FIGURE CAPTIONS

Fig. 1. Test Geometry for Simulated

Barge Train near a Boundary.

A - Flat Plate
B - Box Barge 2 x 3 Train

Fig. 2. Typical Test Trace Showing the

Forces and Moments on the Barge.

Fig. 3. Test Results.

Fig. 4. Geometry of Skegs fitted to Ship

Hull.

281
 A - Flat Plate
B - Box Barge 2 X 3 Train

Figure 1. Test geometry for simulated

barge train near boundary.

282
 Two Walls (20 44 sec) Run 24BU
1 .3

1 .2

cm 1 1
Û.
_J

0.9

0.8
E

\ 0.7

0.6

o 0.5
E.)

0.4
00

0 0.3
o
0.2
0
o . 1

0.1
o 20 40 60
Run Duration (sec)
Cy My (CyLp)

Figure 2. Typical test trace showing the

forces and moments on the barge.
 CY
1.6
Fy
o
- 1/2 p V2LT1.4

V:1.5 ft s
1.2 - L.= 3.9 ft 2\ = 1/50
h/T 1.1 1/s = 0.05

0 1 2 3 inch
MY 0.6
My o
V2L2T
0.56 o
0

0.52

0 1 2 3 inch
BARGE- BARGE GAP

Figure 3. Test results.

284
 33.6 ft WL LOAD WL

STERN

SKEG
30

---15

AP FRAME180
20°
10° 1ft

°31t SKEG

33.6 WL 20ft 20ft

AP FRAME180

Figure 4. Twin skegs.

Figure 5. Photo of towed model tests.

285
 a= 13 de9 V= 2.24 ft/s

1.5

.5 -

KEY h%
1 1.25
2 2.0
3 DEEP

22 GO 92
Xo

Figure 6. Variation of towed vessel trajectory

with shallow water.

286
 SESSION III-B

NEVV FACILITIES, TECHNIQUES,

ICE TESTING, AND VALIDATION-II
 Development Studies for a Six Degree-of-Freedom
Marine Dynamic Test Facility
M. Mackay, G. Unsworth, G. Panton, and Y. Thériault
(Defence Research Establishment Atlantic, Canada)

ABSTRACT these static frame tests, completed in

mid-1991, that are the subject of the
This paper outlines development present paper.
studies for a large six degree-of-freedom
dynamic test rig proposed for the IMD The following sections give a brief
Clearwater Towing Tank. A full size pilot introduction to the MDTF, describe the
test rig, fully functional in the vertical pilot test rig and the development test
plane, was built for this purpose. Tests program, and finally outline future and
were done on a static frame in air and parallel work.
water, using a 2 m long submarine hull
model. A number of the principal test THE MDTF CONCEPT
results are presented and discussed.
Requirements for the full 6 degree-
INTRODUCTION of-freedom (DOF) MDTF called for a device
capable of imparting large amplitude rates
The six degree-of-freedom Marine and motions to a model up to 6 m in
Dynamic Test Facility (MDTF) project length, along an arbitrary trajectory at
originated with the need for dynamic model depths from deeply submerged to fully
experiments in support of new submarine surfaced [1]. Amplitudes up to ±1 m and
acquisition for the Canadian fleet and of ±1 m/s, incidence up to ±30 deg, and a
longer term operational requirements. tactical diameter of three ship lengths
While the submarine application resulted were specified. Flow interference con-
in the Defence Research Establishment siderations resulted in a tail-mounted
Atlantic (DREA) being the lead agency in sting being identified as the primary
initiating the MDTF proposal, the device model support for submarine work, with
has broader applications, including ROV/ alternative arrangements available for
AUV work, surface vessels, and offshore propulsion testing and for surface ship
structures, all of which have been accom- models.
modated in the concept. When operational,
the MDTF will be installed on the carriage The proposed 6 DOF MDTF is sketched
of the 200 m Clearwater Towing Tank at the in Figure 1 as it would be installed in
Institute for Marine Dynamics (IMD) in St. the 8x3 m test frame of the IMD towing
John's, Newfoundland.

The project was initiated with a

feasibility and contract definition study
conducted by Canadair, Montreal. This
work was summarized in a paper to the 22nd
ATTC, Reference El]. The feasibility
study recommended building and testing a
full-size development model of the MDTF to
prove the concept and to develop the
technology for implementation.

The development, or pilot, test rig

is fully-functional in the vertical plane,
but limited in the horizontal plane. The
tests were done on a static frame at
Canadair in both air and water. It is

Copyright 1992 Department of National

Defence Canada for H.M. the Queen. Fig. 1. MDTF 6 DOF General Arrangement
Reprinted with permission.
289
carriage. Two subcarriages carrying
vertical struts move transversely on rails ROLLWAY ROUNDWAY
spanning the test frame. A beam joining BEARINGS (2) BEARINGS (4)
the bottom of the struts terminates at its
forward end in the model sting support.
Independent motions of the subcarriages
(V) and struts (W) generate sway, yaw,
heave, and pitch. Actuators in the beam
provide roll (p) and, if necessary,
incremental surge (Au). The leading strut
is pivoted about its transverse axis to FAIRING
accommodate complex motions. Total force SCREW STRUT
and moment measurements are made with an
internal 6 DOF balance located near the CG
of the model. Fig. 3. Strut Assembly Cross-Section
THE MDTF PILOT TEST RIG
is shown in Figure 3; for ease of manuf-
Physical Description acture, neither the strut nor the fairing
The MDTF pilot rig was built to full are hydrodynamically faired. On the top
size and was designed to meet the full end of the screw, mounted on the strut
performance envelope in the vertical fairing, is a 36 kW servodrive (electrica-
plane. It is sketched, installed on the lly-controlled hydraulic motor) with a
static test frame, in Figure 2. fail safe brake system. The strut assem-
blies are equipped with linear position
Each vertical strut assembly con- transducers and over-travel limit
sists of a strut fairing within which the switches.
strut is supported by dual roundway
bearings fore and aft, and by lateral The forward subcarriage is mounted
linear rollway bearings. Strut motion is on transverse rails, and sone lateral
achieved with a precision 47.6 mm lead motion on the test frame was obtained with
ballscrew actuator passing down its cent- a hydraulic piston jack. The aft sub-
carriage was attached rigidly to the
re. A cross-section of the strut assembly
frame. The forward strut assembly is free
to pivot about its transverse axis in the
subcarriage, but the sting support beam is
a simple structural member. This arrange-
ment permitted evaluation of the struc-
tural and vibrational characteristics of
the design but was only fully functional
in the vertical plane.
For these tests, a 2 m long axisym-
metric hull model was constructed. The
sting was slightly tapered with a diameter
at the model of 25 percent of the maximum
hull diameter. Forces and moments were
7.4 m measured with a waterproofed strain-gauge
monobloc sting balance attached near the
centre of the model. Accelerometers were
located in the nose and tail.

In order to test with the model in

water, steel plates were welded over the
30° bottom and sides of the test frame. The
water level was at the lower end of the
strut fairing, corresponding to a depth of
I
about 3.5 m.
With modifications to the sting and
If beam, and an appropriate balance, the
pilot MDTF can accommodate a model up to
N\ \\\\\\\\\\\\\%\\\\\\\-\\\\\\\\\\\\\\\\\\\\\\\ 6 metres long. It is consequently large
height to the top of the strut fairings
Fig. 2. MDTF Pilot Rig

290
is about 7.4 m on the static test frame ballscrew actuators was the subject of
(see Figure 2) and heavy, although prior concern because to the authors'
rigorous weight control was not a design knowledge they had not, with the exception
criterion. The strut assemblies together of Reference [2], been previously used in
weigh about 2 t, and the subcarriages this type of application. In the event,
another 1.5 t. Weight saving is a ques- their mechanical operation was satisfact-
tion to be addressed in the final engin- ory; smooth operation required fine-tuning
eering design. the servo control loops. Once set up, the
actuators performed reliably for the whole
Motion Control System test program.
A schematic of the motion control
system used in the pilot MDTF is given in The hydraulic power supplies pro-
Figure 4. Only one strut assembly is vided less than the maximum required flow
shown; the other strut assembly and the rate, and limited the actuator envelope to
horizontal actuator are arranged similar- a minor extent as noted below. Nonlinear-
ly. The actuator control loops are simp- ities in the hydraulic power supply
lified, and use only position transducer resulted in higher harmonics of the drive
signals for feedback. The accelerometers frequency in some of the motions.
in the model provide comparative data.
Because of deterioration of the
Thefail safe brakes are spring railway surfaces, the lateral strut
actuated and electrically released. They bearings were removed early in the test
are applied when power is off, or during program, leaving the roundway bearings to
normal operation when an over-travel limit provide lateral stiffness (see Figure 3).
switch is tripped, by an underpressure This proved to be adequate for experiments
condition in the hydraulic system, by an in the vertical plane, but lateral loading
excessive error signal to the servodrive, tests were not performed.
or by an operator-commanded emergency
stop. TEST PROGRAM
System Performance Outline
Most components and subsystems of The main objectives of the develop-
the pilot rig are off-the-shelf non-custom ment test program were to demonstrate:
items. On the whole, they performed satis- satisfactory structural stiffness,
factorily although a few compromises were vibrational modes, and resonance freq-
made in the scope of the tests. uencies,
achievement of the required kinematic
Early problems with the servodrives envelope, and measurement of model motion
necessitated some redesign of the hydrau- and reaction forces, to acceptable preci-
lic circuits. Smooth operation of the sion under dynamic loads, and
stability and accuracy of the motion
control system.
HYDRAULIC NOTE: ONLY ONE STRUT Deflections under static loads and
POWER ASSEMBLY IS SHOWN the basic actuator calibrations were
determined in air. All other functional
tests were performed in both air and
SAFETY / water. Actuator commands were derived
HYDRAULIC from a signal generator; hence motions
CONTROL
consisted of simple harmonic oscillations.
LIMIT
SWITCHES
The dynamic test program comprised
ACTUATOR resonance/modal tests using locally-app-
lied excitation forces, servo loop stabil-
POSITION
BRAKE
TRANSDUCERI
ity with random noise applied to the strut
servos, kinematic envelope measurement,
and model reaction measurements over much
of the envelope. Additional tests were
done to try to determine selected hydrody-
SIGNAL
namic coefficients for the 2 m model.
DATA
GENERATOR HP 9000
SYSTEM (ANALYSIS) Discussion
The number of results it is possible
Fig. 4.
to discuss here is necessarily limited,
Control System Schematic

291
 tion of the model under load will arise
30 principally from sting bending as shown in
Inclinometer Figure 6. In this figure, deflection of
_ Fwd Accelerometer the model under a vertical load was at
0 20
least an order of magnitude greater than
-o
Aft Accelerometer
was measured on the beam at the base of
the struts; the difference is almost all
due to sting bending.

O Target natural frequencies of the

rig were 5 Hz or higher to keep the
dynamic amplification sufficiently small
2 -10 at the highest operating frequency.
o Natural frequencies and modes of vibration
-20 were measured in air using electro-dynamic
shakers, and in water by exciting the
servodrives and by mechanically "strum-
-30 ming" the rig. The tests were done in the
-30 -20 -10 0 10 20 30 worst-case configuration - that is, with
both struts fully extended.
Nominal Pitch Angle, deg
The lowest frequencies found were a
Fig. 5. Static Measured vs Nominal lateral bending mode of the forward strut
Pitch Angle at just over 4 Hz, and bending modes of
the sting between 2.4 and 2.8 Hz. The
but the following examples are represen- former may be low because of the absence
tative of the MDTF pilot test rig perf- of the lateral strut bearings (and there-
ormance. fore easily remedied), but sting vibration
will require further attention. Simple
Accuracy in pitch at the model is things can be done: the 2 m model was
illustrated in Figure 5 which plots nom- excessively heavy in the flooded condition
inal (commanded) steady-state pitch angle - its inertial mass should be reduced to
against values measured with an make it neutrally buoyant - and a shorter
inclinometer and with the model's accele- sting length may be possible without
rometers. The solid line represents excessive interference. Also, some vibra-
equality; standard deviation of the tion damping techniques used in static
measured values is between 0.3 and 0.4 testing [3] may be applicable. DREA has
degrees. successfully used the tuned mass damper
technique in static wind tunnel tests [4],
There was very little backlash or but in dynamic tests this would introduce
hysteresis under static loads, and deflec- a frequency dependence to the inertial
tions of the MDTF components were gen- properties of the model and considerably
erally within acceptable limits. Deflec- complicate data reduction.
The approaches suggested above will
50 alleviate difficulties associated with
sting resonance, but future implementation
E 40
will address the provision of greater
stiffness in the sting/beam subsystem.
o Balance stiffness, which was not specified
o
30 for the development studies, will also be
taken into account.
t 20
As previously noted, the achievable
o 10 actuator envelope was limited by the
o
hydraulic power supplies available for
I') 0 these studies. Figure 7 gives the target
Beam, Aft dynamic envelope with values attained in
-1 0 the course of testing shown as data
2000 3000 4000 5000 points. Limits imposed by the power
0 1 000
supplies under load (model in water) and
Vertical Load at Model, N no-load (no model, in air) are approxi-
mated by the dashed lines.
Fig. 6. Vertical Deflections under
Static Load

292
 EVALUATION AT IMD
With model
O Without model Some additional evaluation should be
Power supply limited: done before implementing a full 6 DOF
No load Under load configuration. This includes integration
E ,0 / of the device with the IMD towing tank
carriage and meeting the objectives stated
above for the static frame tests, but with
/
>7,
01;1...41
o Actuator the addition of forward speed.
o )
4:66: envelope
The pilot test rig has been shipped
to IMD for testing on the carriage in
early 1993; the experimental program will
0.1 be similar to that outlined in this paper.
These tests will also allow us to obtain
0.001 0.1 1 10 data for optimizing the hydrodynamic
Frequency, Hz configuration. For example, we expect a
significant advantage in eventually using
Fig. 7.
faired strut sections rather than those
Strut Dynamic Envelope sketched in Figure 3. This advantage is
illustrated in Figure 9 in terms of the
Finally, we consider estimation of Reynolds number limit imposed by prelimi-
the heave added mass
coefficient from nary estimates of strut drag loads on the
oscillations in heave with, and without, carriage test frame. The value 15x106 is
a superimposed angle of pitch, a. In a desirable lower limit for scaling up to
Figure 8 we have plotted the in-phase peak full scale [6].
hydrodynamic normal force Z() against the
peak acceleration 14. Added mass is repre- PARALLEL ISSUES
sented by the slope of the data - conven-
tionally it is negative. A slope pre- Sting Interference
dicted by Watt's ESAM code [5] is included The problemof sting interference is
for comparison. under investigation. Static test correc-
tions are straightforward [4], but for
Uncertainties in the data arise from dynamic tests there appear to be only
using simply the observed (low-pass filt- limited data which are appropriate to our
ered) peaks for this calculation, and application. However, there is at least
because the inertial mass of the model was one instance of dynamic testing with a
not measured, but estimated from its submarine-like form in which a sting mount
internal arrangements. Nevertheless, the has been used successfully [7], albeit
agreement shown here is quite encouraging, with fairly low-amplitude motions. We
and we expect further improvement using expect that evaluation tests with our
more rigorous experimental and analytical model at IMD will throw more light on this
procedures. subject since there are unpublished Planar

25
MDTF, a=0 o
o MOTE, oc 0 Faired Struts
x 20
_o
E
ESAM
P 15
0 Re=15x106

Pilot Test Rig Struts

o 5
1 2 3 4 5 1 2 3 4 5 6 7
, m/s2 Model Length, m
Fig. 8. Estimate of Heave Added Mass Fig. 9. Achievable Reynolds Number
limited by Strut Drag

293
Motion Mechanism (PMM) data on the same Watt, G.D., Tanguay, B., and Cooper,
hull available for comparison. K.R.: "Submarine Hydrodynamics in the Wind
Tunnel: The DREA Static Test Rig", RINA
Data Reduction Warship '91 Symposium on Naval Submarines
A major (but not the only) role for 3, London, May 1991.
the MDTF will be the determination of
hydrodynamic coefficients. A very simple Watt, G.D.: "Estimates for the Added
method of analysis has served to demon- Mass of a Multi-Component, Deeply Sub-
strate the principle for these development merged Vehicle", RINA Warship '88 Sympos-
studies, but for arbitrary, or non-harmon- ium on Conventional Naval Submarines,
ic, trajectories, we need to look beyond London, May 1988.
the well-established procedures for PMM
testing and analysis [8,9]. Feldman, J.: "Straightline and Rotat-
ing Arm Captive Model Experiments to
DREA has contracted Aerospace Engin- investigate the Stability and Control
eering and Research Consultants Ltd., Characteristics of Submarines and other
Toronto, to investigate suitable parameter Submerged Vehicles", DTRC/SHD-0393-20,
estimation methods. The first phase of November 1987.
this work has shown promising results with
the maximum likelihood approach [10] using Iitaka, H.: "Fundamental Research of
state reconstruction, or digital filter- the Dynamics and the Control of an
ing, on simulated data. The need for Unmanned Submersible for Underwater Explo-
careful experimental design is evident; ration (III)" (in Japanese), Researches of
one aspect of this is that an increased the Electrotechnical Laboratory, No. 878,
data window may be achieved by including Sakura, January 1987.
data from the acceleration and
deceleration phases of each test run. Goodman, A.: "Experimental Techniques
These studies will continue with actual and Methods of Analysis used in Submerged
data from the pilot test rig. Body Research", 3rd Symposium on Naval
Hydrodynamics, Scheveningen, 1960.
CONCLUDING REMARKS
Booth, T.B. and Bishop, R.E.D.: The
The proposed 6 DOF MDTF is a complex, and Planar Motion Mechanism, Admiralty Experi-
in some respects ambitious, project, but ment Works, 1973.
the development studies described in this
paper indicate to us that the target
Eykhoff, P.: System Identification,
specifications for it can be met. The Wiley, New York, 1974.
biggest difficulty will be achieving and
maintaining a good level of structural
stiffness over the large dynamic envelope
that is required. The project is proceed-
ing with further evaluation of the pilot
test rig on the towing tank carriage at
IMD.

REFERENCES

Mackay, M. and Walker, R.: "A Six

Degree-of-Freedom Dynamic Test Facility:
Concept and Feasibility", 22nd ATTC, St.
John's Nfld, August 1989.

Ashcroft, F.H., Troesch, A.W., and

Sullivan, P.: "Design, Construction, and
Initial Testing of a Vertical Motion
Mechanism", 22nd ATTC, St. John's Nfld,
August 1989.
Mabey, D.G., Welsh, B.L., and Pyne,
C.R.: "A Review of Rigid Body Response on
Sting Supported Models at High Angles of
Incidence", Progress in Aerospace Scien-
ces, Vol. 28, No. 2, 1991.
294
 A New Apparatus for Captive Model Testing
of Submerged Bodies in Coning Motion
E.M. Lewandowski
(Davidson Laboratory, Stevens Institute of Technology, USA)

ABSTRACT significantly expands the matrix of

possible test conditions by permitting
A new apparatus has been designed the examination of simultaneous yaw,
and constructed to measure the pitch and roll angular velocities in
hydrodynamic forces and moments on a captive model tests. Such devices have
captive model in coning motion. The been used for many years in wind tunnel
apparatus permits examination of the studies of rockets and aircraft
effects of roll angular velocity in (Reference 1); however, the CMA
combination with other angular and described herein is the first to be used
linear velocity components on these underwater. The apparatus, consisting
forces and moments. The design of the of a body of revolution model, six
apparatus is described, and some typical component force and moment transducer
results are shown. and associated signal conditioning
equipment, yaw adjustment sector and
INTRODUCTION drive shaft connector, was designed and
built at the Davidson Laboratory in
The motions of a maneuvering 1988. An extensive test program was
submersible can consist of simultaneous completed using the apparatus in
rolling, pitching and yawing; indeed, it 1991.2.3
would be surprising if the motions were
restricted to a single plane even during CONING MOTION APPARATUS
routine maneuvers, due to coupling that
exists among the various motions. To Description
predict these motions, knowledge of the A schematic sketch of the Coning
hydrodynamic forces and moments on the Motion Apparatus (CMA) is given below.
body is essential. However, there exist The driveshaft rotates with angular
at present no practical theoretical velocity cd about an axis aligned with
means of determining the forces and
moments that arise from these combined
rotations. Furthermore, the currently
available test facilities cannot
duplicate these conditions: In combined
rolling, pitching and yawing, the cross-
flow velocity varies along the body in
both magnitude and direction, whereas in
maneuvering tests employing rotating
arms or planar motion mechanisms, the
crossflow velocity vectors are coplanar
along the length of the body. As a con-
sequence, the roll-dependent hydrodynam-
ic forces and moments are not modeled
adequately in current motion simulators.
The Coning Motion Apparatus (CMA) Coning Motion Apparatus

295
the flow velocity. The model is set at sector support using shoulder bolts
the coning angle ac relative to the flow which pass through the sector support to
clamp the sector in place. The sector
by adjustment of the sector within the has a series of holes to permit
sector support. In the resulting adjustment of the coning angle ac in 2
motion, the same side of the model
always faces the axis of rotation (this deg increments from 0 deg to 20 deg.
is actually a subset of coning motions
called lunar motion) so that the veloc- Towing and Support Structure
ity distribution and the hydrodynamic The support structure consists of
a torpedo-shaped horizontal enclosure
forces and moments are steady relative
to body axes for constant forward speed containing the main drive shaft to which
The roll, pitch the sector holder is fitted. Power to
and angular velocity.
and yaw angular velocity components are the shaft is delivered through a timing
related to the coning angle, angular belt, which is driven by a 7.5 hp
velocity and roll angle about the x-axis brushless DC motor through a trans-
(0) as follows: mission to achieve 230 RPM. A dedicated
motor controller provides 0.17. speed
p = w cos ac regulation. The horizontal enclosure is
supported by three vertical cylindrical
q = -W sin
ac
cos 0 tubes covered by an airfoil-shaped
fairing. Four airfoil-shaped ribs
r= w sin ac sine
provide support for and give shape to
Design the sheet-metal fairing. At the top of
Figure 1 is an assembly drawing of the fairing, above the water surface, is
the Davidson Laboratory design. The
a horizontal plate which serves as a
apparatus consists basically of the six-
mounting base to interface the lower
component strain gage balance and sting portion of the apparatus with a towing
assembly, the model, the coning angle carriage in the Davidson Laboratory High
adjustment sector and sector support Speed Test Basin. The electric motor
assembly, and the support and towing and transmission is located above this
structure. Each assembly is discussed plate as indicated on Figure 1.
in the following sections. The horizontal torpedo-shaped
enclosure is watertight. A moisture
Balance sensor was installed in the lowest part
The six component strain gage enclosure to alert the operator
of the
balance was supplied by the Convair of water leakage. The instrumentation
division of General Dynamics (Balance cable was passed through the center of
C-12-1.35-C), who also fabricated the the main shaft into a cable storage
sting and a watertight shroud completely compartment. At the aft end of the
encasing the balance. compartment is a waterproof bulkhead
connector. A set of slip rings is
Sector and Sector SuPport located aft of the bulkhead in the dry
The sector and sector support were portion of the enclosure, through which
machined from stainless steel. The signals from the rotating balance pass
sector holds the sting as shown on (non rotating)
to the stationery
Figure 1; instrumentation cables pass From the stationery end of
apparatus.
through the center of the sting, through the slip rings, the instrumentation
the sector and in a groove along the aft cable passes up through the aftmost
side of the sector. From here the vertical tube to signal conditioners,
cables pass through the sector support which are mounted on the towing
into the drive shaft, eventually carriage. After amplification the
terminating at slip rings. signals were transmitted by cables to a
The sector is mounted in the shore-based computer for processing.

296
 A high-power timing belt was used values. It is emphasized that these
to simplify the mechanical power results pertain to an unappended body of
transfer from the motor to the main revolution; out-of-plane forces and
shaft, as well as to permit the use of a moments will be greater for an
hollow shaft with a bore of sufficient asymmetrical body.
size to pass the cable connector
through. CONCLUDING REMARKS
An absolute shaft encoder was
installed on the lower unit, just above The Coning Motion Apparatus has
the main tube, which is driven from the already produced significant data on the
main shaft through 1:1 gears. The effects of roll angular velocity in
encoder was interfaced with a digital- combination with the other linear and
to-analog converter; the output voltage angular velocity components on the
is recorded along with the balance hydrodynamic coefficients of a submerged
signals for use in determination of body. These results will be used to
angular location as well as rotation improve the fidelity of maneuvering
rate. simulations for these vehicles.
The counterbalance mass (shown
schematically on Figure 1) was installed ACKNOWLEDGEMENT
to compensate for the asymmetry of the
rotating apparatus about the rotation The author would like to
axis, to reduce vibration. The radial acknowledge the support of the Electric
location of the mass was adjustable. Boat Division of the General Dynamics
Corporation, for the design and
RESULTS construction of the apparatus.

Typical results for a body of REFERENCES

revolution model are presented on
Figures 2-5 (from Reference 2). Because Tobak, M., Schiff, L.B., and
this data is proprietary, the vertical Peterson, V.L., "Aerodynamics of
scale has been removed from these Bodies of Revolution in Coning
figures. However, they clearly illus- Motion", AIAA Journal, Vol. 7,
trate the remarkable effect of roll No. 1, January, 1969.
angular velocity on the hydrodynamic
coefficients. The table below illus- Lewandowski, E.M., "Tests of a
trates the effects of a dimensionless Submarine Model in Coning Motion",
spin rate of w' = 0.085 at a coning Davidson Laboratory Technical
angle of 14°: Report 2660, May 1991.

Z' = 2.3 Z(') Normal force 3. Lewandowski, E.M., "Tests of a

Y' = 0.23 Z' Side force Body of Revolution with a Fin in
o Coning Motion", Davidson Laboratory
M' = 0.74 M(;) Pitching moment Technical Report 2670, December 1991.
N' = .41 M' Yawing moment
o 4. Malcolm, G.N., and Schiff, L.B.,
"Recent Developments in Rotary-
where Z', M' are the values at zero spin Balance Testing of Fighter
o o
rate (the corresponding values of Y', N' Aircraft Configurations at NASA
are zero). For a submersible with a 36 Research Center", Unsteady
ft hull diameter moving at 15 knots this Aerodynamics-Fundamentals and
dimensionless spin rate corresponds to a Applications to Aircraft Dynamics,
roll angular velocity of 6.65 deg/sec AGARD Conference Proceedings
which is within the range of observed No. 386.

297
Rail

Spin axis

FIGURE I CONING MOTION APPARATUS

Y.

0.02 0.04 2.06 0.08 0.10

-0.10 -0.08 -0.06 -0.04 -0.M 0

md/20

BEHAVIOR OF NORMAL FORCE COEFFICIENT WITH FIGURE 3 BEHAVIOR OF SIDE FORCE COEFFICIENT WITH
FIGURE 2
SPIN RATE AND CONING ANGLE SPIN RATE AND CONING ANGLE

298
 u u 000000
Coning aagla:
M' N'

0 -6'

-0 10 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0:10 -0.10 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.10
o' odd/2V
aod/2V

FIGURE 4 BEHAVIOR OF PITCHING MOMENT COEFFICIENT, FIGURE 5 BEHAVIOR OF YAW MOMENT COEFFICIENT WITH
WITH SPIN RATE AND CONING ANGLE SPIN RATE AND CONING ANGLE

299
 A Standard Method for the Conduct and Analysis
of Ice Resistance Model Tests
D. Spencer (Institute for Marine Dynamics, Canada)

ABSTRACT Ice density, kg/m3.

of Flexural strength of ice, N/m2.
A standard method for conducting ice
resistance model tests as adopted by the INTRODUCTION
Institute for Marine Dynamics is
described. The standard procedure Resistance in level ice has
involves splitting the total resistance traditionally been one of the main
into components; ice breaking, clearing, performance criteria for
buoyancy and open water. This allows each evaluating
vessels operating in ice covered waters.
component to be treated separately when Physical modelling of ship performance in
scaling to full-scale. The analysis ice has led to the construction of several
procedure is then outlined showing how ice basins throughout the world. The
dimensionless numbers are used to scale primary objective of these facilities is
the results to full-scale. The basis of to provide design information for a
the analysis procedure is also given prototype. To achieve this goal, scaling
showing how each component of total laws need to realistically describe the
resistance is affected by model and ice physical processes. Also, it is necessary
parameters. The result is a rational to validate these models by conducting
basis for the conduct and analysis of ice full-scale tests in ice. The ice tank at
resistance model tests. the Institute for Marine Dynamics (IMD) is
one of the newest and most advanced in the
NOMENCLATURE world and has allow it to develop
a,b techniques for the conduct and subsequent
Intercept and slope of C, vs Sn analysis of model test data that are
curve. realistic and accurate. This paper will
Model or ship beam, m. describe some of these techniques and
Cb Buoyancy coefficient. their experimental justification.
C, Breaking coefficient.
C, Clearing coefficient. IMD's Model Ice Basin
Thickness Froude number. The ice basin at IMD is the worlds
Gravitational constant, 9.806m/s2. longest with a usable test length of over
111 Ice thickness, m. 70 m [1]. The water depth is 3 m and the
m, Mass of an ice element, kg. tank has a width of 12 m. IMD uses a
Rn Buoyant Or ice submergence refrigerated model ice containing Ethylene
component of ice resistance, N. Glycol [2] as the main dopant to reduce
R, Breaking component of ice ice strength. Ice sheet growth commences
resistance, N. with a wet seeding technique while the air
R, Speed dependent or dynamic ice in the basin is about -20°C. This
clearing component of ice generates fine ice crystals which land on
resistance, N. the water surface and nucleate growth
Rn Total clearing component of ice resulting in a fine grain ice with a
resistance, N. columnar structure similar to that found
R,.. water resistance, N. in nature. After the ice has grown to the
R, Total pre-sawn ice resistance, N. desired thickness it is gently "warmed" or
Total ice resistance, N. tempered at +1° C. This causes internal
r,s Intercept and slope of C, vs F melting in vicinity of trapped glycol
curve. pockets reducing ice strength. Ice
Sn Strength number. strength is continuously monitored during
Model or ship draft, m. this phase and when target strength is
Model or ship speed, m/s. achieved model testing commences. To date
Density difference between ice and
IMD has grown over 600 ice sheets and
surrounding water, kg/m3. tested over 40 ship hull forms.

301
RESISTANCE TESTS IN MODEL ICE resistance. To estimate this component
two ice resistance tests are conducted; a
Model Preparation level ice resistance test in a virgin ice
Ice resistance tests are conducted sheet and, a test in which the ice has
by towing the model at constant speed with been pre-sawn to remove ice strength
propeller(s) removed. Rudders may or may effects. Together these tests divide the
not be present depending whether they total ice resistance into two fundamental
would interfere with the flow of ice. IMD components; a breaking component and an
uses a rigid towing system with the model ice clearing component. The assumption
only free to roll, pitch and heave, see here is that the pre-sawn tests contains
Figure 1. Tests have been conducted with all ice clearing forces, thus subtracting
rigid and soft (free to surge) towing it from the total leaves only a breaking
systems, and have shown no discernable component.
difference in results. The model is towed
at its centre of gravity so the towing Rbr=R_Rp (1)
apparatus does not interfere with model
motions. However, towing at the centre of This assumption was verified in two ways.
gravity as opposed to thrust block does First, pre-sawn tests were conducted over
induce a bow trimming moment. a wide range of ice strengths. These
results indicated that the pre-sawn
resistance had no ice strength
dependence[3]. The second method was to
conduct level ice resistance tests at
several strengths and from the resulting
resistance - ice strength curve estimate
where the zero strength intercept lies in
relation to the pre-sawn ice resistance,
see Figure 2.
150 Legend

+ Level Ice
0 Pre-sawn Ice
Fig. 1. Schematic of standard model
towing arrangement. R 100

Ballasted is placed in the model so

it has correct draft, trim and metacentric
height. Ballast weights are also adjusted
to provide correct longitudinal and N
transverse radii of gyration. 50
Model surface preparation is of
great importance as it controls the Ice Thickness: 46mm
friction coefficient between hull and ice. Model Speed: 0.88m/s
IMD uses IMRON, two component epoxy paint
with special additives to vary surface o
roughness, hence hull-ice friction.
IMD an extensive ship-model 0 10 20 30 40
At
correlation program is being conducted and
results indicate that the dynamic
coefficient of friction on the model
Ice St rength(kPa)
Total Resistance versus ice
should be about 0.05 to correlate well Fig. 2.
with reasonably smooth ships operating in strength.
snow free conditions.
In this Figure, the measured presawn ice
Components of Level Ice Resistance resistance is also shown, with an assumed
Since ice breaking is a complex zero strength. This plot demonstrates
process the strategy used by IMD is that the presawn resistance appears to
separate ice resistance into components coincide with the intercept. It also
be studied and treated demonstrate that total ice resistance
which can
separately [3]. varies linearly with ice strength. Thus,
by subtracting the pre-sawn ice resistance
Breaking Component. The first component from the total level ice resistance yields
that can be separated out is the breaking a breaking component that is strength

302
 dependent while the pre-sawn ice In Figure S normalized breaking resistance
resistance is strength independent. is plotted against model beam. It shows
However, the breaking component is also that the breaking resistance did not
speed dependent as can be seen in Figure increase significantly with beam.
3.

500 3

400

2
4-
300
Rbr

200 Louis St. Laurent dth

N Model Speed. 1.0m/s
Ice Thickness: 98mm
100

0 .25 .50 .75 1.0 0 .500 1 1.50

Model Speed (m/s) Model Beam (m)

Fig. 3. Breaking Resistance versus Model Fig. 5. Normalized Breaking Resistance
Speed. Ice Strength = 30 kPa. versus Model Beam @ Vm = 1.0m/s.
This occurs because of hydrodynamic Clearing Components. Pre-sawn resistance
pressure ahead of the model, which itself has several components;
provides additional support to the ice hydrodynamic, ice block rotation,
sheet requiring the model to exert acceleration and submergence.
additional downward force to break it. Hydrodynamic resistance can easily be
In the past, it was generally removed by subtracting the open water
assumed that the breaking resistance resistance from pre-sawn leaving a net
varied linearly with beam. To test this clearing resistance due to ice forces
assumption tests were conducted using only.
three simple hull forms, as shown in
Figure 4, with beams of 0.7, 1.0 and 1.3 (2)
m [3].The simple shape allowed waterline
angles and draft to remain constant while In fact, the open water resistance may not
varying the beam. be the same when an ice cover is present
thus equation 2 may not be strictly
correct. However, since the open water
resistance is often only a small part of
450 the total, small errors in it do not
1.3 significantly alter the final result.
This ice clearing term, equation 2,
contains both dynamic and static forces.
The static forces are due to ice buoyancy
50 acting on the model forebody. At very low
model speed, dynamic forces are negligible
and only buoyancy and associated friction
remain. Thus, by conducting pre-sawn test
at low or creeping speed the static
buoyant resistance may be estimated.

Fig. 4. Bow of Simple Hull Forms. 1.0 m

Rb 0.02mis,
(3)
Beam not shown for clarity.
This component should be proportional to

303
the buoyancy of submerged ice, as in this resistance is proportional to the
Figure 6. If the depth to which the ice mass of ice cleared, or pBh, In Figure 8
can be submerged is proportional to model this resistance, normalized by ice
draft, T, then ice buoyancy is thickness, is plotted against model beam.
proportional to Ap,g8h,T. It shows that the effect of beam is indeed
approximately linear.

2500

Model Speed: 0.25m1s

2000

R1 1500

1000

Fig. 6. Schematic of Ice Buoyancy acting

on a bow segment. 500

In Figure 7, this is plotted against

creeping speed pre-sawn resistance for a o
model in ice of different thickness and O .500 1 1. 50
density.
150
Model Beam (m)
47mm e 870 kg/m Fig. B. Velocity Dependent Clearing
Resistance versus Model Beam.
Thus, total ice resistance has been
100 four distinct
empirically divided into
Speed components; hydrodynamic, breaking,
P. 0. 02 m/s submergence, and dynamic clearing.
Creeping = Rao, + Rbr + Rb + R (5)
50
An advantage of having resistance
separated into components is that
47mm 8 930 kg/m
different areas of the hull influence each
component differently. For example, the
breaking resistance is primarily affected
by the waterline region of the forebody,
o 25 50 75 while the buoyancy is integrated over the
entire forebody. Also, designs may be
ApigBhj optimize by comparing relative performance
Fig. 7. Creeping Speed Pre-sawn on a component by component basis.
Resistance versus Ice Buoyancy. Comparison between different designs
and sizes is made much easier if
It clearly shows that the observed dimensionless coefficients are used.
resistance is due to ice buoyancy.
Subtracting buoyancy, R, from the net Dimensionless Numbers. Dimensionless
clearing resistance, R, yields a speed coefficients are used to scale model data.
dependent or dynamic clearing component. Dimensionless coefficients are convenient
for two reasons; they allow scaling to be
= Rb - Rb (4) made at conditions other than those tested
and, spurious or erroneous data points are
more readily identified.
This component undoubtable represents the breaking resistance,
several processes which, to date, have not Both
However, because equation 1, and the dynamic clearing
been further resolved. were made
of its dynamic nature, it is expected that resistance, equation 4,

304
 dimensionless by dividing them by a term Linear regression is used to smooth this
which is proportional to the rate of data and find its slope, b, and intercept,
change of momentum, or impulse, of the a. This implies that the breaking
broken ice, equation 6. coefficient is related to the strength
number by;
d(my)
= V = Vp,13h,V = p,Bh,V2 ( 6 ) C, = a Sfl"' (10)
dt dt
If b were exactly 2 it means that, R, =
This defines a breaking coefficient, C,; aafh. However, generally b is somewhat
less than 2, in part because the breaking
R, resistance also has a speed dependence,
C, = (7) albeit weak. The linkage between speed
p;BhiV2 and ice thickness in the strength number
mean that the influence of speed somewhat
In a similar manner a clearing coefficient decreases the influence of ice thickness.
is defined as; To resolve this dilemma work is underway
to split ice breaking resistance into a
strength and speed dependent components by
C, = (8) estimating the breaking resistance at zero
p,Bh,V, speed.
Equation 8 defines a ratio of
If the breaking coefficient is plotted on clearing resistance to ice impulse. If
a log-log scale against a dimensionless all the clearing resistance were due to
strength number defined as; acceleration of ice pieces from rest to
some speed proportional to model speed it
V would be expected that C, would be
Sn = constant. This would mean that the
(9)
clearing resistance, R, would be

N
-
I cfh,

pB
proportional to speed squared.
most test results suggest
However,
that the
clearing resistance increases, more or
a linear curve results. Figure 9 shows less, linearly with speed. Hence the
such a plot for the three simple hull form clearing coefficient is a function of
described above. speed and is plotted against a
characteristic speed, the thickness Froude
number.

Legend V
F, = (11)
+ 0. 7 m Beam Ngh,
6
0 1.0 m Beam
0 1.3 m Beam If plotted on a log-log basis the
resulting curve is linear so that;

4 C, = rFh- (12)

where r and s are the intercept and slope

3
of the least squares regression line.
Finally, plotting the buoyancy
component, R, against the ice buoyancy
yields a linear curve, as in Figure 7.
The slope defines a buoyancy coefficient,
Cb.For a single ice thickness, plotting
is not feasible, so C is defined as;
-4 -3 -2 -1 1 R,
C, = (13)
ApBh,T
Log Sn
Fig. 9. By using these three dimensionless
Lo Breaking Coefficient versus Ln coefficients, breaking,
Strength Number for Simple Hulls with Beam clearing and
of 0.7, 1.0 and 1.3m. buoyancy components are related to speed

305
and ice mechanical properties. This cut laterally with a spacing equal to 3
allows model resistance data to be scaled times the ice thickness.
or corrected to target ice conditions.
Open water resistance is scaled following
standard ITTC practice and added,
resulting in a total ship resistance.
Pre-sawr
STANDARD TEST PROCEDURE Ice Tract
Standard procedures for conduct of
ice resistance tests is continuously
evolving and this latest standard [4]
reflects IMD's current practice. As more
knowledge of the ice breaking is acquired,
and better testing techniques develop,
updated standard procedures will follow.
Procedures used to collect ice mechanical
properties are not described here but may
be found in rererence 1.
Immediately prior to towing the
model through level ice an ice flexural
strength survey is made. Typically, three Fig. 10. Schematic of Pre-sawn Ice Test
or four locations down the length of the Conducted Parallel to Level Ice Channel.
tank are sampled using six to eight
cantilever beams at each location. The The pre-sawn resistance test is conducted
ice strength obtained is time corrected to at the same speeds as in level ice but
the actual moment the model is towed with an additional slow or creeping speed.
through the ice. At this time, an ice Once again after the test a ice thickness
density measurement is also made. survey is made and ice density recorded.
Immediately after the level ice test an
ice thickness survey is made by measuring STANDARD ANALYSIS PROCEDURE
ice thickness on each side of the channel
every two metres. Plots are made of ice The first step in the analysis
thickness and strength versus tank involves calculating mean resistance and
position so that mean ice strength and model speed. From the position of the
thickness can be determined for the model in the tank ice thickness and
precise section of tank where the model strength is also found. Next the open
resistance was recorded. water resistance is removed from the
The ice-hull friction coefficient is measured presawn resistance leaving a
regularly measured on a special jig which clearing resistance, equation 2. By
uses a board painted at the same time, and plotting this against model speed the low
in the same manner, as the model. This speed clearing or buoyancy may be
flat board is pulled under an Ice sample estimated. If the model has been tested
at constant speed. Normal load, N, is in several ice thickness the buoyancy
deadweights and the resistance may be plotted against ice
provided using
tangential load, F, on the ice sample is buoyancy. C, is the slope of the
measured by a load cell. The normal load resulting curve. In the case of only one
is incremental increased, typically to 300 ice thickness, equation 13 is used. Using
Newtons on a 0.14 m x 0.14 m ice sample. this coefficient, R, is then calculated
The dynamic friction coefficient, Ad, is and subtracted from the net clearing term,
the slope of the resulting F-N curve. as in equation 4. Clearing coefficient,
After the level ice test, supports equation 8, and thickness Froude number,
are placed across the open channel so it equation 11 are then plotted on a log-log
does not close during pre-sawn ice scale. Linear regression of this curve
testing. The model is then moved yields slope, s and intercept r. These
laterally 3 m so it can be towed on in a are used in equations 8 and 12 to re-
track parallel to the previous test. This calculate the clearing resistance. Next,
track is sawn prior to testing in a model pre-sawn resistance is computed
chevron pattern similar to that observed using C, C, and open water resistance for
in level ice, see Figure 10. The pattern the ice thickness, and density and model
usually consists of five longitudinal speed which prevailed during level ice
cuts, one at the centreline, two slightly tests. This resistance is subtracted from
larger than the beam and two at an the total level ice resistance yielding a
intermediate distance. Next, the ice is breaking resistance, equation 1. The

306
breaking coefficient, equation 7 is then
plotted against the strength number,
equation 9, on a log-log basis. Linear
regression smooths the data and the slope,
b, and intercept, a, of the resulting line
is used to evaluate Cbr
At this stage, the original model
resistance data is plotted against that
predicted by the above coefficients. If
the agreement is good then these
coefficients represent a numerical model
describing either model or ship
performance. Knowing certain ice
properties and speed the Froude number and
strength number are calculted, hence the
clearing and breaking coefficients. The
buoyancy coefficient is a constant. From
these, the three components of ice
resistance are calculated.

SUMMARY

In summary, the total resistance of

a vessel proceeding in level ice has been
divided into four components. Each of
these components relate to ice and hull
paramters differently. Tests developed by
IMD are designed to quantify each
component and couple it to relevant ice
and hull parameters. The result is a
methodology which has rational basis
founded on experimental findings.
ACKNOWLEDGMENT

The author would like to acknowledge

Dr. Bruce Colbourne for his contribution
to this work.

REFERENCES

JONES, S.J., "Ice Tank Test

Procedures at the Institute for
Marine Dynamics", NRC/IMD LM-AVR-
20, November, 1987.
TINCO, G.W., "EG/AD/S: A New Type
of Model Ice For Refrigerated
Towing Tanks", Cold Regions Science
and Technology, 12, 1986, pp 175-
195.

COLBOURNE, D.B., "A Three Component

Method of Analyzing Icebreaking
Resistance", PhD Thesis, NRC/IMD
IR-1989-07, October 1987.
SPENCER, D.S., JONES, S., and
COLBOURNE, B., "A Proposed Standard
Method for Conduct and Analysis of
Ice Resistance Model Tests",
NRC/IMD LM-1992-02, February 1992.

307
 CD Model Ice - Preliminary Observations on the
Effect of Density on Tank Testing
D. Spencer (Institute for Marine Dynamics, Canada)

ABSTRACT 12, Mean propeller torque while milling

ice, N-m
A brief description of the apparatus Mean propeller torque in open water
for producing controllable density (CD) at same shaft and model speed as
model ice is given. Ice density is shown for Q, N-m
to affect nearly all aspects of ice tank V Volume of a small piece of ice, e.
testing. Some preliminary results are
given for
Fraction of air in ice by volume.
ice-propeller interaction, AQ, Relative increase in
presawn ice resistance, propeller
rubble ice torque due to ice milling, %.
resistance and level ice resistance. The Density of air, kg/e.
level of ice-propeller interaction is P.
reduced using low density PC Composite density of ice containing
level ice
primarily by increasing the minimum model trapped air bubbles, kg/e.
speed before ice in ingested into the Pi Density of natural model ice, kg/e.
propellers. This leads to a significant
reduction in propeller milling at low INTRODUCTION
speed. Rubble ice resistance is shown to
be dominated by the ice buoyancy with an There has always been some debate
approximate 400% increase in resistance as and speculation on the effect of density
ice density was reduced from 930 kg/m' to in model testing. Several modelling
750 kg/m3. Presawn ice resistance has a materials have been developed in which the
buoyancy dependent component which is density may be controlled. However, with
important at low speed. these materials other parameters are
Ice buoyancy is
seen to significantly increase this altered also, which clouds the effect of
component. The effect of ice density on density alone.
level ice resistance is quite insidious in Until recently [1] refrigerated
that it can alter the icebreaking process model ice had an inherent problem in
significantly increasing the resistance of correctly modelling ice density. In an
certain bow forms. The increased buoyancy attempt to correctly scale the flexural
of the broken ice cusps causes them to and other strength parameters model ice is
float higher as they are rotated by the gently "warmed". This leads to internal
advancing bow. These pieces may then have melting in the vicinity of dopant pockets
a higher incidence of jamming between the which were trapped in the growing ice.
hull and the unbroken ice sheet leading to This internal melting reduces the strength
crushing events and large loads in the but the relatively dense melt water also
shoulder region. increases the density of the ice. For
example, natural sea ice has density of
NOMENCLATURE about 910 kg/m3 while typically it is
between 930 to 940 kg/m3 for model ice.
Fx Horizontal component of total force Although, at a glance, this difference
acting on a structure, N. seems small it must be remembered that ice
Total resistance of a model or ship buoyancy is proportional to the density
proceeding in level ice, N. difference between ice and surrounding
Rr Ice resistance of model proceeding water. Sea water typically has a density
in pre-sawn ice, equal to Rb+Rc+Ro, of 1025 kg/e, thus a submerged ice piece
Ft, Speed dependent clearing component of one cubic metre has a buoyancy of 115
kg. Modelling tanks solutions typically
of ship/model resistance, N. have a density of
Rb Speed independent or ice about 1000 kg/m3
submergence component
resulting in an ice buoyancy of 60 to 70
of ice
resistance, N. kg, approximately half that of sea ice.
Row Open water Obviously, as an icebreaker or
resistance of
ship/model, N. structure submerges ice, part of the load
or resistance is due to buoyancy. In the

309
case of a downward breaking structure the function of the fraction of air it
rubble formation in front of the structure contains, equation (1).
provides an upward force to the underside
of the incoming level ice inhibiting pcV = (1-x)Vp, + x\ip (1)
breaking and resulting in higher loading.
As an icebreaking ship proceeds, ice By cancelling V and because p,>>p, equation
is cleared under the bow. This submerged (1) reduces to;
ice generates buoyant forces normal to the
bow which is resolved into ship pc = (1-x)p, (2)
resistance. In addition, ice buoyancy
acting perpendicular to the motion has a It is evident that density reduction is
frictional resistance component. An proportional to the fraction of air
extreme example of this type of resistance trapped in the ice. Equation (2)
is an ice breaker penetrating an demonstrates that the highest density that
unconsolidated rubble field. Increased can be achieved is the "natural" or un-
ice buoyancy has a positive aspect also. bubbled (x=0) density, 13,. At the other
After the ice is submerged by the hull it end of the scale if x is large (ie. close
has a tendency to float up along the to 1) the composite density is very low.
sides. If the ice buoyancy is too low However, in practice the minimum density
this tendency is much reduced and the ice that can be achieved and still have "ice"
eventually becomes ingested by the is about 700 kg/m3.
propellers increasing shaft torque, and
power. A final aspect of the effect of
density on ice modelling is concerning
pack ice loads. The ability of pack ice
to transfer load to a structure or ship is
to some extent controlled by the static
stability of the individual ice pieces.
High density ice has low freeboard, hence
as they start to tip they quickly flood
and become unstable. As they do the pack
begins to rubble which leads to more
instability. With the development of CD
ice by the National Research Council of
Canada the effect of density alone could
be explored and this paper presents some
initial findings.
DESCRIPTION OF APPARATUS

A complete description of the Fig. 1. Vertical Thin Section of CD model

apparatus is presented in reference [1]. ice showing trapped air bubbles.
The basic principle for ice density
modification is to produce an ice that The apparatus that generates the
contains a small fraction of air trapped bubbles is a series of long stainless
with in it. The refrigerated EGADS model steel tubes with 38 pm diameter holes at
ice [2] used by the Institute for Marine a spacing of 5 mm. Air is fed to the
Dynamics (IMD) has a columnar structure. tubes at low pressure generating a line of
That is, the ice grows from the top down bubbles across the 12 m tank width. These
with long "dendritic" ice crystals. tubes are mounted on a carriage that
Ethylene glycol(EG) is trapped within traverses the usable length (72 m) of
these crystals as they grow. When the ice tank. In doing so, the entire under
is gently warmed at about 0' C the glycol surface of the growing ice sheet is swept
internal melting which with air bubbles. The bubbles are
pockets cause
results in a reduction in ice strength. generated at a constant rate and density
To reduce ice density tiny air bubbles are control is achieved by varying the amount
released at the bottom of the tank. These of time the apparatus is in use. This is
bubbles rise to the surface and become done by controlling the "wait period"
trapped under the ice sheet. As the ice between passes under the ice sheet.
continues to grow the bubbles become Overall run time is a function of the
incorporated into it. The result is a final ice thickness desired. It has been
composite of ice and air, Figure 1. found [1] that bubbled ice has a higher
the same "warming" or
By considering a small volume of this ice strength for
This led to a
it is easy to calculate its density as a tempering time.

310
 modification of the apparatus so that it effect on the breaking or flexural
bubbles the top and a bottom layers only component of ice resistance. But model
resulting in an improvement in ice tests on a 1/30 scale model of MV Arctic
flexural properties, as well as density. in CD and ordinary EGADS ice demonstrate
Many of the test results herein apply to quite different ice hull interaction [6].
broken ice or rubble so flexural
properties are of minor consequence.
Another, less quantifiable, improvement is Louis St. Laurent - Rb+Rc
the general ice appearance and visibility. 200
The bubbles make the ice "whitish" and
opaque which makes for better viewing
above and below the surface.
Since its development, about 40 CD 150
ice sheets have been grown at IMD allowing
the Institute to pioneer research into the
effect of buoyancy on model testing. The 100 Rb
next section present some preliminary and
interesting findings and opens the way for
increased knowledge into icebreaking Legend
processes. N 50
+ 880 kg/m3.
RESULTS 0 935 kg/m

o
Level Ice Resistance
In an level ice resistance test the .250 .500 .750 1
model is towed at constant speed through
uniform level ice. There are many
components to this resistance. First, the
Model Speed (m/s)
inclined bow rides up on the ice inducing
a flexural failure. The resulting pieces Fig. 2. Clearing and Buoyant Resistance
are then rotated and accelerated downward. on 1/15 Scale Model of "Louis St. Laurent"
As the pieces become submerged buoyant
forces are generated which resist motion. Figures 3 and 4 show time series plots of
At IMD extensive use of a pre-sawn [3,4] model resistance for very low density CD
test is made to separate the initial ice and "natural" density model ice
breaking term from the subsequent ice respectively. Of particular interest in
clearing forces. In this test the ice is these plots are two very high resistance
"pre-broken" by sawing it in a chevron events in Figure 3 at about 75 and 100
pattern similiar to that seen in level ice seconds. Examination of the video records
experiments. This removes the breaking for that test indicated that broken ice
component leaving a clearing resistance pieces had become jammed as they rotated
component only. As mentioned previously, tangent to the bow surface.
this clearing force itself has many
components, but of these only the buoyancy
is not speed dependent. Thus, at low or Level Ice - Low Density
creeping speeds, where inertial forces and 1000
hydrodynamic forces are small, only
buoyancy and associated friction are
present. In Figure 2, this clearing R 750
resistance is plotted against model speed pre-sawn
for a 1/15 scale model of the Louis St.
Laurent [5]. 500
Both curves have the same
slope but the low speed intercept is N
higher for the low density CD ice. As one 250
would expect the change in the buoyancy
resistance appears to be linearly
dependent on ice buoyancy. It also
apparent from this plot that the buoyancy o 50 100 150
resistance is a fairly significant part of
total clearing resistance, and model tests
have traditionally under-predicted this Time (seconds)
constituent. Fig. 3. Time Series of Ice Resistance for
At a glance it may be speculated 1/30 Scale Model of "MV Arctic" in Low
that ice buoyancy should have little Density Model Ice.

311
 Level Ice - High Density added to increase rubble depth and then
the rubble was re-compacted. These tests
1000 were conducted in ordinary model ice with
a density of about 940 kg/m3 and in CD
750
model ice with a density of 750 kg/m3.
R It
These results are plotted in Figure 5.
shows the mean rubble resistance versus
500 ridge depth for both ice types.
N
250
Rubble Resistance
1000

o 50 100 150
750

Time (seconds)
Fig. 4. Time Series of Ice Resistance for 500
1/30 Scale Model of "MV Arctic" in High
N
Density Model Ice.
250
Additional ice buoyancy forced ice pieces
up, making it more difficult for the hull
to submerge them, thus, in lower density
ice there is a higher probability of 0
jamming. A direct consequence of this is 100 200 300 400 500 600 700
0
that traditional high density model ice
does not realistically simulate this
aspect. Also, certain hull forms which Rubble Thickness mot
are prone to ice jamming may unduly
benefit from model tests in high density Fig. 5. Resistance versus Rubble
model ice. Another aspect of level ice Thickness for a Model at two Ice
testing that may be influenced by ice Densities.
density is piece size. When an elastic
beam is deflected on a water foundation It is clear that the CD ice has had a
part of the restoring moment is due to its dramatic effect, increasing the resistance
buoyancy. However, if the deflection is by a factor of about four which is
such that the tip of the beam becomes remarkably close to the ratio of ice
flooded this hydrostatic moment can no buoyancy per unit volume, given by
longer increase with deflection. This equation (3).
moves the maximum moment toward the root
of the beam resulting in a bigger cusp. (P. - Pc) (1000 - 750)
It appears that CD ice does produce = 4.1 (3)
smaller pieces but to date this has not be (P. - (1000 - 940)
quantified.
From this it is apparent that nearly all
Rubble Ice Resistance unconsolidated rubble resistance is due to
In rubble ice resistance experiments ice buoyancy and that any tests of this
the ice sheet is broken up into pieces nature must be conducted at the correct
with dimensions about 2 - 5 times the ice density, or if not, empirical density
thickness. These pieces are then confined corrections applied.
to produce rubble with a depth typically
several times the parent ice thickness. Structural Loading
This rubble field may be left for some Ice loading on a downward breaking
time at a temperature below freezing to structure [7], as seen in Figure 6, is
form a consolidated top layer. If the similar to that for a ship except that in
ridge is left un-consolidated the main the case of a ship the forward motion of
forces acting on a penetrating model may the vessel clears the rubble from in
be due to ice buoyancy. Model tests were front. In the case of structure the
conducted on a 1/30 scale model of MV rubble is pushed ahead under the incoming
Arctic penetrating rubble fields of intact ice sheet. Thus, in the case of a
various depths (6). After each downward breaking structure the ice sheet
penetration through the ridge more ice was is supported, to some extent, by the

312
 rubble below. Also, the structure floats up and is ingested into the
experiences forces due to internal rubble propeller. The net effect of propeller
friction. ice interaction is an increase in torque
and power, and a loss of propeller thrust.
In general, there is a increase in
propeller ice interaction with increasing
speed, ice thickness and strength[8]. The
ice trajectory from the bow to midship is
relatively stable [9], but may become
unstable somewhere between midships and
the propeller disk. The probability that
the flow may become unstable dramatically
controls the amount of ice entering the
propeller. Jones et al.[10] using
synthetic ice showed that ice density has
a major role in inducing this instability.
Their tests, however, could not make a
Fig. 6. Schematic of Level Ice Sheet quantitative assessment on the effect on
Moving against Conical Structure. the propulsive characteristics of the
propeller. Quantitative propeller ice
Higher ice buoyancy in the rubble interaction experiments were carried out
increases both the breaking load and at IMD on a model of a twin propeller ice
rubble clearing forces resulting in breaking ferry [11]. In each case the
significantly higher loads evident in model was tested at the same speed, shaft
Figure 7. revolutions, ice thickness and strength
but the ice density was varied from 930
Downward Breaking Structure kg/m' to 880 kg/a0. Visually, the tests
400 didn't appear to show major changes in ice
flow. However, in the high buoyancy CD
ice there appeared to be more unstable
events where little or no ice would enter
300 the propeller disk. At higher speeds both
flows appeared virtually identical. In
Figure 8 ice torque defined as the
increase in mean torque due to ice milling
200 over mean open water torque, equation (4),
is plotted with model speed.
N

Legend
4, -
100 AQ, - x 100% (4)
+ 870 kg/m3. Q.
0 920 kg/m 3 Ice Propeller Interaction
o 150
o 20 40 60 1.0 m 0450 kPa

Ice St rength kPa e

1 100
Fig. 7.Horizonatal Load on Conical Legend
Structure versus Ice Strength for Two Ice + 935 kg/m
Densities. a
0 880 kg/m
Propeller Ice Interaction 50

Level Ice Self-Propulsion. IMD

conducts level ice self-propulsion
experiments with a model in an overload
condition. That is, a model is attached O
to a tow post and towed at constant speed 0 2 4
through level ice with propellers turning 6 8 10
at constant revolutions. The ice is Ship Speed (knots)
broken and submerged by the bow. As the Fig. 8. Percentage Increase in Propeller
model passes over the ice some of it
Torque due to Milling versus Ship Speed.

313
From this graph two things are apparent, in the higher density ice. However, after
first that the minimum model speed (speed the propellers became totally immersed in
intercept) at which no ice enters the rubble the increase in torque while in the
propeller appears to be slightly higher in CD rubble was only about 50% that of the
the CD ice and, secondly, the rate of high density ice. Hence, in this case,
increase in ice torque with speed appears the increase in resistance in CD ice was
lower with CD ice. It must be stressed partially offset by a lower level of
that these are preliminary results based propeller ice interaction.
on a small amount of data available at
this time, but the effect on minimum model Pack Ice Stability
speed seems reasonable. When the flow
becomes unstable ice pieces are free to To date no tests have been performed
rise to the surface under their own to access the effect of ice density on
buoyancy. More buoyant CD ice pieces pack ice stability and load transmission.
accelerate faster and clear before the However, work in flume tanks using
passing of the propeller. The change in synthetic ice [12] has demonstrated that
slope could have several causes. First, the critical water speed which induces ice
at higher speeds ice is more or less being piece instability increases with ice
continuously milled, but in case of CD ice buoyancy.
the ice trajectory appears slightly more
outboard resulting in less ice being SUMM.ARY
sucked inboard and trapped by the bossing.
The second effect may be due to ice In summary, the development of CD
properties, in particular, fracture model ice has allowed, for the first time,
toughness and crushing strength. As ice density to be accurately scaled in a
previously both model refrigerated model basin. Preliminary
mentioned research conducted at the Institute for
experiments were conducted at the same ice
flexural strength. For similar flexural Marine Dynamics indicated the following.
strengths, CD ice, due to its layering
technique, appears to fracture somewhat Ice buoyancy represents a significant
easier and has a slightly lower crushing component of the total resistance for a
It is doubtful whether these ship or structure in level ice. At low
strength.
small changes in ice properties could ship speeds this component represents
nearly all of the ice clearing
account for the significant change in
level of propeller ice interaction seen. resistance and is proportional to the
It can be seen from Figure 8 that at low density difference between ice and
model speed the effect of density is most surrounding water. In the past, this
component has, to some degree been
significant. For example, at 3 knots the under-predicted by model tests using
torque increase in CD ice is about 10%
while in high density ice it's about 30%. incorrect ice buoyancy scaling.
This may explain why there was a factor of Ice buoyancy has a direct influence on
3 difference between ice torques measured the probability of broken ice pieces
on the ship MV Kalvik and those predicted becoming jammed in the ship shoulder
region. Its magnitude can not be
by model tests [8]. The effect of ice on density
propeller thrust is much less, about one- ascertained without correct
sixth the torque influence. Also, while scaling and its influence probably
ice milling increase torque, it reduces depends on hull form. In general, hull
forms with a high likelihood of ice
delivered thrust. jamming are not penalized as they should
Rubble Self- be when tested in higher density ice.
Un-consolidated Unconsolidated ice rubble resistance is
Propulsion. Overload experiments with the
nearly all ice buoyancy dependent.
same model in unconsolidated rubble show
a combination of effects cited above. The Thus, in the past, model tests severely
rubble field was equal to the draft of the under-predicted this component of ridge
model. Ice density was 880 kg/m3 and 935 or rubble resistance.
for the CD and high density ice Ice-structure loading is dependent on
kg/m3 the buoyancy of rubble formed ahead of
respectively. The model was towed at
the structure. This is particularly
constant speed corresponding to 1 knot. for downward breaking
The propellers were turning at a rate so important
structures where buoyant forces act to
that maximum power was being delivered in resulting in
After penetrating the confine incoming ice,
the open water. higher loading.
rubble field fully, but before ice was Finally, the amount of ice entering a
ingested by the propellers, resistance in ship propeller is significantly reduced
the CD ice was about 60% higher than that

314
 if the ice density is correct. The SPENCER, D.S., et al, "Model Tests
effect appears to be two-fold. First, in Level Ice and Multi-Year Ridges
the minimum speed before the onset of of Model 440", IMD/NRC TR-1990-14,
ice milling is increased when ice October 1990, protected.
density is correct. Secondly, the rate
at which ice milling increases with SPENCER, D., MOLYNEUX, D.,
speed appears reduced when the density KEINONEN, A., "Model Tests of
is correct. Thus, in the past ice model M.V.'s "Terry Fox"/"Kalvik" and
tests generally have over-predicted ice- their Correlation to Full-Scale",
propeller interaction. IMD report TR-AVR-12, August 1988.
It is clear from these preliminary HARDIMAN, K.C., RITCH, R., and
tests that ice density scaling can no ABDELNOUR, R., "The Effect of Ice
longer be ignored in modelling. CD model Density on the Trajectory of Ice
ice represents a significant advance Pieces along a Vessel's Hull",
modelling and will allow for greater NRC/IMD CR-1989-12, 1989.
understanding of the ice breaking
phenomena. JONES, S.J., HARDIMAN, K.C., RITCH,
R., AND ABDELNOUR, R., 1990, "The
ACKNOWLEDGEMENTS Effect of Density on the Trajectory
of Ice Pieces around a Ship's
The author would like to express Hull", IAHR Ice Symposium 90, Esppo
sincere gratitude for technical support, Finland, August 1990.
patience and keen interest of Brian Hill,
Spence Butt and Craig Kirby during the SPENCER, D.S., SIMOES RE, A.,
development of CD ice. Also, many thanks "Model Resistance and Propulsion
to Mr. V. Santos-Pedro of the Canadian Tests of a 1/30 Scale Ferry
Coast Guard who provided partial funding. 'Straitway II' in the Ice Tank",
NRC/IMD TR-1991-17, December 1991.
REFERENCES
TATINCLAUX, J.C., et al., "A
SPENCER, D.S., TIMCO, G.W., "CD Laboratory Investigation of
Model Ice - A Technique to Produce Mechanics and Hydraulics of River
Controllable Density Model Ice", Ice Jams, CRREL Report 77-9, p. 15-
IAHR Ice Symposium 90, Esppo 17.
Finland, August 1990.
TIMCO, G.W., "EG/AD/S: A New Type
of Model Ice For Refrigerated
Towing Tanks", Cold Regions Science
and Technology, 12, 1986, pp 175-
195.

COLBOURNE, D.B., "A Three Component

Method of Analyzing Icebreaking
Resistance", PhD Thesis, NRC/IMD
IR-1989-07, October 1987.

SPENCER, D.S., JONES, S., and

COLBOURNE, B., "A Proposed Standard
Method for Conduct and Analysis of
Ice Resistance Model Tests",
NRC/IMD LM-1992-02, February 1992.
5 SPENCER, D.S., "Model-Scale Full-
Scale Ice Resistance Correlation
for the CCGS Louis St. Laurent",
NRC/IMD LM-1992-XX, May 1992.
6. "RIDGERAM - A Numerical Model for
Ship Performance and Transit
Analysis in Arctic and Sub-Arctic
Waters", Fleet Technology, (in
press).

315
 The Deep Water Multi-Directional Wave Basin at the
Offshore Technology Research Center
R.P. Johnson (Offshore Technology Research Center, USA)

BSTRACT centers. It is a joint venture

between Texas A & M University and the
The Offshore Technology Research University of Texas. The purpose of
Center (OTRC) deepwater, multi- the OTRC is to educate graduate
directional wave basin, located in engineers in preparation for work in
College Station, Texas, was commissioned the offshore industry and to conduct
in March of 1991. This model testing engineering research into the unique
facility supports OTRC engineering problems associated with operating in
science and research investigations into water depths exceeding 1000 m.
the unique challenges associated with Academic research projects are
deepwater operations. Scale model conducted by professors and graduate
studies of wave kinematics and wave- students located at both universities,
structure interactions in deepwater have as well as by selected researchers at
been conducted. other institutions throughout North
America.
The basin is 45.7 m x 30.5 m x 5.8 m.
A 9.1 m x 4.6 m pit has an adjustable The wave basin is located in the
floor to allow depths from 5.8 m to 16.7 Texas A & M Research Park in College
m. Forty-eight individually controlled, Station, Texas. It is designed to
hydraulically actuated, hinged accommodate scale model tests of free-
waveboards, located on one end of the floating, moored, fixed and compliant
basin, are used to generate regular, structures subjected to waves,
random, long-crested and short-crested currents and wind.
seas. A progressive wave absorber
consisting of variable porosity vertical 2.0 Description of Facilities
screens is located opposite the
wavemaker to minimize wave reflections. 2.1 Wave Basin. The centerpiece
of the multi-directional wave model
Wave generation and data acquisition testing facility is the basin, 45.7 m
is accomplished using an integrated x 30.5 m, in plan. The primary water
software package operating on a single depth is 5.8 m. There is a 16.7 m
workstation computer. deep well, 9.1 m x 4.6 m, in plan,
located along the centerline of the
Capabilities to simulate surface and basin in the area of homogeneous
sub-surface currents is currently being directional waves. The well has a
installed. Side wave absorbers and wind vertically adjustable floor so that
simulation capabilities are planned for depths from 5.8 m to 16.7 m can be
set.
installation next year. Video
techniques will be developed to assist
Tie-down points for model moorings
in measuring and characterizing short- and for securing equipment have been
crested seas. cast into the floor on a 3.0 m x 3.0 m
square grid.
1.0 INTRODUCTION
Observation wells with glass
The Offshore Technology Research windows are set along the side walls
Center is one of eighteen National at two different depths. There is an
Science Foundation engineering research additional observation well under the

317
shelf that supports the wavemaker. power.

The control room and an observation The wavemaker was manufactured by

balcony are on the second floor W. R. Davis Engineering, Ltd under a
overlooking the basin. Experiments are license from the Hydraulics Laboratory
supervised from a computer console of the National Research Council of
located in the control room. Both the Canada.
wavemaker controls and the data
acquisition system are operated from a Figure 2 presents the regular wave
single computer workstation employing an height performance envelope for the
integrated software package. entire range of wave periods that can
be made. The maximum regular wave
A loading dock and a staging area height achievable is approximately 0.8
dedicated to model construction and m over a range of periods from 2.3 to
model preparation is located at the wave 2.9 seconds.
absorber end of the basin. An overhead,
travelling crane, with a capacity of 5 To obtain the desired types of
tons, has access to the staging area as waves, the wavemaker control signals
well as to the entire surface area of are created on the primary laboratory
the model basin. computer. Software exists to make a
variety of standard sea states.
Other basin support equipment
includes a personnel/instrument carriaae 2.3 Wave Absorber. The wave
which spans the tank. This carriage can absorber is located along the 30.5 m
be positioned anywhere along the length wall of the basin opposite the
of the model basin. Instruments, models wavemaker. It consists of 19 parallel
and test fixtures may be supported from rows of vertical expanded metal
this platform. panels. As the waves travel from the
basin test area into the wave
Mechanical and electronic shop absorber, they encounter panels with
facilities are available for model and progressively decreasing porosity and
instrument fabrication, modification, spacing. This space-efficient wave
repair and maintenance. The Texas A & M absorber design occupies approximately
machine shops or commercial modelmakers 10.3 m of the model basin.
are used for complex model construction.
Experience with the wave absorber
A layout of the laboratory facility over the past year indicates that
is shown in Figure 1. performance can be improved by adding
more rows of screens. Current plans
2.2 Wave Generatina system. A 48 are to complete upgrading to the wave
segment, him:red flap, multi-directional absorber by the end of 1992.
wave generator is located on the 30.5 m
side of the model basin. The waveboards Side wave absorbers of the sane
are located on a shelf approximately 3.3 type are planned to minimize
m above the floor of the tank. The transverse waves.
waveboards are approximately 3.0 m high
and 0.6 m wide. 2.4 Data Acauisition and Analysis.
The primary laboratory computer is a
The waveboards are individually VAXstation 35007m from Digital
driven by servo-controlled linear Equipment Corporation. This computer
hydraulic actuators. Each waveboard is is used to both generate wavemaker
controlled by an independent, computer- command signals and to acquire data
generated, command signal. The servo- during experiments.
control system uses algorithms
implemented on three single-board For wave generation, the waveboard
computers to control the waveboard displacement command signals are
motions. computed on the Vaxstation. These
signals are then downloaded via a
The wavemaker consumes 447 KW of serial line to three single-board

318
 computers which each control sixteen 3.3 Multi-Directional Wave
segments of the wavemaker. Real-time Characterization. Development of
control of the wavemaker is exercised by techniques for measuring and
the single-board computers. This leaves characterizing short-crested seas are
the VAXstation available for other real- planned. Stereo imaging of the water
time duties, such as data acquisition. surface will be obtained with a pair
of color TV cameras and recorders. The
A NEFF System 620'm acquires the
pairs of images will be used in the
analog instrument signals and converts traditional stereo separation mode to
them to digital format for subsequent determine water surface elevations.
storage and analysis. Currently a
Discrimination of small surface-
maximum of 64 channels of data may be
marking floating targets is simplified
acquired simultaneously. This system is by using a number of different colors
expandable to 128 channels.
for the targets. The output from one
of the cameras will be used to make 30
Data reduction and immediate post- Hz samples of surface velocity by
test analysis is conducted on a
determining the difference in location
VAXstation 4000/Model 601.14. Data may be of targets between adjacent images.
transferred, via the OTRC local area The system may readily handle a
network, to other desktop computers and
hundred or more targets in the image
workstations for specialized
so that extremely high resolutions are
computations, analysis and archiving. possible for both surface slopes and
velocity vectors. Twin VCRs with good
The wave generation and data single framing stability will be
acquisition software package is a slaved to a PC. Target filtering and
licensed as GEDAPTm. This software was
developed at the Hydraulics Laboratory location identification will done
of the National Research Council of rapidly so that storage of more than
one full color image at a time (more
Canada for application to long-crested
than a megabyte) is not required so
and short-crested wave model testing and that the pertinent information may be
data analysis.
reduced by three or four orders of
3.0
magnitude.
Plans for Upgrading the Facility
3.1 Current Generatinc System A
local current generating system is
currently under construction. A maximum
flowrate of 6400 gpm is possible with
the existing pumps. The flowrate may be
controlled by adjusting the speed of the
variable speed pump or by pneumatically
actuated flow control valves on the
discharge lines.

The discharge manifold will be

moveable within the basin so that any
combination of wave-current directions
is possible. The manifold may also be
raised and lowered to generate localized
surface or sub-surface currents.

3.2 Wind Generating System. A local

wind generating system is planned for
installation next year. The system will
be an "open loop" consisting of a fan
baffles, flow straighteners and a short
section of tunnel. The fan will be
servo-controlled to produce specified
target low frequency wind spectra.

319
 150 F
............
*.i (45-7m). ----. Observation :::-:
Window -:-.¡:-_-.
Motorized -

Instrumentation .. 48 Hinge
-:' Platform Waveboards:-
..............
100 Ft.
(30.5 m)
... -,:::-.:,:i.K.:?:::.:::-:::::::.:.::,
:.

Loading Progressive
. 15 Ft.
"-- Deep
Wave
. 30 Ft.. .
Dock - - 4.6 m P it
Absorber ....
... - 30 Ft. .:.: - - -
Observation ...¡
- .:--
30 Ft. 9.1 m ::::

.
.: :
Window

Observation Second Floor

Window :.--. :.:. Off ices

16 Ft. Control Room & Observation First Floor

(Second Floor) Machinery

Shops Offices Lobby

Figure 1. 0. T. R. C. Model Basin Layout

e
0.9

0.8
u
1
0.7
a
m 0.6
,I11111111111
111= IMLNI
W
a
e

e
0.5

0.4
IIIVA
III/11
El=NMI
0.3

02
KEN
r
111111111
0.1

o
o 1 2 3 4 5
Regular Wave Period (Seconds)

Figure 2. Regular Wave Performance Envelope

320
 Radar Imaging of Surface Ship Wake Flows
K.R. Nicolas, C.S. Weller, W.T. Lindenmuth, B.H. Cheng
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT when needed, over the radar beam sensing

area. The radio controlled model, trailing
Promising results have been achieved in its wake, passes through the radar sensing
the development of radar diagnostic tech- area. The RIMS radar records the time
niques for the mapping of water surface evolution of this wake as the various wake
velocities in the laboratory environment of features develop and then damp out.
the Maneuvering and Sea Keeping Basin The RIMS instrumentation radar system is
(MASK) at DTMB. High resolution measure- a step-chirp radar with a frequency cov-
ments, with the Radar Image Modeling System erage of 8 to 18 GHz, 10 to 1000 watts peak
(RIMS ) instrumentation radar, mapped the power and pulse-to-pulse polarization
structure of the transverse surface veloci- agility, which is coupled to control, data
ties of the wake from a self-propelled ship acquisition and analysis computers. For
model . The wake center has a broad the data presented here, the RIMS radar
distribution of transverse velocities with transmits vertically polarized bursts of a
speeds of typically 10 cm/sec. The outer frequency stepped waveform from a base
wake exhibits higher transverse velocities, frequency of 8.4 GHz. The waveform consists
e.g. 25 cm/sec, which are interpreted as the of 64 frequencies, equally spaced over a
sum of the water velocity and the capillary 1 GHz bandwidth. Noise reduction is
wave phase speed. accomplished by averaging 32 returns at
each frequency, yielding an effective burst
RADAR DIAGNOSTIC TECHNIQUE rate of 150 Hz. Each frequency burst is then
Fourier transformed to obtain the radar
Radar diagnostic techniques for the backscattered intensity as a function of
mapping of the water surface velocities of downrange distance. This gives a one-
a ship model wake have three advantages dimensional "snapshot" with 64 adjacent
which can complement other methods: (1) downrange resolution cells. Each cell has
wide area of measurement, e.g. synoptic dimensions of 15 cm downrange by 1 m cross-
views, (2) fast time response, and (3) rapid range.
data processing.
The technique, as depicted in Fig. 1, is DERIVATION OF VELOCITIES
described as follows. A self-propelled
ship model maneuvers through the radar The motion of the scatterers in each
footprint (on the water's surface) in the range cell from burst to burst imparts a
240 x 360 ft Maneuvering and Sea Keeping phase shift to the backscattered intensi-
Basin (MASK) . Along one side of the MASK ties. This shift is computed by subtracting
are a wind turbine and the Radar Image the phase for successive bursts in each
Modeling System (RIMS) .The RIMS radar downrange cell. The resulting velocity,
antenna beam illuminates a region on the which is directly proportional to the phase
water's surface 10 m downrange (across the shift, is interpreted as the sum of the
model track) by 1 m cross-range (along the surface flow velocity plus the phase speed
model track) at a look-down angle of 15 deg. of capillary wave scatterers when present.
The wind turbine creates a uniform wind, By stacking these velocity profiles seri-

321
ally side-by-side, a time history and of the transom. The centerline wake flow
spatial image of the wake cross-track fields follow next. Flow is predominately
velocity is created. outward within one model length, with some
reversal near the center farther down-
WAKE SURFACE FLOWS stream. The velocity field is interpreted
as the sum of the surface flows and the phase
Fig. 2 displays a typical water surface speed of the capillary wave scatterers.
velocity field from a model ship wake. The The graph at the bottom of the figure
ship model is at the bottom and the shows the velocity profile across the wake
observing radar is to the left. Time and at a location 3 model lengths downstream.
the evolution of the wake progress upward. As indicated in the shaded image, the
Dark regions denote line-of-sight (perpen- velocities in the outer wake are directed
dicular to the model motion) velocities outward, typically at 25 cm/sec. A reversal
toward the radar. Lighter shades indicate to inward flow occurs near the center, with
velocities away from the radar. Beginning speeds averaging 10 cm/sec.
at the bottom of the figure, the bow of the This radar diagnostic technique for
model entes the radar sensing area with water surface velocities should be a useful
the bow wave showing motion away from the complement to other techniques and offers
model. The model appears next (saturated) new insights to the structure of wakes.
followed by the stern wake immediately aft Sponsored by the Office of Naval Tech-
nology (ONT-211).

WIND GENERATOR RIMS

ANTENNA

WIND
ROUGHENED
SURFACE

MANEUVERING
AND SEAKEEPING BASIN

Fig.

322
 10m

Fig. 2

323
 SESSION IV-A

NEVV FACILITIES, TECHNIQUES,

ICE TESTING, AND VALIDATION-III
 Repeatability of Resistance Experiments in Ice with a
1:20 Scale Model of the Canadian R-Class Icebreaker
S. Newbury (Institute for Marine Dynamics, Canada)

ABSTRACT Careful measurements were made of the ice sheet properties

so that the effect of these variations could be minimized.
A series of experiments was initiated at IMD's model ice Typical variations in ice sheet strength and thickness are
basin to determine the experimental error that may be reported.
expected in the results of ice reistn.r.f. tests. A set of nine
ice she,ets was dedicated to the study, in which a 1:20 scale ICE TANK FACILITY
model of the R-Class icebrealcer was tested in EG/AD/S
model ice. The test length for each model resistance test The model experiments were performed in the ice tank at
varied from one and one-half model lengths to two arad one- the Institute for Marine Dynamics. The tank is 96 m long,
half model lengths for the higher velocities, as well as 12 m wide, and 3 m deep. The main towing carriage
several tests through the full length of the ice sheet houses the test instrumentation system, which provides
(approximately 70 m). The test results were corrected to regulated excitation voltage and transducer pickup for e,ach
nominal ice conditions according to standard IMD instrument channel. The transducer outputs are digitized by
procedures, and these results were then compared to a NEFF System 620 Series 100 amplifier/ multiplexer, and
determine the resistance variation. the data is stored on a DEC Microvax II.

INTRODUCTION The model ice used in the IMD basin is the EG/AD/S type
[4]. The ice is grown from an aqueous solution containing
The Canadian R-Class icebreaker has been extensively tested chemical dopants through careful seeding, freezing, and
at many model ice basins around the world, as it has been tempering procedures. The percentage concentrations of
the subject of an ITTC study of comparative model ice dopants during this test series were 0.39/0.036/0.04. The
resistance [1]. The full scale ship has also been tested in density of the model ice was 940 kg/m3, as determined by
the past, and more recently by IMD [2]. Thus a model of measuring the force required to submerge a piece of ice.
the R-Class vessel was chosen to investigate the repeatability Several other characterization tests for each ice sheet were
of the model ice resistance experiments at IMD. routinely performed. These included cantilever be,am tests
to determine the flexural strength, a plate deflection test to
The aim of ship model ice experiments is to accurately determine elastic modulus, a crushing test to determine
predict the performance of the full scale icebreaker. One uniaxial unconfined compressive strength, and a punch test
aspe,ct of this prediction which must be considered is the to determine shear strength.
precision of the model test results, or the amount of
variation which may occur in the test results. These The ice sheet parameters which may have the most
variations may occur for several reasons, such as small significant influence on the resistance results of a model
changes in the thickness and strength of the model ice, icebreaker are ice thickness and flexural strength, so close
which may not be measurable by standard measuring attention was paid to these. Ice thiclmess was measured at
techniques during a model test. Other small variations may 2 m intervals along both sides of the broken channel after
occur naturally due to the nature of the icebreakirig process, the model test. The flexural strength of the ice at the
as it is not continuously periodic, but is a more chaotic location of the test section to be used for the data analysis
process [3]. was also measured. Five cantilever beams were cut in the
ice sheet and failed with the application of a measured load
This study quantifies the variations in the experimental on each side of the broken channel immediately after each
results from a series of resistance tests for this vessel. test.

327
TEST PROGRAN1 Table 2 lists all of the level ice test runs that were
performed. Several identical test runs were performed in a
The 1:20 scale R-Class model hull was constructed several second nominally identical ice sheet, and some were even
years ago of glass reinforced fiber. The hull was refinished carried out in a third and fourth sheet. The multiple test
with an Imron based paint preparation prior to this test runs in the single ice sheet are from the long test runs which
series. This paint was prepared according to IMD were later divided into separate runs. The numbers in the
procedures [5] to yield a specific hull roughness in order to brackets refer to the ice sheet number in this set of
generate a specific ice/hull friction coefficient during the "RCLASS" ice sheets.
model tests. The nominal ice/hull friction coefficient was
determined to be 0.028, as measured on a flat plate which Presawn ice resistance tests were also performed for each
was prepared at the same time and under the same speed in each ice thickness. Sawing of the ice sheet into the
conditions as the model hull surface. The test conditions of approximate breaking pattern eliminates the breaking
the model are presented in Table 1. component from the total ice resistance of the model, and
facilitates the analysis of the test data.
The model was attached to the main towing carriage through
a rigid tow post which extended down to the model's center RESULTS
of gravity. A gimbal in the model allowed freedom of
motion in pitch and roll, while movement of the tow post in A data segment was selected for each test run (more than
linear bearings allowed motion in he.ave and yaw. Since one segment was selected for the long test runs at constant
yaw motion was not desired, a yaw restraint was attached speed). The segment selected was taken after the model had
near the model's bow at the vertical center of gravity. A fully entered the ice and the motions had stabilized from the
300 lb force transducer was used to measure the resistance initial ice contact. Because of small variations in ice
of the model. The signals from the transducers me,asuring thickness and flexural strength in each ice sheet and between
model resistance, roll and pitch angles, heave displacement, different, but nominally identical ice sheets, the model
carriage spe,ed, and longitudinal model position were resistance data was corrected to the nominal ice strength and
sampled through the data acquisition system. thicicness. A direct comparison between the ice resistance
data for different test runs could then be made. A typical
Test results from nine ice sheets make up the database for time history of the acquired data during a test run is shown
this comparison of re,sistance results. Of these nine ice in Figure 1. This test includes three speeds in 35 mm level
sheets five had nominally identical properties of 35 mm ice as well as three speeds in presawn ice. Examples of
thiclmess and 40 1cPa flexural strength, two ice sheets were data segments selected for analysis are shown by the cross-
grown to 50 mm thicicness with a target flexural strength of hatched sections.
40 kPa, and the other two were 22.5 mm ice she,ets which
were tested at both 40 and 20 kPa. A range of model
velocities, from 0.15 to 0.90 m/s, was also investigated.

The test length for the single spe,ed runs varied from one 67 ^ao .C,PC,
and one-half model lengths for the lowest spe,ed to two and 60 00
+0.00
one-half model lengths for the highest speed. This allowed
the model to fully enter the ice and the model motions to 70'0 000
stabiliz,e prior to obtaining the data segment which was later 0 902
0 SOC
selected for analysis. Several single speed run were made 0 ec,

through the entire length of the ice sheet in order to o =cc

o co, .000
compare resistance results in the same ice sheet. These long co 000
-12 0.1, (004
test runs were later divided into several runs of shorter ooc

length.
000 20.000
The model was also run through ice at the quarter point of o 6+, 672.6 (000 070,
0 000
the tank after the centerline tests to compare resistance
results in an effectively infinite ice sheet to results in ice _00c
with a nearby broken channel. During tests at the quarter 0 6220.000 20.000 40.000
(sec .1070)
60.000 00.000
7007IOCLA5550 or.-2 6ro0-10,
point the center channel was pinned with wooden "staples"
in order to prevent the quarter point channel from opening
up as the model proceeded through the ice. Fig. 1. Time History of Test Data

328
 Typical variations in ice thickness and ice strength are charuael at the center, in Table 5. Results from multiple
small; variations between different but nominally identical runs in the same ice sheet were averaged before the
ice sheets may be somewhat larger. Figure 2 shows a plot comparisons in Tables 4 and 5.
of the typical variations- in ice thickness and ice strength for
one of the 35 mm ice sheets (No. 49). DISCUSSION AND CONCLUSIONS

A total of 67 level ice resistance tests were carried out in

nine ice sheets. The results have be,en tabulated and
411 0
FLEXURAL STRENGTH AND THICK-NESS VARIATION WITH LOCATION
441 0
comparisons have been drawn between the different tests.
The thickness and flexural strength measurements were very
40 0 44.0 constant throughout each of the 70 m ice sheets. The
42 0
standard deviations of the average thickness measurements
O. 420
at 34 locations along the model's channel were about 0.5,
40.0 o 40 0 0.75, and 1 0 mm for 22.5, 35, and 50 rom thick ice sheets,
STAL7,CTN
(0,)3. o IMICIC/a333
o 1.uo,
respectively. The standard deviations of the average
flexural strength measurements at seven locations along the
311
300 length of the ice sheet were about 1.5, 1.8, and 3.8 kPa for
340
the 22.5, 35, and 50 mm ice sheets, respectively.
34.0

320 o Wit111C 1333 =A, The standard deviation of the resistance results was used as
300
an indicator of the variation in the test rsults. The
00 IO 420 16 0 24.0 300 NI 0 CO. 440 640 600 percentage deviation in this value from the mean resistance
COl ',141C1.13.1/011
was calculated to obtain a percentage variation in the results.

Fig. 2. Variations in Ice Thickness and Strength The percentage of the standard deviation of the resistance
results were mostly small, in the range of zero to five
percent, with a few exceptions. For the single speed test
The model's level ice resistance was corrected to nominal runs in the same ice sheet at the tank center the variations
ice conditions by standard IMD procedure [6]. The in resistance were shown to be about four percent. For the
bre,aking component of the measured level ice resistance is test runs at the quarter point in the same ice sheet the
calculated by subtracting the presawn resistance for that ice variations in resistance were about three percent. The
thickness from the total measured resistance. This breaking comparison of resistance results in different but nominally
component is then corrected to the target ice strength by identical ice sheets showed variations in the range of about
linear interpolation between results at the measured ice two to four percent, except for two cases which had
strength and at zero ice strength. Correction of the breaking variations of 8.7 and 11.8 percent. The comparison of
resistance for ice thickness is then achieved by plotting the resistance results from tests carried out at the center and
breaking component of the level ice resistance, corrected for quarterpoint showed variations which were typically within
ice strength, against measured ice thickness. The breaking the range for the results from different but nominally
resistance, corrected for strength and thickness, is then identical ice sheets. There was no clear trend of the
interpolated from this curve. The total level ice resistance influence of the open center channel on the results from the
for the target ice strength and thickness is finally calculated tests at the quarterpoint.
by adding the presawn ice resistance, corrected to the target
ice thickness, to the corrected breaking resistance. REFERENCES

The final results and comparisons are presented in Tables 3 "Report of the Performance in Ice-Covered Waters
to 5. The ice sheet number is given, the flexural strength Committee" (1984). Proceedings, 17th
(a) and ice thickness (II) at the time of the test, the International Towing Tank Conference, Goteborg,
measured resistance (R,), and the ice resistance corrected Sweden.
to the nominal ice conditions (R,c). The mean and standard
deviation of the corrected results are then presented. In Williams, F.M., D. Spencer, S.T.Mathews and I.
Table 3 only test runs in the same ice sheet are compared. Bayly, Full Scale Trials in Level Ice with the
The same speed tests in different, but nominally identical ice Canadian R-Class Icebreaker, SNAME 1992
sheets are compared in Table 4. Center point tests are Annual Meeting, New York, November, 1992.
compared to tests at the quarter point with an open, pinned

329
3. Ettema, R., M.B. Sharifi, K.P. Georgakakos and TABLE 1
F. Stem, Chaos in Continuous-Mode Icebreaking,
Cold Regions Science and Technology, 19 (1991) MODEL TEST CONDITIONS
131-144.
Forward Draft (m) 0.335
Timco, G.W. EG/AD/S: A New Type of Model Aft Draft (m) 0.358
Ice for Refrigerated Towing Tanks, Cold Regions Displacement (kg of f.w.) 957.000
Science and Teclmology, 12 (1986) 175-195. KB (m) 0.194
BM (m) 0.244
Newbury, S. and J. Bell, Effect of Friction on KG (m) 0.388
Icebreaker Model Tests at IMD, POLARTECH 92 LCG (m aft 0) 0.023
Conference, Montreal, January, 1992. Roll Radius of Gyration (m) 0.325
Pitch Radius of Gyration (m) 1.224
Newbury, S. Modelling Technique,s for ice-
breaking Hull Forms, NRC/IMD Report Number
LM-1991-10, October, 1990.

Table 2

endition Number of.Buns Performed

One Sheet 2nd. Sheet 3rd. Sheet

1/4 Pt. center 1/4 Pt. 1/4 Pt. 1/4 Pt.

center

.15 40 22.5 1(40) 1(44)

.15 20 22.5 1(40) 1(44)

35.0 1(41) 1(42) 1(50)

.15 40

.15 40 50.0 1(43) 1(43)

.3 40 22.5 1(40) 1(44)

.3 20 22.5 1(40) 1(44)

7(48) 1(41) 1(42) 1(50)

.3 40 35.0 7(48)

40 50.0 7(47) 3(47) 1(43) 1(43)

.3

.6 40 22.5 1(40) 1(44)

.6 20 22.5 1(40) 1(44)

1(41) 1(42) 1(50)

.6 40 35.0 7(49) 3(49)

.6 40 50.0 1(43) 1(43)

40 35.0 1(42) 1(50)

.45
40 35.0 1(42) 1(50)
.75
35.0 1(42) 1(50)
.90 40

330
 TABLE 3

COMPARISON OF RESISTANCE RESULTS IN SAME ICE SHEET

TEST or It R. Rt. MEAN STANDARD

NO. DEVIATION
(kPa) (mm) (N) (N) (N) (N)

CENTERPOINT RUNS
35m 40kPa 0.3m/s
48 44 34.6 44.5 39.5 42.2 2.0
48 41 34.4 45.8 43.3
48 44 33.7 43.0 39.9
48 42 35.0 47.8 43.1
48 42 34.6 48.1 44.3
48 42 34.5 47.7 44.1
48 44 34.6 46.4 41.3

35m 40kPa 0.6m/s

49 42 36.3 70.3 64.3 63.1 2.2
49 44 36.2 68.7 61.4
49 40 35.6 66.8 64.9
49 42 35.2 64.4 61.9
49 41 35.2 68.1 66.4
49 44 35.3 67.8 63.1
49 42 34.8 61.2 60.0

50m 40kPa 0.3m/s

47 42 52.7 103.5 91.9 88.8 4.0
47 49 52.3 105.3 83.8
47 47 50.7 97.1 84.2
47 40 52.0 99.8 93.3
47 42 51.6 100.2 91.5
47 42 51.5 93.9 86.2
47 38 51.6 92.4 91.0

QUARTERPOINT RUNS
35mm 40kPa 0.3m/s
48 38 34.0 40.0 40.2 42.1 1.1
48 38 33.4 39.6 41.1
48 38 33.5 40.5 42.0
48 37 34.7 43.0 42.8
48 38 34.3 42.8 42.7
48 39 33.9 42.9 42.9
48 36 34.2 41.7 43.3

35m 40kPa 0.6m/s

49 41 35.0 58.6 57.8 60.3 2.1
49 41 35.4 63.5 61.5
49 39 35.1 60.9 61.5

50m 40kPa 0.3m/s

47 32 51.8 88.1 98.0 95.4 2.3
47 39 52.3 99.3 93.7
47 40 50.7 96.7 94.5

331
 TABLE 4

COMPARISON OF RESISTANCE RESULTS IN NOMINALLY IDENTICAL ICE SHEETS

TEST ar H, Rt. Rtc MEAN STANDARD

NO. DEVIATION
(kPa) (mm) (N) (N) (N) (N)

22.5mm 201dPa
0.15m/s

40 27 20.6 13.2 12.9 13.7 1.2

44 22 21.4 14.3 14.6

0.3m/s

40 26 21.6 17.8 16.8 17.2 0.6

44 23 21.9 18.2 17.6

.6m/s

40 24 21.7 30.7 30.0 30.6 0.8

44 21 21.4 29.6 31.1

22.5mm 40kpa
0.15m/s

40 40 21.7 16.7 17.7 17.8 0.1

44 37 21.7 16.0 17.8

0.3m/s

40 46 22.5 21.8 20.2 22.1 2.6

44 38 22.2 22.7 23.9

0.6m/s

40 46 21.8 35.5 34.7 35.8 1.6

44 43 21.8 36.5 36.9

35mm 40kPa
0.15m/s

41 38 35.0 37.0 38.3 38.5 0.8

42 40 35.7 39.1 37.8
50 42 36.1 42.9 39.3

0.3m/s

38 34.6 40.5 42.6 43.7 1.5

41
42 44 36.4 51.7 45.3
50 42 36.1 50.8 44.5
48 43 34.5 46.2 42.2

0.6m/s

38 34.5 61.3 64.7 62.9 1.8

41
42 41 35.2 65.1 63.5
50 44 35.2 64.4 60.5
49 42 35.5 66.8 63.1

50mm 40kPa
0.3m/s

43 51.8 98.9 88.8 90.4 2.2

47
43 37 51.7 91.9 91.9

332
 TABLE 5

COMPARISON OF CENTER AND QUARTERPONT RESISTANCE RESULTS

TEST LOCATION (7, HI Rtc MEAN STANDARD

NO. DEVIATION
(kPa) (mm) (N) (N) (N)

35mm 40kPa
o.3m/s

48 center 43 34.5 42.2 42.2 0.1

48 quarter 38 34.0 42.1

0.45m/s

42 center 4G 34.6 54.2 55.1 1.3

50 quarter 37 35.3 56.0

0.6m/s

49 center 42 35.5 63.1 61.7 2.0

49 quarter 40 35.2 60.3

0.75m/s

42 center 41 34.4 72.5 72.2 0.4

50 quarter 37 35.6 71.9

0.9m/s

42 center 40 33.3 91.8 85.7 8.7

50 quarter 41 35.2 79.5

50mm 40kPa
0.15m/s

43 center 41 51.5 69.6 68.1 2.1

43 quarter 39 50.6 66.6

0.3m/s

43 center 37 51.7 91.9 91.5 0.6

43 quarter 35 50.9 91.0

47 center 43 51.8 88.8 92.1 4.7

47 quarter 37 51.6 95.4

0.6m/s

43 center 37 51.0 113.8 116.2 3.3

43 quarter 34 51.1 118.5

333
 2D Tests with One and Two Cylinders in Waves
M. de Conti (University of Sao Paulo, Brazil)

ABSTRACT analyzed in a micro-computer or in some

specialized equipment, for instance a
A small new facility has been spectrum analyzer.
installed in the Department
Naval of
Architecture of the University of Sdo
Paulo, Brazil. It is a wave flume, 14 THE MODELS
meters long, 1 meter wide and 1 meter
deep. Regular and irregular waves, as well Firstly one single cylinder was
as packages of waves, can be generated in tested. Its diameter is 0.15 m and the
the range 0.5 - 2.5 Hz, 0.5 - 5.0 cm. The draft was 0.225 m.
wave generator is plunger-type, driven by Then, two cylinders were tested,
a servo-motor. varying the distance between them, and
2D tests with one and two their draft. The diameter of both
cylinders have been conducted, varying the cylinders is 0.15 m. The distance between
immersion and the distance between the the cylinders' axes is here called d; the
cylinders. Forces acting on the cylinders, draft is t. The following conditions were
wave elevation and plunger motion have tested (see figure 1):
been measured. d=0.44 m; h=0.225 m;
The paper describes the facility d=0.44 m; h=0.260 m;
and presents the results for the d=0.36 m; h=0.225 m;
stationary cylinders acted transversely by d=0.36 m; h=0.260 m;
waves. d=0.28 m; h=0.225 m;
d=0.28 m; h=0.260 m.
THE FACILITY The cylinders were rigidly
connected. Two vertical columns with a
A wave generation equipment was very thin foil shape were attached to each
installed in a small tank 14 meters long, cylinder, and connected by a plate above
1 meter wide and 1 meter deep, already the free surface. The influence of the
existent in the Department ofNaval columns was kept small, because of their
Architecture of the University of Sdo thin shape in the direction of the flow.
Paulo.
The wavemaker is a plunger driven DESCRIPTION OF THE TESTS
by a servo-motor; the plunger motion
measured by a potentiometer is used as The tests with the single
feedback. Regular and irregular waves in cylinder were conducted with a set of
the range 0.5 to 2.5 Hz can be generated regular waves. Then tests with two rigidly
with amplitudes from 0.5 to 5 cm. It is connected cylinders were run with regular
also possible to generate packages of waves as well as package of waves.
waves with an approximately flat spectral The models were fixed at a
distribution of amplitudes. distance of 5.5 meters from the wave
The beach is a wooden generator, that is, approximately the
construction on parabolic arc, completed middle way between the wave generator and
with transverse beams. Its length is 3.5 the beach (see figure 2). The cylinders
meters, so that the effective length of were positioned horizontally; their length
the tank is about 10.5 meters. is 0.98 meters, thus something less then
The facility is intended for the tank width of 1 meter.
tests of fixed and free models in waves. The vertical force was measured
Floating models can be installed on a XY by a strain-gage transducer. The
type mechanism; vertical, longitudinal and transducer was fixed between the plate
angular motions can be measured by that connected the columns of the
potentiometer. Forces on fixed models can cylinders and a beam that laid over the
be measured by strain-gage transducers. side walls of the tank.
The signals are amplified, The wave elevation was measured
filtered, etc, and then A/D converted and by a resistance wave probe, located at a

335
distance of 2 meters from the wave
generator (see figure 2).
The plunger motion was measured
by a resistance potentiometer, and this
signal was used as a feedback control to
the servo-motor.
The packages of waves had an
approximately flat spectral distribution
of amplitude. The sweep for the plunger
motion was from 0.1 to 3.0 Hz. The time of
generation for each sweep was 25.0
seconds; the plunger was kept stationary
for 5.0 seconds from sweep to sweep. A 2-
channel spectrum analyzer was used for the
data acquisition. A rms average of 64
acquisition periods of 25 seconds was made
for each test. The wave elevation and the Figure 1 - The model (double cylinder).
vertical force on the model were analyzed;
a Hanning passband was used. The transfer
functions were recorded.
Each condition was tested four
times. The averages of these results was
obtained. $ 5*5m
A set of regular waves tests was wove
conducted for the single cylinder and for .0m absorber 20m
the two cylinders - condition 1.

RESULTS

The results for the single

cylinder showed very good agreement with wave
theoretical predictions, and the curves generator
almost coincided. 14.0m
The height of the waves was
varied for some regular wave tests, and no
non-linearity was observed.
Packages of waves were generated
without the model. The resulting amplitude Figure 2 - The tank.
spectrum was very similar to those
obtained when the model was present. It
concluded that reflection was not 2
was force/ wave amplitude ( Kg/cm
intense for the tested conditions. ..,%
The results for regular waves and
packages of waves (condition 1) were ----
\
1

compared; the corresponding curves 3

011, 5
resulted very similar.
The results for packages of waves
are plotted in figures 3 to 7. In figures ,.
3 and 4 the draft is and the
fixed,
distance between the cylinders is varied.
\\
.

//
/ ...
In figures 5, 6 and 7 the distance between
the cylinders is fixed and the draft is /
varied. \\-, / .....
......,
ACKNOWLEDGEMENTS 7 10 15 2.0
The author acknowledges wave frequency(Hz)
Technicians Joao Martins, Carlos Souza and
Marlei Araújo, who conducted the tests. Figure 3 - Vertical force per wave height
versus exciting frequency; double
cylinder; draft 0.225 meters; distance
between the cylinders 0.440 meters
(condition 1), 0.360 meters (condition 3),
0.280 meters (condition 5).

336
 2
force / wove amplitude ( Kg/cm ) force/ wave dmphtude ( Kg/cm )

\
2
\ --_-__ 3
4

.... .... ....

s,
d

o
/' N..
%

0Q7
07 I0 I5 20 10 1.5 2.0
wave frequency (Hz) wove amplitude ( Hz)

Figure 4 - Vertical force per wave height Figure 6 - Vertical force per wave height
versus exciting frequency; double versus exciting frequency;
cylinder; draft 0.260 meters; distance
double
cylinder; distance between cylinders 0.360
between the cylinders 0.440 meters meters; draft 0.225 meters (condition 3),
(condition 2), 0.360 meters (condition 4), 0.260 meters (condition 4).
and 0.280 meters (condition 6).

force/Wove amplitude' ( Kg /cm )

........,,
%

\ /
--- . I

do
2

0,.....----
\\\
....
2

- \III
force/ave amplitude ( Kg/cm)

0'
0.7 1.0 1.5 2.0 .7 10 1.5 2.0
wave frequency ( Hz) wove frequency (Hz)
Figure 5 - Vertical force per wave height Figure 7 - Vertical force per wave height
versus exciting frequency; double versus exciting frequency; double
cylinder; distance between cylinders 0.440 cylinder; distance between cylinders 0.280
meters; draft 0.225 meters (condition 1), meters; draft 0.225 meters (condition 5),
0.260 meters (condition 2). 0.260 meters (condition 6).

337
 A Precision, High Capacity, High Speed,
Right Angle Drive Fully Submerged
Propulsor Dynamometry System (FSPD)
K.J. Anderson, G.A. Hampton
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT determining the ship propulsion

conditions used in water tunnel
This paper explains the propeller cavitation evaluations.
capabilities of a new right angle drive
propeller dynamometry system which has BACKGROUND
recently been put into service at the
Carderock Division of the Naval Surface In the past, the model propellers
Warfare Center (CDNSWC). The towing usually used at CDNSWC were less than 12
carriage mounted Fully Submerged inches in diameter. The CDNSWC
Propulsor Dynamometer (FSPD) was propeller boat (Model 4500) was designed
designed to make precision steady state 36 years ago for the experimental
measurements of propeller thrust and requirements of that time. Model 4500
torque while moving through the water at had a maximum propeller center line
high speeds in the towing basin. submergence depth of 16.5 inches (41.91
Experiments may be performed at various cm), a maximum speed of 12.0 ft/sec
depths and shaft inclination angles. (3.66 misec) and a maximum torque limit
The capabilities introduced by this new of 200 in-lbf (22.6 N-m). In recent
instrument will make investigations into years the trend has been toward using
many phenomena and the measurement of larger propulsion models and propellers.
their effects on propeller performance Usual practice in CDNSWC
possible.
cavitation experiments conducted in the
The FSPD has been
used for 36-inch Variable Pressure Water Tunnel
obtaining the high speed open water (VPWT) and the Large Cavitation Channel
performance for the large diameter (LCC) involves the use of large model
propulsors used in cavitation propellers, over 14 inches (35.6 cm) in
experiments at CDNSWC. Experiments have diameter. The propellers are operated
been performed to determine the effects
of the variation of propeller depth of
submergence and Reynolds Numbers on open
water performance. Some preliminary
experimental results are presented in
this paper.

INTRODUCTION 7

The accurate measurement of

propeller thrust and torque in Open
Water is one of the most important
evaluation and design support tools in
model basin experimentation. The
results of Open Water Experiments are
frequently used in determining the S
propulsion characteristics of hull
designs. Open Water characteristics are FIG. 1. THE FULLY SUBMERGED PROPULSOR
also used by the Naval Architect in DYNAMOMETER (FSPD) DESIGN CONCEPT

339
and cavitation performance is evaluated carried to the carriage via overhead
at water velocities which are equal to trolly lines.
the full scale speed of the ship. It is Drive Motor and Control
felt that model propeller cavitation System. The drive motor needed to be
characteristics obtained at these high powerful enough for open water data to
Reynolds numbers result in improved be obtained at the highest Reynolds
correlation with full scale Numbers possible. The drive motor would
observations, by minimizing scale have to be accurately controlled so that
effects. propeller shaft speed could be precisely
It was obvious that the Model 4500 maintained. This required a motor
was inadequate to accommodate the large controller and feedback system which
diameter propellers which were required could regulate the drive motor shaft
for use with the new experimental speed by changing the shore generator
techniques. A new device for conducting output. All of this needed to be done
open water experiments was needed which while the FSPD was moving along the
would fill the immediate and future basin trolly system on board the
requirements of CDNSWC. carriage at varying distances from the
DC generator.
DESIGN CONCEPT Precise control of the motor shaft
speed would result in a more steady
The design of the Fully Submerged thrust and torque on the propeller
Propulsor Dynamometer (FSPD) was shaft. A steady shaft speed would
influenced by both technical desires and result in a more precise measurement of
practical limitations. These two the forces generated by the propulsor.
elements were balanced in order to The control of motor shaft speed would
create a workable design which would be a critical factor influencing the
satisfy our present and future overall performance of the FSPD design.
requirements. A sketch of the FSPD
design concept is shown in Figure 1. Shaft Thrust and TorquQ
Measurements. The FSPD would need to
Technical Needs measure shaft thrust and torque as
It was decided that many of our accurately as possible at low as well as
needs would be satisfied by an high loads. Strain gaged transducers
instrument which could accurately for measurement of thrust and torque
control shaft speed, measure thrust and would have to be used because of their
torque on a deeply submerged single accuracy and compatibility with our
propeller shaft, while moving at high existing signal conditioning and data
speeds. The FSPD would be used for open acquisition equipment.
water evaluations of larger diameter
propulsors and other propulsor related Special Requirements. We wanted
research. an FSPD which would be durable,
dependable, require minimal maintenance
practical Considerations and operate for a long time in the harsh
The FSPD would have to be as environment of the model basin.
compatible with existing instrumentation The FSPD needed to be able to
and facilities as possible in order to accommodate large propulsors weighting
minimize the cost of the project. up to 100 lbs (45 kg). It needed to be
able to make measurements at various
Power drive motor
Suriplx. The depths. The capability to measure
had to be compatible with the power propulsor thrust and torque at different
supply which was available to the towing inflow angles was needed. A method for
carriages. The only power source which attaching a Kort nozzle type of
could supply the motor power required propeller arrangement was required.
for the FSPD was a variable voltage DC
power supply (50 kW,0-400 volts, 125 FSPD DESIGN
amps, 67 HP). The generator is located
in an on shore location and the power is The FSPD was designed and built by
Kempf & Remmers GmbH of Hamburg Germany.

340
 It was based on the Kempf & Remmers H 39 the FSPD. The structure has been used
right angle drive water tunnel to support force measuring equipment on
dynamometer design. which a duct was attached. A ducted
propulsor arrangement used in recent
Drive Motor towing tank open water experiments is
The FSPD is driven by a shunt shown in Figure 2.
wound DC motor rated at 50 kW at 3000
rpm. The motor speed may range up to FSPD ACCEPTANCE
3000 rpm in right and left hand
rotation. The FSPD was delivered to CDNSWC
The shaft speed of the motor may in January of 1991. In February of 1991
be set by using either analog Or digital the acceptance procedure was begun by
control. In analog control the shaft CDNSWC. The acceptance procedures were
speed is set by turning a potentiometer agreed to by Kempf and Remmers GmbH.
until the desired shaft speed is
reached. In digital control the desired Thrust and Torque.
shaft speed is input by using a key pad, Part of the FSPD acceptance
entered and the motor ramps to the procedure was to determine how well the
desired shaft speed at a preset rate. thrust and torque sensors performed.
The thrust and torque sensors were
Shaft Spepd evaluated by performing a series of
Shaft speed measurements are made static calibrations. The results of the
using an incremental angle encoder and a static thrust and torque calibrations
gear wheel with an inductive pick-up ,910
which are coupled to the drive motor.

Thrust and Torque

The load sensors for the
measurement of thrust and torque are
full-bridge strain gage type
transducers. The maximum thrust
capacity is ±674 lbf (3 kN) and the
maximum torque capacity is ±1328 in-lbf
(150 Nm).

Shaft Depth
The propeller shaft depth may be
set anywhere between 0 and 5 ft (1.52 m)
below the water surface when the FSPD is
mounted to the vertical rails of the
CDNSWC towing carriage and the shaft is
in the horizontal position.

The propeller shaft inclination

angle may be set to 10 degrees above and
below the horizontal position using the
motorized inclination device which is
part of the FSPD. The shaft inclination
angle range has been expanded to 30
degrees above the horizontal by adding
special hardware between the FSPD and
the vertical rails of the CDNSWC towing
carriage.

,$1.1pport Structurp
Fig. 2. THE FSPD ON CDNSWC TOWING
A support for suspending equipment
CARRIAGE II WITH A DUCTED PROPULSOR
over the propeller may be attached to

341
are shown in Figure 3. The thrust and control system interfered with the
torque were required to be measured regulation of the DC power supply and
within ± 0.10 % of the full scale the thrust and torque signal channels of
reading. the dynamometry.
Analysis of the calibration data Kempf and Remmers GmbH redesigned
shows that thrust was measured to within the control system in order to isolate
a precision index of ±0.1 %. Torque was the electrical noise caused by solid
measured to within a precision index of state switching so that it would not
±0.05 %. cause interference with the strain gage
signal channels and the control signal.
Shaft Speed. Kempf and Remmers installed the
redesigned control system in August
The shaft speed regulation was
evaluated with the FSPD on shore. At 1991.
first, it was found that the shaft speed
The design requirements for shaft
speed were to control rpm within ± 1 rpm
control did not function correctly. The
electrical noise generated by the solid up to 1000 rpm and within ± 2 rpm above
1000 rpm. The shaft speed was recorded
state switching components of the
-19 during open water experiments which were
c,o
performed at a nominal shaft speed of
o () TORQUE CAL. SLOPE 540.1 rpm. The shaft speed and the mean
co
shaft speed are plotted in Figure 4.
Analysis of the shaft speed data
-- MEAN (in-lbf/volt) indicates that rpm was measured to
within a precision index of ± 0.01 %.

Open Water Data.

The FSPD was used to perform an
open water experiment with a model
-20 propeller. The open water results were
a o compared to the existing results
obtained with the propeller boat Model
o
4500. The open water experiments were
o performed at close to the same Reynolds
numbers. The depth to the center of the
-200. FSPD propeller shaft was 48 in
1 5 9 13 17 21 25 (121.92cm) and the depth to the shaft of
SAMPLE NUMBER Model 4500 was 16.5 in (41.91 cm). The
diameter of the model propeller was
16.326 in. (41.468 cm). The results
from these experiments are shown in
0 THRUST CAL. SLOPE Figure 5.
The data in Figure 5 show that
open water torque coefficients acquired
MEAN (lbf/volt) with the FSPD were within ± 1.5 % of the
torque coefficients ac uired with
q 540.2

P5401
4=
540.0
11 16 21 26 1 6 11 16 21 26
SAMPLE NUMBER SAMPLE NUMBER
SHAFT THRUST AND TORQUE Fig. 4. FSPD SHAFT SPEED CONTROL AT
Fig. 3.
CALIBRATION RESULTS 540.1 RPM

342
Model 4500 in the range of propeller RaRluLt-abj-LLLy
advance numbers between 0.150 and 0.950. Open Water experiments were
Thrust coefficient was between 1.5 and performed with FSPD using a 15.856 in.
2.25 % higher when recorded with the (40.274 cm) diameter propeller on two
FSPD. different occasions. The faired results
The reasons for differences in the from these open water experiments are
measurement of propeller thrust between presented in Figure 6. Torque
the FSPD and Model 4500 are not clear coefficients were within ± 0.8 % for all
and are being studied at this time. values of J greater than 0.150. Thrust
coefficients were within ± 0.8 % for all
FSPD EXPERIMENTATION values of J between 0.20 and 1.55.
Depth
Since August 1991, the FSPD has Open water experiments were
been used for many propulsor open water performed with FSPD shaft center line
evaluations. These have included a depth at a 16.5 in. (41.91 cm)
preliminary evaluation of how depth and (depth/propeller diameter,d/D=1.04) and
Reynolds number variations effect 48 in. (121.92 cm) (d/D=3.03). The
propeller performance, the open water experiments were performed with the same
evaluation of a ducted propulsor, and propeller which was used to check the
the evaluation of propeller performance repeatability of the FSPD in Figure 6.
at different shaft angles. Examples of Faired results from the open water
the types of preliminary results which experiments are presented in Figure 7.
we have obtained from the FSPD are Preliminary trends in the open
presented in the sections that follow. water data in Figure 7 indicate that
propeller thrust and torque coefficients
are indeed effected by variations in
depth of submergence. Preliminary data
indicate that propeller thrust and
1.8 - - -KT EXP. 3

1.6 KT EXP. 5

- - -10KQ EXP 3
1.4
- 10KQ EXP 5
1.2
0

a3 0.8

0.6
0.4
0.2
11111E1'111 II 111111

0 0.2 0.4 0.6 0.8 1.2

1

PROPELLER ADVANCE NO (J)

11111 1111111 If
Fig. 5. OPEN WATER COMPARISON OF
0.4 0.8 1.2 16
EXPERIMENTS WITH THE FSPD & MODEL 4500 PROPELLER ADVANCE NO (J)
Fig. 6. REPEATABILITY OF OPEN WATER
EXPERIMENTS PERFORMED WITH THE FSPD

343
torque coefficients increase with the Experiments were performed with a
depth of submergence. The KT and 10KQ 5 bladed propeller which was 14.633 in.
were increased by over 1 % at a J of (37.168 cm) in diameter with NACA/DTMB
modified blade sections. The Re was
0.550. At a J of 1.250 (near peak
propeller efficiency) the KT was computed by using the blade section
increased by 5 % and 10KQ by 3 %. chord length and the combined forward
open water and rotational speed at the 70%
Additional Plots of KT and 10KQ
experimentation must be performed to propeller radius.
validate the trends noted in this against Re are presented in Figure 8.
limited and preliminary study The data in Figure 8 show that
10KQ increased with increasing Re to a
geynolds Number Effects maximum value at about 6 x 105. KQ then
A few experiments have been gradually decreases downward by about
performed to determine how propeller 2.2 percent between 6 and 19 x 105 and
open water characteristics are appears to level out at a Re of 2 x 106.
influenced by changes in blade Reynolds The data show that KT increased
number (Re). The experiments were with increasing Reynolds Number to a
performed by increasing model propeller maximum value at about 6 x 105. KT then
shaft speed and carriage speed while gradually decreases downward by about
keeping the design propeller advance 1.2 % between a Re of 7 and 19 x 105 and
coefficient (J) constant. In this way
appears to level out at 2 x 106.
the effect of only changes in Reynolds
The propulsor efficiency (no)
number on propulsor thrust and torque increased by about 1.2 % between a Re of
could be determined.
1. 0.7
- - KT d/D .1.04
o
X**
- KT d/D=3.03
ea
1.5- - - 10KQ d/D=1.0.
10KQ d/D=3.03

1.2-
0.6
10 14 18 22
Rn x 10E-05

0.6-

0.3-

10 14 18 22
o , ,

1.6 Rn x 10E-05
0.4 0.8 1.2
Fig. 8 PRELIMINARY RESULTS:EFFECT OF
PROPELLER ADVANCE NO (J) REYNOLDS NUMBER ON OPEN WATER
Fig. 7. PRELIMINARY RESULTS: EFFECT OF PERFORMANCE
DEPTH ON OPEN WATER PERFORMANCE (NACA/DTMB MODIFIED BLADE SECTIONS)

344
5 and 21 x 105. The propeller SUMMARY
efficiency appears to level out near a
Rn of 21 x 105. The FSPD gives the CDNSWC modern
Further investigation is needed so and unique capabilities for studying and
that the trends in the limited data evaluating marine propulsors. The FSPD
presented here can be validated. is a precision instrument capable of
performing open water and similar model
Ducted Propulsor Performance basin experiments:
The FSPD has been used to evaluate at High Reynolds Numbers
the performance of a ducted propulsor. at Deep Submergence Depths
The experimental arrangement and the at Shaft Inclination Angles
propulsor is shown in Figure 2. Rotor with Ducts and Complex Propulsors
thrust and torque were measured along Our future plans include pursuing
with the axial force of the duct and the work involving Reynolds Number and
stator unit. The rotor thrust (KT), depth effects on propulsor performance
rotor torque (KQ), duct axial force presented in the previous section.
(Kf), and total propulsor thrust
(Kft=KT+Kf) coefficients are presented ACKNOWLEDGEMENTS
in Figure 9.
The purchase of the FSPD was
Performance At Angles funded by the U S Navy's Asset
The FSPD was used to evaluate a Capitalization Program. The authors
would like to acknowledge individuals in
propeller at different shaft angles (a). CDNSWC Codes 15, 152, 1522, 1505, 3322,
The results of these experiments are
presented in Figure 10. The data show 301 and on the Federal Executive Review
that changes in the shaft angle between Board for their support in the funding,
10 and 25 degrees produce a much greater procurement and on site preparation for
increase in the propeller thrust and the FSPD. We thank Mr Paul Douglas of
torque than similar changes in.shaft Vector Research Inc. for his valuable
angle between 0 and 10 degrees. consultation on electrical issues. We
2 acknowledge Kempf & Remmers GmbH for the
design and manufacture of a quality
instrument to suit CDNSWC's unique
requirements.
0.

O.

Y 0.

O.

O.

0.5 1 1.5 2 2.5

1 2 3 PROPELLER ADVANCE COEF (J)
PROPELLER ADVANCE NO (J) Fig. 10. PROPELLER OPEN WATER
Fig. 9. DUCTED PROPULSOR OPEN WATER PERFORMANCE AT VARIOUS SHAFT ANGLES ON
PERFORMANCE THE FSPD

345
 SESSION IV-B

HIGH SPEED CRAFT-I

 The High Speed Craft Committee Report
J.G. Hoyt, III (Carderock Division, Naval Surface Warfare Center, USA)

The High Speed Craft Committee was forever in his debt.

fortunate this year and received nine
contributions. Two of these, RECENT WORK
"Development of a Waterjet Propulsor for
the Marine Corps High Speed Amphibious Activity in the various research
Craft" and "Run Length and Statistical facilities involving high speed craft
Error Estimation for Seakeeping Tests has been limited. This is the same
and Trials", were placed in other trend as was discussed by the committee
sections. Of the remaining papers, four report of the 22nd ATTC. A brief
of the seven discuss aspects of the discussion of the research work
mono-hull planing craft and two papers performed by several of the research
involve" air cushion and surface effect facilities is as follows:
craft.
ARCTEC Offshore Corporation
The High Speed Craft Committee Mike Musatow
has, as it did for the 22nd ATTC,
included a paper that looks at our AAAV. Throughout the period from
industries' past. During the 22nd ATTC 1990 to 1992 Arctec continued it's
we had a contribution by Professor support of the Advanced Amphibious
Thomas Gillmer discussing the design of Assault Vehicle (AAAV) development by
the Baltimore Clipper "Pride of conducting Phase III and Phase IV test
Baltimore II", and this year we are programs for FMC. The test programs for
pleased to present a paper that the hybrid concept, 29 foot LOA, foil
discusses the origins of the Frigate assisted planing vehicle were performed
Constellation, which is currently a at 116th scale.
floating museum in Baltimore's Inner
Harbour. As much as this committee The Phase III program involved
looks into the marine industries' resistance testing in calm water and
future, we have the duty to remember its head seas of sea state 3 up to 28 and 25
past. knots, respectively. The test matrix
was designed to collect data for
Finally I would encourage you to analysis of effects of changes in gross
read a special contribution from Mr. vehicle weight, longitudinal center of
Peter ward Brown. Peter needs no gravity, and foil angle of attack.
introduction due to his numerous Acceleration measurements at the vehicle
contributions, especially in the area of center of gravity and the drivers
planing craft. He has submitted to me, compartment were used to estimate the
and I have included in this committee likelihood and level of seasickness and
report his formulation for an explicit discomfort based on ISO 2631 guidelines.
approximation for the ATTC friction
coefficient. All of you that have The recently completed Phase IV
extrapolated by hand, created computer program involved conducting self-
"lookup" tables or interpolation schemes propulsion tests of the vehicle and a
of Schoenherr friction line will be revised foil optimization in calm water.

349
Self-propulsion of the AAAV presented a conducting underwater flow visualization
significant engineering problem. The for alignment of the propeller shaft
solution was to develop a belt driven Z- struts and analysis of flow patterns
drive type system. In general the around the turn of the bilge and in way
propulsion system worked well. Further of a bow thruster. These tests were
will be undertaken to performed from 16 to 30 knots. A
refinements
improve the system. significant improvement in the form
factor and resistance to weight ratio
Also, a light weight free-to-surge were recognized.
towing subcarriage made of carbon fiber
was developed. British Columbia Research Corporation
Gerry Stensgaard
Planing Hull. In June 1991, calm
water drag tests of a 47" double chine Ocean Engineering Center High
planing hull having an adjustable, full Speed Vehiclesearch. Since
beam transom extension were conducted on 1989 we have been working
behalf of the Army Corps of Engineers collaboratively with the National
Marine Design Center. Tests of 20 Research Council's Institute for Marine
configurations were executed at speeds Dynamics (St. John's, Newfoundland) to
between 15 and 30 knots. The stern first verify and then extend Fridsma's
work with simplified planing hulls. In
extension was adjustable to 0°, 10, 30,
and 5°, below the baseline. Tests were addition to the original Fridsma hulls
executed for full load, 50% load, and (L/B = 4,5,6, Deadrise = 10,20,30
10% load cases and at various locations degrees), we have added hulls of the
of the LCG. Additional tests were following proportions: (L/B = 2,3,4
performed for the original hull form, Deadrise 10,20,30 degrees).
Experiments have been run in calm and
40' without the transom extension.
photography, surface rough water. The work is on-going and
Underwater
photography of the wave profile, and has so far resulted in a paper entitled
"Model Experiments on Simplified Planing
video records were maintained to
facilitate determination of wetted areas Hulls in Regular and Irregular Waves",
and lengths for use in data analysis. A by Molyneux and Roddan which was
standard thrust unloading method was presented at the Marine Dynamics
employed for these test. The tests were Conference held at the Institute for
Marine Dynamics in May, 1991.
preformed at 1/ 8th scale in deep water.
These tests are the subject of a paper
On the commercial side, since the
to be presented at the 1992 High last ATTO, we have tested numerous high
Performance Marine Vehicles Conference.
speed motor yachts, several SES's and a
In 1991 high-speed military amphibian.
High Speed Motor Yacht.
Arctec conducted bare hull resistance
Davidson Laboratory
tests of a 182 ft motor yacht in calm
Edward Lewandowski
water and head seas of sea state 4 at
speeds from 16 to 27 knots. The tests In
Maneuvering of Planing Hulls
were conducted for G.I. Robinson Yacht
support of the US Coast Guard's pursuit
designs at a scale of 1112th. Following of R & D projects that will enable it to
these tests, the design length was evaluate advanced marine vehicles, the
increased to 197 ft and a stern/transom Davidson Laboratory has completed an
was undertaken at the
redesign extensive study of the directional
A test program to
suggestion of Arctec.
was stability of planing boats using
evaluate the design changes rotating arm tests of five models. The
performed. The bare and appended hull model series is generally representative
resistance program was enhanced by of a 100 ton, 100 ft hard-chine patrol

350
boat. Two of the models were taken in apparatus construction, the Davidson
all conditions, and video recordings of Laboratory is again providing
all runs were made. measurements and analyses of amphibious
aircraft performance. The following
The seakeeping behavior of the WPB tests have been used to evaluate several
hulls in headseas up to 15 ft new aircraft during the last few years:
significant height was determined, and I. Calm Water
their beam-sea rolling response at zero Low Speed Resistance and
speed was measured using a new free-to- Spray - Free to Trim and
sway apparatus. Heave
High Speed Resistance and
A new free oscillation apparatus Spray - Fixed Trim
was used to measure the roll exinction Dynamic Longitudinal
characteristics at high speed of the Stability (Porpoising) -
prismatice hulls on straight course at Simulated Wing and Tail
fixed trim and yaw. Empirical Aerodynamics
expressions for the added inertia and Lateral and Directional
roll damping were obtained. Stability - Fixed Yaw and
Roll
The results of the stability tests II. Irregular Seas - Scaled
are presented in a data base of well Reproducible Sea States
over 5,000 runs, including tabulated Motions Response Floating
values of forces, moments, drafts, and at Various Headings
wetted lengths in both dimensional and Impact Loads and Motions
non-dimensional form. Measurements Simulating
Landing and Take-Off
Seaplane Testing. When any new Operations
water-based aircraft design is started Low Speed Bow Spray
and a preliminary geometry has been Observations to Determine
determined, towing tank tests must be Limiting Sea State
carried out at an early stage in hull Capability
design to assure acceptable hydrodynamic
performance and stability on the water. David Taylor Model Basin
Resistance must be low enough to permit John Hoyt III
take-off with the available thrust or
more thrust must be provided. Spray AAAV. The advanced amphibious
must clear the canopy, propellers or jet assault vehicle program has continued
inlets, wing flaps, and horizontal tail. through the time period since the last
Longituinal stability must be assured ATTO. The AAAV is an amphibious
throughout take off and porpoising must personnel carrier with an expected gross
not develop within the general operating vehicle weight between 60,000 to 70,000
hull trim angles required for take-off lbs, and a desired overwater speed in
from water. Lateral or roll stability excess of 25 knots. David Taylor has
appendages such as wing tip floats or been heavily involved in this program.
sponsons must be represented where the
appendage can strike the water or In support of the AAAV program the
influence the loads and moments. full scale trial of the high water speed
Landing and take-off impact loads and technology demonstrator (HWSTD) has been
motions must be determined as a function successfully completed showing the
of the sea state operational conditions hydrodynamic feasibility of a craft of
anticipated for design load this sort. The HWSTD proved that speeds
considerations. in excess of 25 knots in seas of 2 feet
were possible. HWSTD achieved a top
After a twenty year pause and new speed of 29.4 knots. The propulsion

351
system demonstrator is currently under heeling moments were applied to the
test by DTMB and has already achieved a model which was free to heel, trim, and
top speed of 28.5 knots. rise. Experiments were conducted at
In addition to the DTMB designed volumetric Froude Numbers up to 2.3, and
vehicles, government testing of the two it was found that for large heel angles
competing concepts for the production (greater than 15 degrees), the
AAAV by FMC Corporation and General transverse stability of the model
consistently increased with speed. For
Dynamics Land Systems has been ongoing.
This program has pushed the boundaries heel angles less than 15 degrees, the
of our known technology with the study transverse stability decreased around
of planing craft of extremely high body hump speed and was gradually restored as
loadings and hybrid foil assisted speed was increased. Details of the
planing craft. tests are presented in Reference A.

Planing Craft. In addition to the Another test program at NAHL

before mentioned planing amphibians, a compared the seakeeping performance of
two semi-planing patrol craft. The
manned scaled evaluation of a twin hull
assisted planing craft, models were nearly identical except that
foil
"Wavestrider", designed by Peter Payne one hull had hard chines while the other
had round bilges. A unique aspect of
has been completed.
this program was that the models were
Currently a joint effort with towed simultaneously, side-by-side, so
SAIC/MARIMTEK has begun to develop a that they encountered the same waves.
computer tool for the design of planing Since the two models experienced
It is the goal of this effort to identical waves, the shortcomings
craft.
provide design information as to the normally associated with representing a
resistance, powering, and seakeeping of model's response to random waves using
statistics, were minimized. Under the
planing craft.
conditions tested, the round bilge hull
.SES. Work on the German SES 700 had consistently larger motions and
fast test craft for the Ministry of accelerations than the hard chine hull.
Defense of the Republic of Germany has It was postulated that the hard chine
Further work on the hull damped heave motions more
been completed.
U.S. Navy's SES 200 is also continuing. effectively than the round bilge hull.
The SES 200 has just recently finished a
The reduced heave motions resulted in
Ongoing reduced vertical accelerations. Details
refit to Kamewa waterjets.
studies utilizing this craft for ride of the tests are presented in Reference
control are continuing, and speeds in B.
excess of 46 knots have been reached by
References. A) Compton, R.,
this craft.
Zseleczky, J. and Abrams, W., "Underway
Inclining Experiments Performed on a
U.S. Naval Academy Hydromechanics ASNE High
Planing Hull Model,"
Laboratory
Performance Marine Vehicles Conference,
John Zseleczky
Washington D.C., June 1992.
High Speed Vehicles. The U.S.
B) Zseleczky, J., Nehrling, B. and
Naval Academy Hydromechanics Laboratory
has been active in testing Compton, R., "Side-By-Side Testing of
(NAHL) Hard Chine and Round Bilge Semiplaning
planing and semi-planing craft over the
In one test program, Models in Waves," ASNE High Performance
past three years.
Marine Vehicles Conference, Washington
the overall transverse stability of
planing boats at post-hump speeds was D.C., June 1992.
investigated using an unappended 1:8
model of a current USCG hull. Fixed

352
 A Note on the ATTO Friction Coefficient
Peter W. Brown'

In many cases it is convenient to

have an explicit expression for the
friction coefficient CF. For instance
during model tests of planing hulls, the
full size resistance should be
calculated after each test run so that
the thrust unloading may be estimated.
For these cases the following note may
be of interest.

The ATTO friction coefficient [1]

is given by the implicit expression:

0.24.2[N1CF = log (Rn Cr.) (1)

A simple explicit approximation

for this coefficient is:

0.25[N/CF = 0.7 + (log Rn - 2)

- 1.553 log(log Rn - 2) (2)

This expression is accurate to

within ±0.000001 over the range of
Reynolds number from Rn = 106 to Rn --

109. It is thought to be as good as the

tabulations in general use.

Reference. 1. Uniform Procedure

for the Calculation of Frictional
Resistance and the Expansion of Model
Test Data to Full Size. SNAME Bulletin
No. 1-2, August 1948.

'Consultant to the Davidson

Laboratory, Stevens Institute of
Technology

353
 Discussion of the Origins of the
Frigate and the Sloop Constellation
K.M. Lynaugh
(Carderock Division, Naval Surface Warfare Center, USA)

PREFACE evaluate three hull forms which represented the USS

Constellation at different time periods in its history and give
Alleged to have been built in Baltimore, Maryland in 1797, an opinion as to whether the ship hull of today is the same
the USS CONSTELLATION was transferred in 1955 to a as in 1795; the Baltimore theory or the 1853 Norfollc
group of Baltimore citizens for preservation and display. theory. The methodology was divided into three phases:
Initiating a protracted, vitriolic and public dispute, some
maritime historians lrave c/aimed since 1947 that the ship Model each set of artifactual data in the computer
presently exhibited is another warship with the same name using bspline curves and surfaces.
built near Norfolk, Virginia in 1855. Has the ship had a Cut each hull at the same locations.
continuous identity since 1797 or are those claims a mistake Overlay the sectional, waterline, and buttock
ora hoax? In March 1989 the U.S. Navy's Curator of Ship curves and note the discrepancies.
Models, situated at the David Taylor Research Center,
located and identified in the Navy's vast collection, the 1853 1.0 EVIDENCE
designer's half hull model representing the design for a new
CONSTELLATION. Further investigation by the Curator's Three sets of hull form evidence exist in the form of
office indicated that essentially all of the historical records drawings, first a design lines plan from 1795, second a
citeid, clearly substantiating the 1797 origin of the present design lines plan from 1853, and thirdly a keel drawing and
ship, had probably been forged by a CONSTELLATION frame survey drawing, both from 1853. Other evidence was
employee between 1956 and 1965: (for a copy of the in the form of an offset booklet from 1853.2 The booklet,
complete report "FOULED ANCHORS") or tables, contained x, y, and z coordinates which were used
to build, or loft the ships frames at full scale.
In addition to his scientific examination of the designer's
half model and the investigation of the historical Two of the hull forms are represented in typical naval
documentation used to support the 1797 origins of the architecture lines drawing format. This format shows the
present vessel, the Curator requested that the author study, form of the ship hull drawn to scale (114 inch to 1 foot and
from a modem naval architect's point of view, three pivotal 1/3 inch to 1 foot respectively) using the method of
archival technical drawings of the CONSTELLATION. The descriptive geometry. The first hull form was a 1795 lines
author's architectural study; the subject of this paper, drawing of the USS Constellation as designed by Joshua
confirmed both the artifactual and the documentary findings: Humphreys and drawn by William Doughty. The second
the ship that is displayed in Baltimore today was built near was an 1853 lines drawing of the USS Constellation
Norfolk in 1855. attributed to John Lenthall and a mould floor offset booklet
which was from the 1853 design of Lenthall. The third
Though not built in Baltimore in 1797, the present ship was piece consisted of a keel drawing, dated January 1853, and
the last sail-only powered warship designed and built by the a set of transverse sections from a frame survey, dated
U.S. Navy. As such, it is a rare artifact of first importance February 1853, both performed at the Norfolk Virginia
and truly deserves to be preserved and displayed for the Navy yard. Secondary evidence exists in the form of
American public. traditional naval architecture mathematical calculations.
This evidence is hydrostatic curves of form drawings, a set
INTRODUCTION of bonjean curves, and curves of statical stability?
Although this secondary evidence exists only for the 1853
The following discussion pertains to the work done for the hull form, analysis has been performed for the 1795 hull
David Taylor Research Center's (DTRC) model curator on and is compared (see section 2.0).
the USS Constellation ca 1795. The author was asked to

355
 SHCP is a general naval architecture program used Navy
2.0 METHODOLOGY AND MODELING OF THE DATA wide for static ship analysis and was used to obtain the
(SOFT/HARDWARE) hydrostatic curves of form (at drafts from 2 to 24 feet).
This program computed statical stability information about
Figure 9 shows an overview of computer hardware and each vessel and produced an AutoCad dxf drawing file of
software used to perform this analysis. The hardware the hydrostatic curves of form. The curves of stability for
consisted of a Digital Equipment Corporation (DEC) micro each vessel are compared in similar fashion (see Figs. 12,
computer (microVaxII) with three dimensional naval 13, and 14).
architecture design software and an IBM /AT compatible
computer with two dimensional drafting and word Bodyplan, XYZ, DXF are fortran programs that were
processing software. written at DTRC. SHCP is a fortran program consisting of
several subroutines which perform standard naval
The course of this naval architecture methodology is as architecture calculations (i.e. Hydrostatics, trim lines,
follows. The three hull forms were digitized into the Naval longitudinal strength, etc.). The IBM AT Compatible
Sea Systems Command (NAVSEA) Hull Form Design programs AutoCad from Autodesk, Inc. ver 10.0 and
System (HFDS). A schematic of the system with the Wordperfect from Wordperfect Corp. ver 5.1 were used
modules used is shown in figure 10. The HFDS consists of extensively to display not only the engineering drawings, but
a large number of fortran computer programs (not all are to construct this report.
shown) that reside on a DEC microVaxII computer at
DTRC and are used for naval architecture desig,n and The investigation centered around the data in the January
analysis. The main programs used to perform this analysis and February 1853 survey performed in Norfolk, Virginia.
were Bodyplan, XYZ, Blines, DXF and Ship Hull It was determined that if one of the hull designs from 1795
Characteristics Program (SHCP). Bodyplan was used to or 1853 matched this frame survey then it would prove
digitize the existing lines drawings and create a SHCP input either:
file (offsets of stations). XYZ was used to take the SHCP
input file and create a input file for Blines. Blines is a If the frame survey of 1853 matched the 1853 Lenthall
commercially available computer code from British desimi then only one design/ship existed and that the design
Maritime Technolog.y (BMT) and has been used for lines was far ahead of typical naval architecture for the 1795
fairing and preliminary ship design since 1982 by the US period. It would also provide the heritage ne,eded to support
Navy. Blines is a wire frame modeler (i.e. it is not surface the theory that the 1955 (and present day) ship is the
but individual curves). The program uses cubic bsplines to original frigate of 1797 built in Baltimore.
represent curves for sections (frames), waterlines, buttocks,
diagonals, stem and stem profiles, knuckle curves, and If the frame survey was similar to the 1795 Humphreys
shear lines. The program interpolates surface points (x,y,z) design, then this is to be the key argument for construction
to fit the curve, uses vertex points to adjust the curve, and of a new vessel in 1853. It would prove the contention that
inverse radius of curvature (2nd derivative of function the original frigate from 1795 was in such disrepair that a
describing curve) to evaluate smoothness or fairness. new and modern design was constructed in 1853-55 at
Norfolk.
The 1795 and 1853 lines plans were digitize,d into the Vax
and were cross-faired on Blines. It was noted that both the Based on the geometry of the survey, both the 1795 and
1795 and 1853 plans required only small adjustments (inches 1853 designs had frame se,ctions taken at the same
or less, full scale) in order to cross fair. The 1853 Norfolk longitudinal sections.
survey was not faired as it is considered as live data dire,ctly
3.0 FINDINGS AND COMPARISON OF THE
off the existing ship. The frame survey drawing labeled
USS Constellation dated February 1853 contains nine (9) DIFFERENT HULLS
sections. The section offsets were measured at every two
(2) foot waterlines from the rabbet line of each station up to It is my opinion that the 1853 survey matches the original
the rail (planking was removed at this point). The data 1795 Humphreys design. It is interesting to note that the
points were fit with bsplines but not faired. This was to 1853 survey not only showed a hog in the vessel of
approximately 19-20 inches, but shows that the vessel had
retain the curves (bsplines) through the data points. At this
point plot files were generated from BLINES and DXF was a slight twist in the aft quarters. This hog and twist can be
used to convert them into AutoCad .dxf format ready for seen in figures 15-16a. To show the sinailarities of the two
AutoCad insertion and further engineering study. hulls, the frames from the 1853 survey were overlaid on the
1795 frame cuts and are shown in figure 16. On

356
 comparison with the same longitudinal section cuts or Some, from the Baltimore theory side, proposed that the
frames from the 1795 design, the curves match rather well. clipper bow visible today was retained from the 1795
This is especially true when the curves for the survey are design. This is not true. The 1795 design has a rounded
moved to the same origin (rabbet line) as in the 1795 design hull form, while that of the 1853 design is of a clipper (or
(see figure 17). However, when the same cuts are made in sharp) form, both forrns typical of their laistoric period. It
the 1853 Lenthall design and compared to the 1853 survey, would not have followed the typical naval architecture
the discrepancies between the frames are noticeable and in practice of making small progressive modifications to
my opinion are not of the same vessel (for possibility of existing designs. If radical design changes occurred, they
converting existing frames to the newer design see section generally happened in the merchant or private design
8.0). community and not in the Navy.

4.0 ORIGINAL SHAPE OF 1795 FRIGATE BELOW GUN Typical naval architecture is a slow evolution which starts
DECK AS COMPARED WITH EXISTING DESIGN from a previous design and makes minor modifications for
the next iteration. It is also noted that during the 1840's-
The 1795 design and 1853 design are compared in one 1850's there appears to have been a renaissance in naval
form. Sectional (frame) slices have been made in the 1853 architecture within the Navy. Ships were being built more
design hull and compared with the 1795 hull design. The for speed and handling agility than for "show the flag" size,
sections are compared only below the gun deck. Figure 18 and stable gun platforms. It is also noted that new naval
shows the body plans of the 1795 design, the 1853 design, vessels were similar in hull shape to comparable new
and the combination or overlay of one on the other. It is merchant vessels of the period. Merchant vessels were
highly unlikely that this marriage of a wooden hull structure constructed for speed, sailing ability, hull strength, and
would be possible, both for te,chnical and fiscal reasons. It stability at a variety of drafts.
is a fact that today this type of conversion is performed, but
it is done with steel and welding, not by nailing and 6.0 OPINIONS ON THE USE OF SURVEY EOUIPMENT
pinning. Also, the present hull would reveal a major scarf AND THE SLIP
zone.
The February 1853 survey was apparently performed at the
4.1 PROPOSED 12 FOOT EXTENSION TO MAKE 1795 North slip in the Norfolk Navy yard. The slip was a stone-
DESIGN MATCH 1853 DESIGN faced incline used to pull ships out of the water or to launch
them. The most important aspect of dry docicing a large
Although the 1795 design is 163'-8" length between vessel is in supporting the keel and bilges. The supports
perpendiculars (lbp) and the ship today is 176' lbp it is prevent localized stresses from damaging the keel (using
assumed that if a extension of 12 feet (a small increment for keel blocks) and keep the ship from rolling over on its side
all the work involved) were performed, it would have been (using bilge blocks) when it is hauled out of the water. The
done at the midships section or slightly forward. The device believed to measure the shape of the keel before the
technique of extending a ship was known to have been used Constellation was hauled out is shown in figure 25.3 The
during the 1850's on at least two naval vessels. Both the process consisted of two barges with a man stationed on
USS Sabine and the USS Santee were lengthened in the bow each barge to pull or release a measuring line as a
and stern.' However, both vessels were new constructions pantograph- type mechanism under the keel adjusted to the
and the modifications took place on the ways before shape of the keel. At the same time the pantograph pulled or
launching. No planking or major construction had yet taken slacked on the measuring line, readings were taken. The
place in the forward and aft frames when they were barges were pulled from the bow to the stem and
modified. No evidence has been found which supports measurements were taken at the respective frames which
modifying an existing US naval wooden vessel into an would rest on keel blocks. Recording the numbers had to
entirely new design. Most major modifications performed have taken place in very calm water and conditions. The
on vessels were to razee or remove upper gun decks. One numbers were called out by each man on the barges and
result of this practice was to increase stability under sail. recorded. This information was then provided to a
The effect of razeeing was moving the effective center of draftsman to draw up the docking plan. The docking plan
gravity lower in the hull. Figure 20 shows an isometric of was then used to set the keel and bilge blocks in the slip or
both vessels and figures 21 and 22 show shaded images of construct the cradle to hold the vessel while it was being
the vessels. It is noted that the two designs are not similar. hauled out. The ship would then settle on the correctly
positioned blocks. These blocks prevented the ship from
rolling or being damaged by uneven pressure on the hull
5.0 ORIGINAL CLIPPER (THINNED) BOW RETAINED? during the overhaul period.

357
 ship will start showing, her age by developing a noticeable
hog. At this time the ship's structural stability is in question
and a survey would be in order to assess action to either
7.0 OPINIONS OF THE USE OF WOOD AS A repair, or scrap the vessel.
STRUCTURAL MATERIAL
8.0 CONCLUSION
Although the comparative abundance of wood has made it
a natural material for homes, structures, and ships, it is by With the examination of the evidence given to me
no means the ideal construction material for ships. But with completed, my conclusion is that the USS Constellation as
America's great supply it was the material of choice not it exists today in Baltimore, Md., is not the ship that was
only in quantity, but with regard to cost and a large desig,ned and constructed in 1795 but a new design and
experienced labor pool. These factors made its use popular construction of 1853.
until the late 1800's, and with some commercial vessel
construction, well into the twentieth century. Subjecting a wooden structure to over half a century in the
marine environment, and the limited knowledge of
The principal types of wood used for ship construction were preservatives during its life, it would not have been justified
white and live oak for the frames (it was also used in fiscally and technologically to modify the old Constellation
selected areas for planking) and southern yellow pine for into a more modern design. Furthermore when wooden
planking. Live oak is fairly impenetrable by liquids and has ships were repaired, it was to local damage of bow and
one of the highest modulus of elasticity out of all the stern posts, keel shoes, rotted or damaged planking, or
domestic woods. It also had many of the natural shapes limited number of broken or rotted frames, and beam knees,
needed for curved structural members. Yellow pine on the not a large scale redesign "cut and paste" effort. Figure 26
other hand, grew straight, was easily worked, and had low shows possible additions to the 1795 frames to make them
shrinkage. These factors made its use ideal for planking,. similar to the 1853 design. It is in my opinion that this
exercise would not have been allowed by the designer, the
A major draw back for wood is that it is best to keep it ship constructor, or the Navy.
under a controlled environment (stable humidity or
submerged). The worst case is when it is subjected to a The theory that the Constellation was a radical design for
continually wet and dry environment. Failure as a structural the 1795 period is not a possibility. The 1795 design and
material is imminent if it is not kept dry or completely 1853 survey data match in form with other ships built in
submerged and protected (i.e. paint, coatings, chemical that era, while the 1853 design matches well with its design
treatment, etc.) from the environment. A practical contemporaries. Hull changes so radical to the historic
everyday example of this is the observation that fence posts period would not be allowed by local individuals, on a
will always decay at the post/ground interface. Another Federally funded project, as has been suggested. The
problem with wood is that it is cellulose and a part of the changes would not be allowed by a traditionally conservative
food chain for many organisms. The following are Naval design community, without supporting calculations
organisms which degrade wood: Fungi, bacteria, insects, and tests, in order to confirrn an action to build and field
and marine animals. Fastenings such as nails, trunnels, and such a vessel on the relentless sea.
bolts are pathways for many of the organisms and aid to its
Finally, the USS Constellation that exists in Baltimore today
eventual failure.
is a fine example of 1853 naval history. It holds a place in
A typical problem with large wooden vessels is the the evolution of US naval architecture. The ship of 1853
flexibility of the material under load. Wood will, with time, demonstrates the major advancements in hull design from
creep and conform to whatever external forces are acting on the state of the art of 1795, into that of 1853 which allowed
ships greater speed and better handling in a seaway. If the
it. Thus, a wooden vessel has a tendency to hog, within a
number of years. This hogging is due to the large buoyancy ship is to be preserved it should be as a superb example of
force acting in the inidships area and little or no buoyancy 1853 period design, not 1795.
force available at the bow and stern. As a historical note,
shipwrights would build a hang (reverse hog) into the keel Acknowledgments
of several inches to offset this large buoyancy force. With
this poor match of opposing forces, especially in the The author would like to thank Dana M. Wegner, Liz B.
midship se,ctions, as the wood loses its strength due to Douglass, Michael D. Condon, and Colin Ratliff for their
absorption of water and fungus attack coupled with the help in preparing this document.
working and loosening of fastenings in a seaway, etc., the

358
ENDNOTES LIST OF FIGURES

"Fouled Anchors: The Constellation Question Schematic of the Hull form Design System and the two
Answered" by Dana M. Wegner with appendices different computers used in study.
by CoIan Ratliff and Kevin Lynaugh. DTRC-
91/CTO6 of September 1991. A copy of the report Detail of specific Hardware and software on both the
is available from U.S. Department of Commerce, VAX and IBM AT/compatible used.
National Technical Information Service, 5285 Port
Royal Road, Springfield, VA 22161, 703-487- AutoCad Drawing of SHCP output graphs for
4650. (ref. NTS ADA 241-916, $35.00 ea). hydrostatic curves of form, 1795 hull design.

A plan dated May 1853 is in Lenthall Collection, AutoCad Drawing of SHCP output graphs for
Franklin Institute. The dimensions taken from the hydrostatic curves of forrn, 1853 hull design.
old ship prior to and after hauling out are 107-13-
4A and B; a June 1853 plan is drawing 28-3-5; the AutoCad Drawing of SHCP output graphs for
offsets are 142-1-7, all in Record Group 19: hydrostatic curves of form, 1795 survey of 1795 design.
Records of the Bureau of Ships, entry 1.26, "Plans
of Ships and Shore Establishments," National Shaded image of 1853 survey of 1795 hull design.
Archives and Records Administration (hereafter
cited as RG-19, E-126, NARA). Body plan of 1853 survey overlaid on body plan from
1795 design.
"U.S.F. 'Constellation' Hydrostatic Curves." L.D.
Polland, 1 June 1961. Drwe.. C-100-6-(A); 16a. Body plan of 1853 survey overlaid on body plan from
"U.S.F. Constellation Curves of Forrn." E. Kangas 1853 design.
and C. Anderson, 23 Oct 1964. Drwg. C-100-
106-(B), both from Polland Papers. 78-11-10, Midship curves from 1795 design and 1853 survey at
"U.S. Ship Constellation." Bonjean curves. 31 Oct same origin and rotation.
1888; 28-3-12, U.S. Ship 'Constellation Curve of
Statical Stability." W.S. Mintoyne, 26 May 1888, Body plans from 1795 and 1853 designs.
both from RG-19, E-126, NARA.
20 Isometric of both 1795 and 1853 designs.
For drawings of ships undergoing lengthening, see
for example: 138-13-12, "Proposed alteration of Shaded image of 1795 desig,n.
U.S. Frigate Sabine...Navy Yard New York, April
26th, 1854"; 79-10-7B, "Proposed alteration of Shaded image of 1853 design.
U.S. Frigate Sabine...Navy Yard New York, April
26th, 1854. B.F. Delano, N.C."; 79-10-7A, A device for measuring keel line while ship was afloat.
"Profile of U.S. Frigate Sabine as a Corvette,
April 26th, 1854"; 107-10-9M, "Plan for A possible scheme for converting 1795 midship frame
lengthening U.S. Frigate Sabine 20 feet. Navy to 1853 frame??
Yard New York, June 15th, 1854. B.F. Delano,
N.C."; 107-11-14E, "U.S. Frigate Santee, Navy
Yard Kittery, ME, May 9th, 1854. I. Hanscome,
N.C.," RG-19, E-126, NARA.

Drawing, "For taking, the shape of ship before

docking," drawn by Charles S. Bruff, undated,
Record Group 45, Naval Records and Library,
entry 405A, "Grice Collection," #87, National
Archives and Records Administration.

359
 spline
Hulliorm
men' fciiing Analysis
digitizer

=N--Aso

VAX

es ,his figure snows oh overvie oi

Huilform Design
computer hardware used to model
Data Base
evidence. Information wcs digitized

in or typed into o oifset file using

the VAX system (1.). The data was

then faired and Analyzed on this

IBM AT/compatide [7i system. Hull form comparisons
CAD data base were performed on the IBM AT/
compatible system (2.).

Detailed Drafting

Modeling

Fig. 9. Compteer hard/soeware used In study

VAX tIFDS PLATFORM

CD VII. rrz 1 ...N. -1
Icrrso ....- --.... CONVERSION , ANts I
AwITILAA 70 RUNES 7 rem..
'''''',...
1

earre.e j
SPCA
Orr sEl i
DAIA Oir
FILE GONvERRS
/Nei r 'GAIN( Snow% A SAP,
70 OAF FRI PORI,. OF 7.f HIJAl FOR.
DES4N SPAM./ (KOS)
SOFTWARE USED P4 DAIS
ANAA AS. ON IRE AAA SYSTEIA
SAGS' ReS Is ALSO TR. FOR 7NE
NADROSTARC RR, AVG...m.1A P1.111011.'
ANALYSIS 051 A Su., PORTION pr
RROGRANI 50,10.01 Asset ABU JS

- CO NERAORA GOwNECIION

IBM AT/COMPAT/BLE PLATFORM

WV-7

woRoIscg,ocA
REPOPA
GENERARON
PROCRAN

REPORT

Fig. 10. HFDS schematic

360
 .
...... ...
.. ...
....,-,.........

"

,g I
. Oa;
6tigii
..

5
i fj!
Ì 5 flit_
ggg
. !1-

I
SHIP HULL CHARACTERM17CS PROGRAM ISHCPI
HYDROST.0 CURVES OF FORM FOR 1795 DEMON

Fig. 12. Hydrostatic curves of form for 1795 hull

... , .

..

4
s

ii 11
iOz.._
,!!.
VI I.
. !..

. - :-..... ...... .
SHIP HULL CHARACTERISTICS PROCRA,SHCPI
HYDROSTATIC CURVES OF 70100 TOR 1803 DESIGN

Fig. 13. Hydrostatic curves of forro for 1853 hull

.. , '

5
lf 't(ta
0 i
I. 1 E

1
a:
........, yil
DATA HELM/ IS SUBJECT TO DUESTION
DUE TO REVERSE CURVATuRE /N mULL
FROM MOGG/NO CONDITION
i

. . -
. . .

sifip ,LL cmARAcTERmT,c PROCRAm RUN FIER :953 suRVET PUL.

FORM or US: CONSTELLATION

Flg. 14. Hydrostatic curves of form for 1853 survey

361
 7 9

2
3 4 5

OVERLAY Of 1705 DESIGN ON 1853 SURVEY

AT SAME LONGITUDINAL LOCATIONS

Flg. 16. Body plan of 1853 survey overlaid on body plan 01 1795

fECTION1 fur 1.7 0.0 rt 07 Or SV1,0

V ATIIirtl TO I./ 31101.411 ItC1101

6 9

IV 'FL

-BASEUNE
CONIP1 'TER IMAGE SHOWING HOC
AND GENERALLY POOR CONDITION
or 1795 111.11. AS SURVEYED IN 1853
1 2 3 4 5

Fig. 15. Computer graptsc image of 1853 survey showing hog and twist

IV. 1t1 C»

:;171tI ?!.;,h1tLiZtVon'
Tias3 "R"T

Flg. 16e. Body plan 01 1853 survey ovedaid os body plan of 1853
 Ak 1- NOTE AS FRAME I1AS [MCC. OvER
1 50 PLUS YEARS IT HAS ALSO
I
MOVE OUTICARD AwAY FROM THE
I CENTER UNE

MIDSHIP SECTION OF 1853 SURVEY

OVERLAID ON 1785 DESIGN

vt.

FRAMES TRANSPOSED TO SA1AE RABBET LINE.

1853 DESIGN
Fig. 17. Curves from 1795 and 1853 survey al same origin and rotalion

1795 BODYPLAN 1853 BODYPLAN OVERLAY

32 FT
1795 DESIGN
28 FT
24 ,7
AU T

16 10
12 FT PERSPECTIVE VIEW OF 2853 DESIGN
(TOP) AND 1795 DESIGN (BOTTOM)
8 FT BOTH AT SAME SCALE AND ANGULAR
VIElf.
4 FT
2,
Fig. O. IsomeIric of both 1795 and 1853 desig.
BODY PLANS OF 1795 AND 1853 DESIGNS
COMPARED AT SAME LONGITUDINAL LOCATIONS
AS MEASURED FROM NUDSIIIPS,

Fig. 18. Body plan of 1883 design compared with body plan 01 1795
 design 1853 to similar 1785 from trame make to addition Possible 28. Fig.

design 1853 of image Shaded 22. Fig.

BAsELINE

W FT

POSSIBLE. TRANsiTIoN
THIS MAKE WOULD
si ,FtRZE,FRo2 mAT1 oF ADDITION IF
1LLINE
OF DETERMINE TO FRAMES
SURVEY. FRoM GEOMETRY (OUTER) 1853 AND
YEAR 56 OF THAT NoT
oLD (INNER) 1795 OF
CEoMETRy, DEsIGN NEw OF SECTIoNS MIDSHIP
COMPARISON A IS THIS NOTE: oF

)COMPARISoN
SHAPE. w NE MAKE TO
Form Hull 1853
scARFS? POSSIBLE S. FRAMEs OF coMBINATIoN 2.

CONSTELLATION U.S.F.

afloat while ship of fine keel measured which System 25. Fig.

PLACEMENT. BLOCK KEEL

FOR KEEL IN HOG MEASURING FOR BEAM
UNDERWATER WITH BARGES SHOWS FIGURE design 011795 =age Shaded 21. Fig.

Form Hull 1795

CONSTELLATION U.S.F.
 A New Systematic Series of Planing Hulls
G. Delgado, G.E. Lacy, C.C. Hsiung
(Technical University of Nova Scotia, Canada)

1.0 INTRODUCTION the more than sixteen useful plots

obtained, four will be shown here.
In the last few decadesthe Velocity is given as beam Froude number
performance of planing hulls has been Cv = v/VgB. Figure 1 shows the
researched by conducting model series such distribution of LIB. It is clear that
as, Series 62 [1]), Series 65 [2], most of the planing hulls have a L/B
Fridsma's rough water series [3], the ratio of 3.0, and that the chosen range
Naval Academy Series [4], and the Dutch of 2.5 < L/B < 3.5 covers the peak of
Series [5]. However, the geometry of the distribution. Figure 2 shows the
planing hulls has changed considerably distribution of deadrise angle. It can
from that tested in the above series, and be seen that the distribution is centred
consequently there is a need to provide around 18° and that more than 90% of the
designers and researchers with new designs considered are within the range
performance data based on modern hull of 12° to 24°. From this analysis, a
shapes. The Naval Architecture Program of decision was made to build 9 models
the Technical University of Nova Scotia, with L/B ratios of 2.5, 3.0, and 3.5 and
in conjunction with the University's with deadrise angles of 12, 18 and 24°.
Centre for Marine Vessel Design and To select the shape parameters for
Research (CMVDR) is completing a program the parent hull, several lines plans
to design and test a systematic series of were studied to determine the trends in
hulls representative of the modern-day form. For many reasons hull designers
planing craft. were not willing to supply us with their
design parameters, but nonetheless, a
2.0 DESIGN OF THE PARENT HULL AND SERIES total of 46 hulls were measured and
entered into another data base from
Generally, the earlier planing hull which the shape parameters were
series tests do not accurately represent obtained. Some of these parameters are
the modern planing hull because the given in Table 1, below.
length-beam ratio is too high, the The towing tank at the CMVDR
deadrise angle is not varied measures 30m long, lm wide, and lm deep.
systematically, and some series use A beam of 230mm for all models was
dated, or prismatic forms. To define a chosen as being compatible with the tank
typical or parent hull for the new series,
contact was made with designers and boat
builders from across North America and
Europe to obtain data on modern hull
shapes. The type of boats considered are: Parameter %LOA
pilot boats, rescue craft, patrol boats,
recreational craft, and racing boats. The Chine Length 91.2
length range is from 15 to 85 feet,the
displacement range from 1,000 to 150,000 Lth, const Deadrise 41.0
lbs, and the cruising speed range from 14
to 70 mph. Lth, Const Beam on Deck 51.2
The information supplied by
these boat builders, together with Bow Length 37.0
information obtained from journals and
publications was entered into a data base Depth, Chine at Bow 10.7
which includes more than 300 hull shapes
from 83 sources. The data were plotted
and analyzed, and from this analysis non- Table 1
dimensional coefficients were chosen. Of

365
 reater than this resulted in an
specifications. Figure 3 gives the body
unrealistic trim by the bow. The
plan for the parent hull. The design was
incorporates developable surfaces to ease vertical centre of gravity (VCG)
the construction of the models. A chine
maintained at the tow point (30% beam)
spray rail was incorporated in the hull within -2%. This height of the VCG
shape as this is typical for most modern appears to be realistic and was the same
planing hulls. No attempt was made to as used by Fridsma [3]. The actual value
incorporate additional lifting strakes - of VCG was measured and recorded for
this in itself being material for a each condition tested.
separate research project. Based on this The towing was performed with two
hull shape, the entire series was different thrust lines:
obtained. For the variation of L/B ratio,
(L/B=3.0, P=18°) was Thrust line passing through
the parent hull the centre of gravity
enlarged and reduced in the longitudinal Thrust line parallel to, and
direction using a computer program. The
variation of deadrise angle was obtained lying on, the keel or base
the keel or line.
by raising and lowering
centre line of the parent form. In
modifying the parent hull in this manner, The thrust line (ii) was chosen as
the main non-dimentional parameters a reasonable compromise to the many
possible configurations of single, twin,
remained unchanged.
triple screw, and jet or sterndrive
propulsion. It is not intended to
3.0 EXPERIMENTAL DESIGN.
The displacement coefficient was represent conventional shaft drive.
chosen as Cd = A/(p*B"3), (where 'p' is With all of the above parameters,
water density and 'B' is beam). Figure 4, the total number of runs will exceed
which represents data from our planing 2500 before this series is complete.
Turbulence was induced in all the
hull data base, was used for the selection
of Cd for each L/B ratio. This information models by means of Rama [6] strips
selected after testing two
is reduced to tabular form in Table 2 of turbulence
below. configurations
stimulators, and observing the results.
The parent hull was tested at LCG=75%
aft, and using Cd=0.1,0.15, and 0.20. In
L/B C, the first test, Rama strips were placed
at station 4 and station 7. This
2.5 .05 .10 .15 .20 configuration showed an increased total
resistance at all speeds. The final
3.0 .10 .15 .20 .25 choice was to install a single Rama
.30
strip over station No. 4, since this
3.5 .10 .15 .20 .25 an increased
configuration produced
resistance at low and medium speeds
(hump speed), but no change at high
Reynolds numbers - in keeping with what
was expected.
Table 2 Shallow water effect was
investigated, and for the Depth-Froude
-Number at which the tests were
performed, this effect is almost
The Towing Tank at the CMVDR has
negligible (<5% increase in wave making
been recently upgraded and carriage speeds
resistance) [7].
in the order of 4.0 m/s can now be In addition, underwater
obtained. This has allowed us to tow the photographs were taken for each run to
series at Cv values up to 2.75 which is determine wetted length of the keel and
well into the full planing regime, and
chine and to compute wetted surface area
covers most of the vessels contained in as well as spray zone. These photos were
the data base (see Fig 5), except for taken using a 35mm underwater camera and
those with high L/B which tend to be black and white film, and cover more
racing boats. than 80% of the runs.
The longitudinal position of the Every run was video-taped above-
centre of gravity (LCG) was chosen for water using a carriage mounted camera,
each displacement within the range 25% to
forward from the to provide a visual record to facilitate
35% of LOA measured
transom. This range appears to cover that data reduction.
ensure repeatability, -50
typically used today (Almeter [9]). Values To

366
 conditions of speed, displacement and LCG in a paper of this length the complete
were repeated at least once on the parent results of the work. This will be
hull. The results showed that the presented as a complete series in the
variation of force, sinkage and trim from near future. Some of the results
run to run is in the order of 1%. obtained are discussed below.
A typical page of the plotted
4.0 TANK CALIBRATION results from this series are as shown in
Figure's 12,13,14, and 15. Each of these
To ensure the reliability of our pages of graphs contain the results
results, it was decided to first test a obtained for a model with a single
hull previously tested by others. To avoid displacement and centre of gravity and
the additional uncertainty introduced by two thrust lines. Error bars have been
testing a geosim hull, it was desirable to included on the graphs, which correspond
test a hull of identical dimension to one to the standard deviation of the
with published data. For this reason, the averaged signal. The condition shown is
Fridsma L/B=4 prismatic hull was selected. for L/B=3.5, p=18,C,=0.15, LCG = 70%
The hull has a deadrise angle of 200, a aft. The model resistance is plotted as
L/B ratio of 4.0, and a length of 36", R/A and also as C, versus C,. Sinkage is
which is suitable for the CMVDR tank. For non-dimensionalized by division by beam.
the 8 pound displacement, LCG's of 60, 65, Trim is given as a relative value. The
and 70% LOA aft were tested, and for 16 static trim angle was also recorded and
pound displacement the LCG positions of will be included with the tabular data.
55, 60, and 65% LOA aft were tested. Each Figure 15 shows wetted surface area
condition was tested at 12 speeds which versus C,. This has been non-
correspond to V/VL's ranging from 1 to 4. dimensionalized by division by the
These tests were also performed a second square of the beam.
time to insure repeatability. In looking at these plots, it is
These results showed quite close interesting to note that the resistance
agreement with the Davidson Laboratory. and trim lines with thrust line
Figure's 6 to 11 show the 8 and 16 pound correction cross the uncorrected lines.
displacement and 65% aft-LCG results. The For resistance this occurs at a C, of
error bars represent the standard about 1.5, and for trim this occurs at a
deviation of the averaged data. The solid C, of 2.25. This is typical of all the
circles correspond to the data points as results collected so far. Figure 16
marked in Fridsma's paper. The solid gives a plot of the resistance curves
squares represent data from B.C. for the parent hull with LCG = 70% aft.
Research's Ocean Engineering Centre (OEC) The 4 curves represent the 4
collected on a somewhat larger scale displacements tested. It can be seen
Prismatic hull of Fridsma form [8]. that all the curves are of the same
Figures 6, 8 and 10 show the 16 pound form or family. This result is typical
displacement condition. Figure 6 shows a of all the data collected.
very close agreement between the CMVDR In reviewing the video tape of the
data and Fridsma. Figure 8 shows CMVDR's 18 degree deadrise data runs, it was
sinkage prediction to be slightly above observed that the transom always
Davidson Laboratory but below OEC, and ventilates at a C, of 1.0. This is a
Figure 10 shows CMVDR to again be in close somewhat value than the 0.5 given by
agreement with Fridsma. Figures 7,9, and Savitsky and Brown [10].
11 show the 8 pound condition at 65% aft As another check on our results, a
LCG. From these graphs, it can be seen comparison was made with Savitsky's well
that the agreement is also quite close, known procedure [11]. This comparison
the resistance given in Figure 7 being was made for the case of the thrustline
within 6% of Fridsma's graphical data, passing through the centre of gravity.
while the sinkage and trim agree very Several conditions were tested, one of
closely with the Davidson Laboratory which is shown in Figure 17. This was
results. model L/B = 3.5, p = 18°, LCG = 70% aft,
It is the intention of the authors and C, = 0.10. Close agreement is
to publish a thorough account of the tank observed.
calibration at the CMVDR, since there is Porpoising was also observed and
much more detail that cannot be included recorded using a video camera. So far
in a paper of this length. this dynamic instability has appeared
only in the two cases of L/B=2.5,
5.0 SERIES RESULTS LCG=75% aft,p=18', and C,=0.15, as well
as L/B=3.0, LCG=75% aft, p=24., and
It is obviously impossible to show C,=0.20. In other words, in the case of

367
heavily loaded vessels with an LCG far Hull Forms With 25 Deadrise Angle",
aft. It is expected that other cases of I.S.P., Vol. 29, 1982.
porpoising will be observed once all the
conditions are tested. Furthermore, it is Hama, F.R., "An efficient Tripping
expected that some of the other conditions Device", Journal of Aeronautical
would exhibit this phenomenon if higher Sciences., March 1957.
speeds could have been attained. A t
this moment approximately 70% of the runs Sturzel, G., Graff, w.,
have been perform2d, and it is expected "Systematic Investigation on Hullforms
that by later in the year the complete for Small Craft in Shallow Water in the
series, together with a method to predict Subcritical and Supercritical Speed
the full scale performance will be Ranges", B.S.R.A, Translation No. 2046.
presented.
Opel, H.S., "Comparison of Calm
Water Experiments on a Simplified
LIST OF SYMBOLS Institute for
Planing Hull", N.R.C.,
C, = beam Froude number = v/VgB' Marine Dynamics Report LM-1989-22.
C, = displacement coefficient = A/(pB))
Almeter, J.M., "Resistance
LOA = length overall Prediction of Planing Hulls. State of
LCG = longitudinal centre of gravity
VCG = vertical center of gravity the A,t", SNAME, Hampton Rds. Section,
L/B = length to beam ratio Febuary 1991.
A = displacement Brown, P.W.,
Savitsky, D.,
p = water density
= beam "Procedures for Hydrodynamic Evaluation
B
L = length of Planing Hulls in Smooth and Rough
p = deadrise angle Water", Marine Technology, Vol, 13.,
Oct. 1976.
ACKNOWLEDGEMENTS Savitsky, D., "Hydrodynamic Design
of Planing Hulls", SNAME, Marine
The authors wish to acknowledge the
support received by Mr. T. Edwards from Technology, 1964.
the Center for the Marine Vessel Design
and Research, Dr. A. Bell from the
Mechanical Department of the Technical
University of Nova Scotia, R. Hetaria for
the construction of the models, R. Hawary Naval
and to Research Assistant.
for assisting on the tests, Architecture Program at the Technical
N.S.E.R.C., A.C.O.A., and N.S.D.E.D. for University of Nova Scotia.
financial support.
Research Naval Architect. Center for
REFERENCES Marine Vessel Design and Research.
Clement, E.P., Blount, D.L.,
"Resistance tests of a systematic series [2] Head of the Naval Architecture
hull forms", SNAME Program and Director of the Center for
of planing Design and Research.
Transactions, 1963. Marine Vessel
Technical University of Nova Scotia.
Holling, H.D., Hubble, E.N., "Model
Resistance Data of Series 65 Hull forms",
NSRDC. Report 4121 May 1974.
Fridsma, G., "A Systematic Study of
the Rough Water Performance of Planing
Boats", D.T.S.I.T Report R75., November
1969.

Compton, R.H., "Resistance of a

systematic series of semi-planing transom
stern hulls", SNAME, Marine TEchnology.,
October 1986.

Kuening, J.A., Gerritsma, J.,

"Resistance Tests of a Series of Planing

368
 100 80 1 1 I I I I I I

80
60

60

40
o.
40

20
20

o
20
7I
2.5
1
1 _I ,I, 1 1-11 1 1 1 1 1 1 1 1 1 ,
3.0 3.5 4.0 4.5 5.0 5.5 60 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Fig 1
L/13 Fig 2 Deodrise Angle

INNWE
MENEM
MENEM
=NW'
Fig 3

0.4 6.0 1 1 1 1 1

5.5 o
5.0 oo
0 00 00
0.3 0
4.5
00
000 o 00
0 0 00
4.0 0
o 00
3.5 °CAP ° o 0 oO
0.2 3.0 830 g (13'
2.5
2.0 93
0.1 1.5
1.0
0.5
0.0 0.0 I

2 3 4 5 6 1 o 9.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 60

Fig 4 Fig 5
L/B L/B

369
0.30 0.200 1

Tuns Data. Run 1

0.175 ThrisData.ThirÇ2-,-""*"
0.25 Fridsme Data
0.150
41,

0.20
0.125
: .
a
E
0.15 0.100
I o Tuns Data, Run 1 i5
a Tuns Data, Run 2
Fridsma Data; 0.075
OEC Data !

0.10
!
0.050
1

0.05
0.025

0.00 0.000
00 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 50
00 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Fig 6 Fig 7 V/sqrt(L)

V/sqrt(L)

0.1 0.1

0.0 a I 3c

0.0

-0.1
Tuns Data. Run 1
Ia Tuns Data, Run 1 a Tuns Data. Run 2
Tuns Data, Run 2 -FricIsma Data
1
Fridsma Data 0.1
OEC Data
-0.2

-0.3 0.2 4 5
4 5 o 1 2 3
o 1 2 3

Fig 8 Fig 9 V/sqrt(L)

V/sqrt(L)

2
4 " o Tuna Data, Run 1
o Tuns Data, Run 1 =, Tuns Data,
Fridsma Data
Fridsma Data
OEC Data 1

o
7

2 3 4 5
4 5 o 1

1 2 3
Fig 11 V/sqrt(L)
Fig 10 V/sqrt(L)

370
 0.30 2.5 0.01
n....No..Th rust. :Corm p-.. . ... ........

v Thrust Comp. I : Q No ThruStline Comp.

0.25 -t
0.00 Thrust tine-Corn.
2.0
1

0.20
-0.01 '

0.15 1.5 o
0.02
0.10
1.0
0.03
0.05
_

0.00 i
0.5 -0.04
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Fig 12 Fig 13
Cv Cv

4.0 3.00

3.5 No Thrusline Comp.

2.75 Thrustrine Comp
3.0
(1)
2.50

-
g.)
2.5

2.0
..
:
..

E 2.25
-i----- ..- cn
1.5 p
o .....;._ : i. , . 2.00
' o I. No Thrustliné CoMpensation o o
1.0 .:.

Thrustline Compensation
1.75
0.5

0.0 1.50
00 0.5 1.0 1.5 2.0 2.5 3.0 00 0.5 1.0 1.5 2.0 2.5 30
Fig 14 Cv Fig 15 Cv

0.30 500

450
0.25
400

0.20 350
-1-
300

0.15 250
o
<7,
200
0.10 Q)
o CWOR Dato
- Cd = 0.10 150
Sovitskj, Predittion
Cd - 0.15 a
Cd = 0_20 100
0.05
Cd = 0.25 50

0.00 o I

0.0 0.5 1.0 1.5 2.0 2.5 3.0 o 1 2 3 4

Fig 16 Cv Fig 17 Cv

371
 Mathematical Simulation of Hovercraft Maneuvering
G. Zilman, T. Miloh (University of Tel-Aviv, Israel)

Abstract In common practice, the longitudinal component of

these forces is called "wave resistance". It's evaluation is
A mathematical rnodel of a inaneuvering amphibious air-
presented in many papers, e.g. [14], [1], [13], [8], [5], but
cushion vehicle (A CV) is presented. The ACV, studied is the consequent analysis of other components, such as the
of plenum-chamber type with constant pressure applied to
lateral force and yawing moment as well as their signif-
the free-surface of the fluid. The main efforts of this paper
icance for hovercraft maneuvering, is to our best knowl-
are concentrated in clarifying the role and the significance
edge, far from being understood. Therefore we intend to
of the hydrodynamical forces exerted by a moving pressure discuss herein some pertinent such as:
distribution. The general method of approach is illustrat-
ed by means of solving the titled problem for a schenzatic The memory effect, imposed by the free-surface of
heavy fluid;
air-cushion planforrn close to the ACV SR.N2.
The explicit solutions for the longitudinal force, later-
al force and yawing moment, obtained within the frame-
INTRODUCTION work of the theory of surface waves;
Despite of many years of theoretical study and abundance The dependence of the longitudinal force, lateral force
and yawing moment on the drift angle and of a hovercraft;
of volumes of experimental data, the analytical descrip-
tion of hovercraft maneuvering still meets formidable d- The influence of the water depth on the above menti-
ifficulties. One of the remarkable features of an (ACV) oned reactions;
is that hydro-and aerodynamical forces, involved in its A comparison between the numerical computations
dynamics, can hardly be separated in a rigorous manner. against available experimental data;
Especially it refers to the hydrodynamical interactions The mathematical model of hovercraft maneuvering
between the skirts, air jets and the free-surface. How- which embraces aerodynamical forces acting on both in-
ever, without a clear mathematical model of a craft any ner and outer surfaces;
experiment with constrained models in a towing tank can The results of the mathematical simulation of the hov-
not be sufficiently useful since in such a case even the ercraft maneuvering.
conjectured range of parameters of the motion is not well
known aprio ri.
The resulting vector of forces acting on the hovercraft
THE ACV EQUATION OF MOTION
can be decomposed into four components, i.e., aerody- A hovercraft is an aggregate of rigid and elastic bodies
namical forces, acting on the hull of the hovercraft, air jet which are subjected to complex relative deflections. N-
forces, forces exerted by wave making on the free-surface evertheless, it can be usually presumed that the inertial
and "residual" forces. Evidently, such an approach is characteristics of the hovercraft do not depend signifi-
very similar to the Froude hypothesis, but it should be cantly on the fluctuation of its geometry. In this case the
stressed that each of the foregoing components can not maneuvering equations of a hovercraft may be considered
be estimated with the same accuracy as the frictional re- as equations of a rigid body and its inertial characteristics
sistance of a conventional ship. can be evaluated at the steady state equilibrium position.
The above-mentioned separation can be considered by Three orthogonal systems of coordinates are used here,
invoking the momentum conservation principle with re- i.e., fixed in space, OG xG yG zG fixed in the body
spect to the air volume, bounded by all rigid inner sur- with the origin OG which coincides with the center of
faces of the hovercraft and flexible skirts. gravity of the vehicle, and a moving local system Oxyz
One of the most significant components, acting on with (x,z) lying in the plane of the undisturbed free-
the inner surface of a hovercraft, is imposed by the free- surface, where Oy pointing upwards and Ox in the di-
surface elevation which is consequently induced by the rection of the vector V, ( of the speed of the air cushion
moving pressure distribution. pressure distribution).

373
 Generally speaking, the ACV moves with six degrees where V .0/1 + V and ß = tani(Vz/Vx) denotes
of freedom, but for the analysis of the horizontal motion the drift angle.
of the ship only four of them may be incorporated. These
degrees of freedom are the state variables of the following AERODYNAMICAL FORCES
equations of motion: The vector F and the moment M of the aerodynamical
forces acting on a maneuvering hovercraft can be condi-
M (1.7x + VzSly) = X(i) tionally decomposed into two main components
M(Vz VxS2y) Z(i) (1)
xAx IxyQy = Mx(t) F = Font +Fin (2)
.13,3421, /xy1:2x = My (t)
M = M.ut +Min (3)
where M is the mass of the ship; Vx, Vz, Oy and S2x
Here the subscripts "in" and "out" refer to the so-
are the forward, transverse, yaw and roll components of called inner-surface Si and outer-surface Soot of the hov-
velocities, respectively; Vx, V2, 2y and 2x are the clarifi-
derivatives with respect to time t of corresponding com- ercraft. This definition requires some additional
ponents; X (t), Z (t), My (t) and M x (t) are the forward cation. with the
The air volume inside the camber(s) interacts
force, transverse force, yaw and roll moment, respective-
ly; /Yy, IXx and Ixy are the yaw, roll and product outer air flow through the inlet orifice of the fan system
moments of inertia of the ship with respect to the Ox and the air gap under the skirt. The planes of inlet ori-
fices and the baseplane 1 of the craft separate the above
and Oy axes. pointed inner and outer surfaces. With some sacrifice in
The orientation of the ship axes may be derived by rigour, one can assume that when calculating the aerody-
its successive rotation through the yaw angle ,rp and the namical forces it is permissible to ignore the mutual in-
roll angle 0, respectively, defined around the axes in their fluence between the inner and outer disturbances of the
progressively changed position. air flow. Simply speaking, one may calculate the force
Note that in equations (1) we do not take into accoun- Fout by assuming that the fan system does not operate
t the trim of the ship and its daylight clearance under and, in turn, calculate the force Fin by assuming that
the skirt. If there exists an intensive interaction between the disturbances of the outer air flow do not exert any
flexible skirts and the free-surface exists, then these two considerable influence on the flow inside the air chamber.
kinematical parameters have a significant bearing upon
the overall behavior of the vehicle. Aerodynamical Forces Acting on Outer Surfaces
The appearance of a large trim in conjunction with
a large transverse velocity is relate to the potential dan- The forces acting on the outer surfaces of an ACV are
ger of a hovercraft rollover. Herein, we consider such a beyond the scope the present study and, thus, we will
situation as a special anomalous case of navigation, s- discuss them only in broad terms. As a general remark,
ince during the operation, no parts of the hovercraft are we note that the study of these forces can be carried out
considerably immersed in the water . by employing scaling models of ACV and aerodynamical
Although the possibility of predicting the vehicle roll- tubes.
over is a problem of great interest, naturally, the predic- Two general approaches can be used for the above
tion of hovercraft maneuvering under normal navigation- mentioned study. The first of them (e.g. [18]) relates to
al conditions with relatively small trim and roll angles experiments with ACV models supplied by movable and
is of major importance. Moreover, a host of numerical immovable control surfaces. The compoments of the re-
experiments with corroborative observations of full scale sulting aerodynamical vector, i.e., the forward force XA,
trial results allows us to presume, that equations (1) can transverse force ZA and yawing moment MAy, can all
be used not only for the evaluation of maneuvering per- be expressed as functions of nondimensional kinematical
formances, but also for predicting the initial moment of parameters of motion, e.g.
drastically increasing the angles of roll and drift.
Thus, under the present consideration, the kinemati- XA = QSCXA(3 9)45i)
ZA = qSCzA(0,wr, , 9, bi)
cal relations defining the coordinates of the center of grav- 9, 6)
M(X,Y)A = qSLCm(x,y)A (0, wx, ,

ity G, (G and the orientation of the ship in the inertial

coordinate system can be written as follows: where q 0.5poV2, pa denotes the water density, L is the
length of the ship and S is some reference area, say, the
ox; = V COO
'The plane which contains the lower contour of skirts
Qy = V sin(0

374
 projection of Sout on the ACV centerplane.The nondi- escape flow by FE. Then, the resultant vector of the air
mensional angular velocities are defined as follows: pressure applied to the craft surface Sin can be written
= OxLIV,wy = L/V and bi is the angle of the as follows
controls deviation.
Such experiments are rather expensive and are appli- F5 FE + FE (5)
cable only to a particular hull/control surfaces/air pro-
where
pellers combinations. Thus, usually it is difficult to ex-
amine the effect of different control surface combinations FE = pa JQE VEn, do (6)
on ACV maneuverability.
The second approach is based to the so-called "modu-
lar" mathematical model which enable separates evalua- FE = Pc nE dE (7)
tions for different lifting configuration. 2 Since a vertical
Here Pc is the air-cushion pressure and nE is the unit
stabilizer with a rudder can be considered as a wing with
vector along the interior normal to the dE.
an aileron, the lift FR of such a configuration can be es-
Next, let us proceed with the derivation of the formula
timated invoking the lifting surface theory as
for the moment ME acting on the inner surface. For this
purpose we shall use the moment-of-momentum conser-
FyR = qSc(k14 RO k2Cit wz +k3C;',"Rwy +k4CT Rb)
vation principle. By a reasoning similar to the preceding
where SR denotes the area of the lifting configuration and one, we may write the following relation:
factors ki are some functions depending on the propeller
loading. Ms, = ME +M (8)
where
Aerodynamical Forces Acting on Inner Surfaces
As long as the outer aerodynamics of the ACV is speci- ME = Pa f Q E VE (r x nc,) do- (9)
fied, we can consider the aerodynamical forces acting on
its inner surfaces. The air, which enters into the chamber, ME = Pc (r x nE) dE (10)
produces the resulting force given by:
where re and rE denote the radius vectors measured from
F1 = PaQ 11 the origin of the attached system of coordinates to the
where Qi is the inlet flow through the fan orifice and surface elements do- and dE, respectively.
V/ is the velocity of the correspondent center of the fan
orifice. Usually, this component is small in comparison Aerodynamical Forces of the Jet Efflux
with other componens of aerodynamical forces and its Several theories relating to the escape flow can be used,
evaluation does not represent any additional difficulties.such as the theory of jet curtain or the plenum chamber
Let us now examine the air volume W bounded by theory. For the sake of brevity herein we consider only
inner-surfaces Sir, and the free-surface E . The inletflowthe latter one. It should be stressed that the following
enters through fan orifices and escapes through the air
analyzes for the plenum chamber theory differs from the
gap under the skirts, i.e., through the surface u. Dur- jet curtain examination only by inessential details.
ing the time interval 6t this volume is displaced and the
According to the Bernoulli equation, the escape flow
change in its momentum KE will be can be expressed as
dK E paót Q E VE nu dcr (4)
QE = CEhN/2Pc / pa (11)
where Q E denotes the escape flow, VE is the velocity
where CE denotes the escape coefficient, h is the daylight
of the jet efflux through the surface do- in the direction clearance under the skirt and the cushion pressure Pc is
along the exterior normal n, to do-. According to the
assumed to be uniform within the whole cushion volume.
momentum concervation principle, dKE is equal to the
Strictly speaking, the escape coefficient depends not
impulse during time 6t of all external forces acting on
only on the inclination of the skirt fingers 0, but also on
the volume W, i.e. of all mass and aerodynamical forces
the type of the subsurface E, its form and the angle of
applied to the volume W through the surfaces Sin and
inclinations of air jets 7 with respect to the horizontal
S. Let us designate the resultant vectors caused by the plane. However, for most practical purposes, accuracy
'Naturally, the interaction between different parts of the total it may be estimated by assuming that the surface E is
lifting configuration is substantially important rigid and the angle 7 is the same along the contour of
375
the cushion z, = bc(x). Just to give an example, note HYDRODYNAMICAL FORCES 3
that for a conventional design 0 = 450, CE 0.567 and The Velocity Potential
cos -yrz-z 0.6.
by
However, the dependence of escape coefficient QE on The ACV makes its presence felt on the free-surface
by the air cush-
the daylight clearance and the free-surface elevations is means of the pressure distribution created
caused
more significant. This dependence can be expressed as: ion. Usually, typical values of the water depression
by the cushion are low in comparison with the, say, length
(12) of the craft. That is why it is a good approximation to
h = ho 19,(x)O 71,
linearize the hydrodynarnic problem. Under the usual as-
sumptions of linearized potential flow [16], the potential,
where 17.0 denotes the daylight clearance in equilibrium 77, (, t) satisfies the Laplace equation
position and Thu is the wave elevation measured from the
undisturbed free-surface plane. A(I) = 0 (19)
The substitution of (11) and (12) in (6) and (9) yields
two compoments of the outlet, i.e, the transverse force and the free-surface condition
FEZ and the yawing moment MEy , which after some =0 (20)
simplification can be written as follows g4), =.-_ (Pc)gl p on
where p is the water density. On the flat rigid sea bottom
we have
FEZ = 2g MCE (0 cos -y (13)

(1), = 0 on 71 = H (21)

where H denotes the constant depth of the fluid.

871, amp
)dS, (14) In order to guarantee the uniqueness of the consid-
MEy = 2PcCE cos 7
Sc ôzOX ered boundary-value problem the appropriate initial con-
ditions have to be set up as follows:
where g is the acceleration of gravity, S, denotes the air
cushion area and
4)° cip(e, 77, c, o) = C, 0)

= Pc(,71,0) pc = 71, o
aqw
Using the Laplace integral transform in the time domain
(15)
SCJ5 az and two-dimensional Fourier transform with respect to
C), one can obtain a corresponding initial-value prob-
For ship maneuvering studies the values only of three lem for the transforms.
components of the vectors F, and 14,- are essential, i.e., The solution of this initial value problem and the sub-
sequent inversion of the transforms render the explicit for-
=
s c
Pc
a.
dxdy m for the potential as a multidimensional integral with
the integrand depending on time. We do not represent
this explicit solution for two reasons: i) for the sake of
brevity; ii) since from the practical point of view and for
intended purpose it is as complicated as useless. Why is
FEZ = Pc-8,71w drdy
sc az it so ?
The kinematical parameters of motion are the argu-
ments of the above-mention integrand and, consequently,
077, aqw
z)dxdy the hydrodynamical forces will be represented as a mul-
/VEY = Pc(x az ax tidimensional integral operator acting on the nonlinear
function of the unknown arguments Vx, Vz, Qy and 7,b.
As we can see now, the aerodynamical reactions acting 3 First of all some considerations concerning the label of this sec-
on the inner surfaces of the ACV depend on the free- tion should be given . Usually, the aerodynamical reactions (16)-
Indeed, the influence of
surface elevation under it. Moreover, the same derivatives (18) are considered as hydrodynamical. mentioned components is
the free-surface elevation on the above
of 77, arise in components of (13)-(14) and (16)-(18). felt through the deforrnation of the water free-surface and this de-
Thus, the next problem, which should be examined, forrnation can be examined by means of employing the theory of
define them as
concerns the physical nature of the free-surface deforma- surface waves. That is the main reason why we
hydrodynamical.
tions and the appropriate methods for evaluation it.
376
The substitution of these operators in the equations of Herein we deal with a particular form of pressure dis-
motion leads to such a system of integro-differential e- tribution, which posses some advantages :i) in correspon-
quations that even its strict mathematical definition rep- dence with the physical meaning the pressure is a finite
resents some difficulties. Needless to say, that its nu- function ; ii) the selected form of the pressure distribution
merical solution can be very difficult. However, we can allows to obtain a comparatively simple explicit solution
not ignore the fact that the time dependent solution of of the problem. We define the air cushion pressure in the
the initial value problem (19)-(21) reflects the influence following form
of memory effects caused by the nonstationary motion of
the fluid. Certainly, if the memory effect is presented, it
has to be established whether it is essential for practical Pc= { Po (1 f2r if f2 < 1
purposes or no. otherwise
where f2 = (x/a)2 + (z/b)2, Po is a constant quantity,
The Memory Effect a,b are the semi-axes of an ellipse and a > 0. The "block
coefficient" of the pressure distribution can be defined by
The set of unsteady ACV trajectories, roughly speaking, the formula
is random and infinite. Thus, in order to examine the
memory effect, it is necessary to select such a maneuver
which can be sufficiently representative for intended pur- irabPo s°= 1 + a
Pc (x,y)dxdy
pose. Herein, among a variety of possibilities, we consider
Usually cp ranges from 0.85 to 0.99 in correspondence
the rectilinear accelerated ACV motion with zero initial with the type of skirts, clearance, etc. For the ACV mo-
conditions.
tion with acceleration e some of the inner integrals (22)
In this case the wave resistance formula can be written
can be expressed in explicit form which yields the fol-
as follows [5],[16], [10], [21]
lowing expression for the coefficient of wave resistance
Cx = FEx(pg)2/2aII
FEX = 472pg fo k2adk 1 cos OdO (22) Do 7r/2
Cx = 4a -2F2 (1 + a)[1 k6 \/-4 dk i D2 E cos Oa (24)
Isin a (t r)A(k,O,t,r)dr (23) O -w/2
x
J
o roe 7r/2 k2
1 k2 cos(k.fq- T) dk j sin OD2 sin(-T2 cos 0)c10]
+41.2pg[k2 cos o-tdk A(k, 0 ,t, r) cos OdO 13 7r/2
where
where
D = Ji+,(Z)/ Z1+'
A(k,0 ,t,r) s!c(,()ck dq Pc(ei,(1)sinw deid(i Z= 0.5k2N/cos2 0 + r2sin2 0
E = (r2/2k)OrAT cos 9) EL, (-1Y+1 ai
w = (" 6) cos + (C (i)sin0 + Vc(t) V(r) and cti = cos ml[C(n)C(m)] sin niF[S(n) S(mi)]
a = -Vgk tanh kH q)/
m1,2 = (kZ7' cos 0 n q/2.,/77,;7
Doctors and Sharma calculated the wave resistance q = tanh k2H; T = t1;7-2; H = H/2a; = elg; r =
for a specific type of pressure path which exponentially bIa
decays at infinity. Obviously, the rate of the exponent de- Here F(x) represents the Gamma-function, Ji+er(x)
cay depends on some parameters which can be selected denotes the Bessel function, and C(x), S(x) denote Fres-
in order to provide the best agreement between theoret- nels integrals,
ical calculations and experimental data. Doctors [2], [3]
2 2
gave also the numerical evidence that the wave resistance C(x) = vrr fo cos(x2) di; S(x) = sin(x2)dt
is not critical to small changes of the values of selected Jo
parameters, but, unfortunately, we found that the trans- The steady motion can be considered as a limiting case
verse force is subjected to considerable fluctuations with t oo of the unsteady rectilinear travelling pressure dis-
respect to the longitudinal and especially to the trans- tribution under the assumption that the speed Ve(t) is
verse smoothing of the pressure distribution [19]. defined as a Heaviside function.
Under such circumstances the problem can be said to Under the conditions H oo and e = 0 this asymp-
be nonstable with respect to small changes of the initial totical steady state solution renders the following expres-
data and, thus, it can be hardly employed in practice. sion for the coefficient Cx

377
 with (24) are plotted in Fig.2.

Cx

oc J174.0[0.5Fn-2 \/(1 + t2)(1 + r2t2)

Cx = K (a) dt
+ r2t2)1+*(1 + t2)°-05
(25)
where K(a) = 42ar2F2(1+a)Fn4a-2 and Fn =17,11/Tiri
The calculated values of Cx for a steady rectilinear ACV
motion are shown in Fig.1

cx
1

O 0.5 1.0 Fn

Fig.2 The coefRcient of wave resistance in unsteady

10 motion (1 - = 0; 2 - 7 = 0.025; 3 = 0.05; 4==
0.075)
It is seen, that the memory effect can cause drastic
changes in the magnitudes of wave resistance coefficients,
especially in the range of Froude numbers larger the 0.4.
0.5 Nevertheless, if Fn > 0.5 (the most important and inter-
e_sting case for ACV maneuvering) one can assume that
for most practical purposes the assessment of ACV ma-
neuvering can be carried out by invoking the quasi-steady
hypothesis.
0 0.5 1.0 Fn Numerous model tests with a rotating arm [15] give
evidence that the considering hydrodynamical compo-
Fig.1 The coefficient of wave resistance in steady nents weakly depend on the angular velocity .ly. This
motion (7 = 0) experimental fact is of g-reat importance, since it allows to
reduce a complicated hydrodynamical problem to a com-
paratively simple one, i.e, to the problem of evaluation
the hydrodynamical forces exerted by a pressure distri-
It is seen that for typical values of a which have the bution moving rectilinearly and steadily with a constant
physical meaning the wave resistance coefficient does not angle of drift.
changes significantly and all calculations may be imple-
mented for any appropriate a in the interval 0 < a < Tran.sverse Force and Yawing Moment
0.025. Hence, we may calculate the ACV wave resis- For the steady rectilinear motion of a pressure distribu-
tance in unsteady motion by using a reasonable quantity co, a = O (c, = 1) direct
tion with parameters H
a = 0.025. 4 The results of calculations in accordance calculations yield three components of hydrodynamical
forces as follow

rr2 /7/2cos(0 + 0) AJ?(E)2 (26)

Cx = 2F n4 cos2 OdO
-7/2

77'2 1.7/2 J?(E) (27)

Cz = sin(0 + 0)
4If 0 the numerical calculations for accelerated motion en- 2Fn4 7/2 A2 cos2
counter formidable difficulties relate to the bad convergence of the
inner integrals (24).
378
 7-2(1 r 2) f7r/2 sin 2(0 + 0) shown in Fig.3-Fig.5
Cmy = (28)
7r j_,/2 A3 cos2 0

0 10 20 30 40 50 60 R,deg
oo
tIc" 1(A1)12(At)
x
Jo 4Fn4t2 + sec4dO
\1111,\

20
7rr2(1 r2) T/2
sin 2(0 + 0)
8Fn2 --x/2

30
Cz T

dB
x Ii(E)J2(E)
A3 cos2

Fif.3 Transverse force coefficient (b/a=0.4)

where
,deg
10 20 30
FE(x ,z)(Pg)2
Cx,z =
(p2 P)

-01

CM y = MY(p9)2
(2pa 17,2 It2) -0.2

-0.3
A = Vcos2(0 + 0) + r2 sin(0 + 0) cmy

Fif.4 Yawing moment coefficient (b/a=0.4)

E A sec2 0/(2F7z2)

.11,2(x) is the Besse] function of first kind,

12(x) and Ki(x) are the modified Bessel function Fig.5 which displays the results of experiments with
and NI(x) denotes the Neumann function. the ACV model SR.N2 in a towing tank (rotating arm)
[15], where the solid lines represent the pure hydrodynam-
Some typical results of correspondent calculations are ical components and the dash lines represent the sum of

379
hydrodynamical and a.erodynamical characteristics. However, since the equations of sway-yaw and roll mo-
tion are often slightly coupled, for the purpose of prelim-
/3,deg
inary design it is permissible to implement the stability
analyzes under the assumption that one can neglect the
10 15
influence of the roll angle. The common analyzes ren-
der the quadratic equation for yaw-sway mode and the
further investigation of the directional stability in prin-
-10 ciple does not differs from the correspondent analyzes
for a displacement ship. Thus, it is possible to suggest
for evaluating the degree of inherent directional stability
20 the nondimensional stability indices and the so-called dy-
namic stability levelthe ratio between damping lever and
static stability lever which, respectively, locate the lon-
30 gitudinal centers of the rotary total force and the static
force acting on the ship.
Cz It should be emphasized, that in spite of this strong
-40 analogy between the directional stability of an ACV and
a displacement ship, the general stability of the ACV
Fig.5 Transverse force coefficient (b/a=0.4) motion with respect to the arbitrary equilibrium position
differs from the correspondent stability of a conventional
A strong qualitative and quantitative resemblance be- ship. Indeed, it is well known, that a dynamically sta-
tween theoretical and experimental results is evident. It ble conventional ship is always stable in the steady turn.
is surprising and notable that for the Froude numbers Moreover, even if it is dynamically unstable on a straight
Fn < 1.5 the aerodynamical forces acting on the hull of course , it becames stable in steady turn except for a s-
the ACV are negligible in comparison with aerodynami- mall range of rudder deflections 6 < b at the vicinity
cal forces. Since this range of Fronde numbers is of main of zero neutral angle. Such a peculiarity for a displace-
interest in the ACV maneuvering problems, one can say ment ship is caused by the typical dependencies of the
that the determination of the hydrodynamical compo- hydrodynamical forces on the angle of drift, i.e., they are
nents is the primary aim of theoretical and experimental always increasing functions of the angle of drift. Fig.3
investigations. Surely, it does not mean that in realis- Fig.5 demonstrate that the hydrodynamical forces of the
tic maneuvering conditions all remaiding aerodynamical ACV are strongly nonlinear functions with a sharp maxi-
components are negligible. As for example, the lifting ma. Thus, if the angle of drift in the steady turn exceeds
forces of the tail units and especially its moment due to the angle of drift which corresponds to the maximum of
the propellers slipstream are of the same order as the hy- hydrodynamical forces one can expect that the behavior
drodynamical forces. Another case, were aerodynamical of ACV can be nonstable.
forces are comparatively large concerns the ACV maneu- Notable, that in contradiction with a displacement
vering under the action of the wind. ship, the ACV nonstability occurs if the controls deflec-
tion exceed some critical angle b =
ACV MANEUVERING PERFORMANCES 5
The ACV Turning Performances
The ACV Directional Stability The radius of the turning circle is a widespread character-
For an ACV to be inherently directional stable (with con- istic of the ACV maneuverability. It can be evaluated by
trols fixed at the zero neutral angle) the latent roots employing the system of steady state equations of motion
of the characteristic equation which corresponds to the (1), where all derivatives with respect to time are equal
linearized equation of motion must satisfy some require- to zero.
ments, well known from the theory of stability of motion. Due to the nonlinearity of the right-hand side of these
Generally speaking, if one takes into account the roll of equations, the solution of this system is nonunique and
the ship, the characteristic equation will be of the third- exists only for some control deflections 6 < be,. Fre-
order. quently, three roots of the system exist. Two of them
correspond to the stable equilibrium position and the
'Due to the limited space and since the required detailed ana- remaiding one refers to nonstable equilibrium, which is
lyzes is too long and tedious, herein we give only the brief descrip- nonrealiazabe under stationary controls actions.
tion of major features relate to the analyzes of ACV maneuvering
equation.
At the same time, it is expedient to simulate the
380
 unnsteady ACV turn starting from the initial point t = 0. Indeed, herein we present only results, which allow com-
As a limiting case of such a simulation, it is possible to ob- paratively simple analytical solutions. Nevertheless, the
tain the steady turn characteristics for different controls, elliptic form of the ACV contour is actually close to the
but if the rudder deflection exceeds the critical value b, real geometry of ACV and, in turn, the general solution
then the steady motion is unattainable. As a rule, this for an any arbitrary contour may be obtained in the same
peculiarity results in the sudden and drastic increasing of manner without principal complications.
the drift angle and yaw angular velocities which, in turn, Exceptions do, however, occur. We are aware of fact,
enhances the tendency towards rollover. that modifications of the present theory are needed, espe-
Figure 6, which are given as an illustration, shows the cially with respect to the interaction between the flexible
time history of the drift and roll angle change, obtained skirts and the free-surface. In this context, the extrac-
from a simulated turning maneuver with the ACV SR.N2. tion of comparatively large components, exited by the
air cushion, is a feasible method for estimating the forces
10 20 30 40 t,sec exerted by the skirts.
An important aspect of this research, which is not pre-
sented here, is a detailed comparison between the theoret-
30knots 40 knots 50knots
ical simulation of ACV maneuvering with full-size crafts.
-10 More efforts should be directed towards modelling the-
oretically the actual configuration of existing crafts in
order to enhance the correlation with full-scale trials.

-20
Acknowledgment
The authors would like to thank Dr. L.Doctors for his
valuable comments.
Fig.6 The drift angle as a function of time (b = 200)
Figure 7 displays the dependence of the simulated
ACV trajectory on the initial speed. References
163M Barrat, M.J. (1965). The wave drag of a Hovercraft.
G

LO
J. Fluid Mech., 22, pp.39-47
08 Doctors, L.J. (1972). The Forces on an Air-Cushion
0.6 Vehicle Executing an Unsteady Motion, Proc. Ninth
Symposium on Naval Hydrodynamics, Paris, France,
0.4
Vol. 1, pp. 35-94
02
Doctors, L.J. (1975). The Experimental Wave Resis-
tance of an Accelerating Two-Dimensional Pressure
Distribution. J. Fluid Mechanics, Vol. 72, Part 3,
pp. 513-527.
Fig.7 The ACV trajectory (5 = 20°)
Doctors, L.J. and Dagan, G. (1980). Comparison
It is remarkable, that the ACV trajectory and its kine- of Nonlinear Wave-Resistance Theories for a Two-
matical parameters of motion are very sensitive with re- Dimensional Pressure Distribution. J. Fluid Me-
spect to the initial speed of the turning maneuver. This chanics, Vol. 98, Part 3, pp. 647-672.
feature distinguish the ACV from other types of maneu-
vering vehicles. Doctors, LI., Sharma, S.D. (1972). The wave resis-
tance of an air-cushion vehicle in steady and accel-
CONCLUSIONS erated motion. J. Ship Res., 16, pp. 248-260
Based on the results of the present study, it is clear that Doctors, L.J. and Sharma, S.D. (1973). The Accel-
the analysis of ACV maneuvering, when operating over eration of an Air Cushion Vehicle under the Action
the free-surface, can be adequately represented by a hy- of a Propulsor. J. Ship Research, Vol. 17, No. 2,
drodynamical theory of a moving pressure distribution. p.121-128.

381
D'yachenko, V.K. (1966). Wave resistance of a sys- Zilman, G.I., Sobolev, G.V. (1975). Hydrodynamical
tem of surface pressures in unsteady motion (Rus- reactions exerted by an elliptical pressure of distribu-
sian). Trans. of Leningrad Shipbuilding Inst, 52, pp. tion moving with an angle of drift (Russian). Trans.
83-91 of Lenin grad Shipbuilding Institute, Hydrodynamics
and Theory of Ship, pp. 30-35
Everest, J.T. and Hogben, N. (1967). Research on
hovercraft over calm water. Trans. Royal Institution Zilman, G.I., Shifrin, E.M. (1981). Wave Resistance
of Naval Architects, Vol. 109, No. 3, pp. 311-322. of Surface Pressure Distribution in 'Unsteady Motion
(Russian). Applied Hydromechanics, Naukova Dum-
Everest, J.T., Hogben, N. (1969). A theoretical and ka, Kiev, pp. 42-49.
experimental study of the wavemaking of hovercraft Zilman, G. (1982). Maneuvering of a Hovercraft
of an arbitrary planform and angle of yaw. Trans. (Russian). Leningad Shipbuilding Institute Press.
Inst. Nay. Architect,111, pp.343-35'7
Haussling, H.J. and Van Eseltine, R.T. (1978).
Waves and Wave Resistance for Air-Cushion Ve-
hicles with Time-Dependent Cushion Pressures. J.
Ship Research, Vol. 22, No. 3, pp. 170-177

Hogben, N. (1966). An Investigation of Hovercraft

Wavemaking, The Aeronautical Journal of the Royal
Aeronautical Society, Vol. 70, pp. 321-329.

Miloh, T., Dagan, G. (1985). A study of nonlinear

wave resistance using integral equations in Fourier
space. J. Fluid Mech.5159, pp.443-458
Murthy, T.K.S. (1970). The Wave Resistance of a
Drifting Hovercraft. Hovering Craft and Hydrofoil,
Vol. 9, No. 4, pp. 20-24.
Newman, J.N., Poole, F.A.P. (1962). The wave resis-
tance of a moving pressure distribution in a cannal.
Schiffstechnik, 9, pp. 21-26.

Ship theory handbook (Russian) (1988). Ed. by

Y.I.Voitkunski, Vol. 3, Manoeuverability of conven-
tional ships. Sudostroenie, Leningrad.

Wehausen, J.,V., Laitone, E.V. (1960). Surface

Waves, Encyclopedia of Physics, IX, pp.446-778,
Springer-Verlag, Berlin.
Yeung, R.W. (1975). Surfa,ce Waves Due to a Ma-
neuvering Air-Cushion vehicle. J. Ship Research,
Vol. 19, No. 4, pp. 224-242.

Zeitfuss, W. and Brooks, E.N. (1972). Prediction of

Static Aerodynamic Characteristics of Air Cushion
Vehicles through 180 Degrees of Yaw. J. Aircraft,
Vol. 9, No. 4, pp. 306-310.
Zilman, G. (1975). Evaluation of the side force exert-
ed by a moving pressure distribution with an angle
of yaw (Russian). Trans. of Leningrad Shipbuilding
Institute, 96, pp. 14-22.
382
 Scaling of Main Spray Generated by Planing Surfaces
S. Ryan (Tulane Medical School, USA)
R. Latorre (University of New Orleans, USA)

ABSTRACT
measurements for different model sizes
of a high speed semi-displacement hull
This paper presents the application [3]. These geosim measurements also
of the Latorre-Ryan spray scaling collapse onto a single line calculated
relationship to the Geosim tests of using the Latorre-Ryan scaling
eleven semi-planing hulls tested at
Froude Number F relationship.
= 1.0. Using the
results of the gpray scaling, it is
SPRAY SCALING RELATIONSHIP
possible to collapse the spray
measurements into a single curve. It is assumed the motion of the
INTRODUCTION
spray sheet is steady and defined by the
following eleven parameters.
One of the phenomena associated
Four physical properties: gravity
with planing hull operation is the g, density p viscosity V , and
,
formation of a spray sheet. In model
tests surface tension a .
the water surface tension
stabilizes the spray sheet and a spray
"blister" forms. In addition to being Three parameters characterizing the
hull: beam b, deadrise angle fi
an interesting hydrodynamic phenomena,
the spray is the source of model-full and trim 7 .
scale differences. The spray trajectory Four parameters characterizing the
is therefore an important parameter.
spray: thickness maximum ,

In the height Z, trajectory angle 0 and

Latorre-Ryan relationship
for spray scaling the spray velocity V.
sheet
trajectory is modeled as that of a
projectile in the presence of gravity. From the method of dimensional analysis,
the following dimensionless parameters
Dimensional analysis of the spray sheet are obtained.
results in a modified Weber number.
This is the surface tension parameter
STR based on maximum blister height [1]. STR =
To = STR Modified Weber
validate its use the spray blister V2 Z based on spray height
measurements of Savitsky and Breslin [2] Z.
were used to check (1)
the scaling LI
relationship. It was shown the scaling
relationship collapsed the spray data Rnb Reynolds Number
Vb (2)
for different test speeds into a single Rnb'
curve. In this paper the scaling V
relationship is used to examine the
scaling of the geosim model Cv Speed Coefficient
spray C. (3)
.J gb

Spray height to beam.

(4)

383
 6
Spray thickness to Eleven sets of data summarized in
beam. (5) Table 1 are examined at Froude Number F
= 1.0. Fig. 3 shows the variation
trim with model size at F = 0.9. It is
(6) obvious that at this test condition,
Deadrise angle. trim
13
smaller models operate at higher
This is in part due to the
values.
Trim angle. (7)
spray formation near the bow.
From photographs of the spray, the
Spray trajectory 4 and
0 lateral spray offset it , in Fig.
angle. (8) I height H in
the maximum spray
Fig. 4 were measured. The plot in Fig.
Using the results of a projectile 5 indicates a higher spray trajectory
in a vacuum, the maximum height of the for smaller models.
spray in Fig. 1 is given by:
OF GEOSIM MODEL SPRAY
SCALING
V2 MEASUREMENTS
Z= sin20 (9)
2g The experimental data was converted
equation 1 and
with into values of STR using
Substituting Z from equation (9) plotted in Fig. 6. The solid line was
the STR parameter gives: based on 1m model. Using the Froude
velocity scaling where:
2gc
STR - 42
sin 0
V
( 10)
V = V0/1-7.- where = L/Lo, the scaling

can be written for V =

relationship
This leads to the scaling relationship: constant as:
4 . 2 4
V sin 0 V
STR
STR = STR, -4-- ° = STR o Irt (12)
STR -
V' sie0
(11)
The scaled STR results in Fig. 6 are in
agreement with the
Using the spray scaling excellent
measurements.
relationship it was possible to collapse
the experimental data of Savitsky and
DISCUSSION AND CONCLUSIONS
Breslin [2] into a single line [1].
The present work is a continuation
GEOSIM MODEL TESTS [3] by
of the spray scaling developed
Latorre and Ryan [1]. In that study they
Recently a resistance test program on
semi-displacement craft investigated the influence of speed
of a high-speed the spray geometry. In the present work
in Figure 2 was completed by of
shown
in Japan [3]. we have investigated the influence
sixteen towing tanks size on the spray geometry. In
small models of 0.5 - model
Comparisons of
both cases the scaling relationship
3m and
0.75 m were made with larger 2m, proved valid. This supports the
effect
4m models. They showed the scale but following conclusions.
is not found in the wave resistance, and
appears in the spray formation
it
trim.

384
 REFERENCES
When models are tested at constant
Froude number, there is a scale problem
with the spray geometry and the vessel Latorre, R., Ryan, S., "Dimensional
trim. and Similitude Analysis of Spray Blister
Sheet from Prismatic Planing Models,"
By introducing the present spray Ocean Engineering, Vol. 16, No. 1, 1989,
pp. 71-83.
scaling relationship it is possible to
account for scale problem in the spray.
Savitsky, D., Breslin, J., "On the
Main Spray Generated by Planing
Surfaces, "Stevens Institute of
Work is underway to extend this
Technology Experimental Towing Tank
analysis to correct the trimming moment.
This Report 678, 1958.
will clarify the scaling
relationship for trim of a planing Tanaka, H., Nakato, M., Nakatake,
surface.
K., Ueda, T., Araki, S., "Cooperative
ACKNOWLEDGMENTS Resistance Tests with Geosim Models of a
High Speed Semi-Displacement Craft,"
Journal of Society of Naval Architects
The authors are grateful to Mrs. Latapie
of Japan, Vol. 169, June, 1991, pp. 55-
for typing the manuscript.
64.

TABLE 1. JAPANESE SYSTEMATIC DATA SET [3]

L V b
Zmax Zmin
No. M M/S M mm mm Japanese Towing Tank

1 0.4 1.97 0.09 29.7 22.7 Nagasaki Inst. of Applied Science

2 0.5 2.21 0.112 55.0 41.3 Hiroshima University
3 0.75 2.71 0.168 73.9 61.9 Hiroshima University
4 0.75 2.71 0.168 54.5 41.3 Nagasaki Inst. of Applied Science
5 1.5 3.83 0.336 11.2 87.3 University of Tokyo
6 1.5 3.83 0.336 11.2 87.3 Susitomo Heavy Industries
7 2.0 4.427 0.448 12.3 90.5
8
Shipbuilding Res. Cen. of Japan
2.0 4.427 0.448 17.5 14.2 Research Institute IHI
9 2.0 4.427 0.448 12.3 90.5 Mitsui Zosen (MES)
10 3.0 5.422 0.672 20.3 14.5 Meguro Model Basin
11 4.0 6.26 0.896 29.7 23.3 Ship Research Institute

385
FIG.1 MODEL OF
SPRAY T RAJECTORY
LPP 23.2 m
Z _T=V2sin20
zg B 5.2 m
2.4m
0.9 m
TT-rim 0.5 m

Y F1G.2 PLANING
"HULL LINES
[3] 0=20°

FIG. 3 PLANING HULL TRIM [3]

Trim
Fn =0.9 Trim = TG-Tfx1000/0
Lpp
4 \CD
S-1 .

CD
\cs) 0
0 R-
3

C
®
2 . 108
106 107
105 VL
REY NOLDS NO V

FIG.4 PLANING HULL SPRAY OFFSET [3]

tY6/7

3
O
L,rn MODEL LENGTH
386
 FIG. 5 PLANING HULL SPRAY HEIGHT [3]

1.0

05

L,m MODEL LENGTH

STR

pZ V2 FIG. 6 COMPAR I SON OF SPRAY STR

CALCULATED FROM FIGS. 4- 5
-3
10 WITH SCALED STR = STR clv4
KEY
1 SCALED STR.
A ,;)) CAL.STR
_ z(T)
From T_ESTS
STR.

-5
10

o 1 3 4 5
MODEL LENGT H L, m

0.7 1 1.41 1.73 2 2.12 yv

387
 SESSION IV-C

HIGH SPEED CRAFT--II

 Correlation of the USMC
High Speed Technology Demonstrator and
Propulsion System Demonstrator
J.G. Hoyt III, B.D. Abramson, D.D. Hayden
(Carderock Division, Naval Surface Warfare Center, USA)

ABSTRACT
A concept currently being developed is a tracked
The model to full scale correlation in resistance and vehicle that will perform hydrodynamically like a planing
powering between two scale model USMC waterjet craft during transit from the point of disembarkation to the
propelled high speed amphibious vehicles is shown. The surf zone where it will revert to its land operation
estimated require,d thrust for a 0.295 scale model of the High configuration. This transformation to sea mode is
Water Speed- Techno/ogy Demonstrator and a 0.245 scale accomplished by deploying: a double angled bow plate to
model of the Propulsion System Demonstrator are compared improve bow shape and decrease wetted length, a transom
to full scale results in a speed range of 5 to 30 knots. flap containing the high speed waterjet propulsors, and
Correlation above 22 knots, after con-ection for air ingestion covering the track area with chine plates that improve
into the waterjets, is found to be excellent. The effects of bottom roughness as well as increase beam. The major
air ingestion into the waterjets, effect of long time constants challenges in the development of such a vehicle, from a
to reach steady state, and an apparent hysteresis effect due to hydrodynamic standpoint, is the delivery of sufficient thrust
acceleration are also discussed. to the vehicle to propel it at the desired speed, and an
acceptable level of hydrodynamic performance while planing,
NOMENCLATURE particularly in waves. This includes tolerable wave loads on
the articulated structure of the vehicle deployed during
DHP Delivered Horsepower planing, ride quality, stability, visibility of the driver, and
GVW Gross Vehicle Weight other seaworthiness related problems.
HWSTD High Water Speed Technology Demonstrator
LCG Longitudinal Center of Gravity The present investigation is focussed primarily on
PSD Propulsion System Demonstrator the correlation between model and full scale results of the
USMC United States Marine Corps resistance and propulsion in calm water and the possibility
Weight Displacement of air ingestion into the waterjets. Model experiments were
Running Trim conducted in calm water and waves to provide data for the
design and verification of the High Water Speed Technology
Demonstrator (HWSTD) and the Propulsion System
INTRODUCTION Demonstrator (PSD). HWSTD was a 0.75 scale vehicle
constructed primarily to demonstrate the hydrodynamic
The U.S. Marine Corps has charged the Marine feasibility of the concept. HWSTD was powered by thre,e
Corps Program Office at the Carderock Division, NSWC 16.1 inch waterjets with a MTU MT-883 diesel engine for
(Code 124) and the Special Ship and Ocean Systems propulsion. An in depth evaluation of the total system is
Dynamics Branch (Code 1562) with developing technology ongoing using the PSD, a full scale test bed vehicle with
for their next generation of amphibious vehicles with greatly ceramic armor, mock-up weapons station, and the capability
improved capabilities. Such a vehicle's operational profile, to carry 15 troops and a crew of 3. PSD is powered by four
from a hydrodynamic point of view, would include the 16.1 inch waterjets and the combined output of a Cumins
capability of "over the horizon" deployment and rapid transit VT-903 diesel engine and a GE T-700 gas turbine engine.
to the beach at speeds in excess of 20 knots in at least a 2 Figure 1 shows schematics of the HWSTD and PSD with
foot significant sea. This, of course, must be accomplished both the as tested and as built dimensions given.
within the constraints of affordability, maintainability, and
without diminishing other desired improvements in combat
capabilities.

391
 A 0.295 scale model of the HWSTD and a 0.245 Sea Spe,ctra. There is little difference between the measured
scale model of the PSD were tested in the CARDEROCK, calm and rough water resistance of this hull form. It is
NSWC's Deep Basin Facility in June 1987 and August believed that this is primarily due to the double angled bow
1988. Both models were evaluated utilizing the thrustline plate, however, change of drag in waves is beyond the scope
database matrix approach (Ref. 1). In addition to the model of this paper. Secondly the estimated require,d thrust below
experiments, an inlet series was studied in June 1984 and a the range of model data was extrapolated assuming a speed
10 inch waterjet / inlet prototype was tested in March 1987. squared relation. ALso the extrapolation of required thrust
These waterjet / inlet tests were performed to validate the beyond the range of model data was made assuming that drag
on plane consisted of friction C), induced drag (A tan T)
design procedures utilized by CARDEROCK, NSWC Code
1562 and Engines Branch Code 2721. The full scale trial of and the experimentally derived "residual" resistance. The
HWSTD was completed in February 1990 while the testing running trim was estimated by a simple linear extrapolation
of PSD started in June 1991 and continues at the present of the mean pitch curve.
time.
DESIGN OF PS D
DESIGN OF HWSTD
In the case of PSD, the design problem was a little
During the design and construction phase of both different since the decision was made to save money by
the HWSTD and PSD, there were numerous design reusing the HVVSTD waterjet design and propel the craft
iterations in order to match available powerplants to with four 16.1 inch waterjets. The target weight for PSD
propulsors. In the case of HWSTD, the choice of the M'TU then became 54,500 pounds with an LCG of 10 feet forward
MT-883 as the propulsion engine narrowed the scope of the of transom for the light condition in order to preserve the
design to a point where a target gross vehicle weight (GVVV) 20% thrust margin. Since PSD was to be capable of
and longitudinal center of gravity (LCG) could be chosen. carrying 18 marines, it's heavy weight was estimated to be
The target design for HWSTD became a vehicle with a 57,500 pounds, which would reduce the thrust margin at
GVW of 36,530 pounds, and a LCG of 7.73 feet forward of hump to only 15%. This risk was considered worth taking
the transom. This vehicle was to have a fixed transom flap in view of the fact that HWSTD had successfully completed
set at 6 degrees and was to be propelled by three 16.1 inch it's full scale trials, essentially as predicted.
waterjets with a maximum speed of 1250 revolutions per
minute and a total of 1155 delivered horsepower.
Initially PSD was to have a fixed transom flap
angle of 8 degrees. However, after the model experiments
In selecting the waterjet propulsion system for the design was changed to a variable transom flap. Again
HWSTD the guidance given by Blount (Ref. 2) to size the the maximum waterjet speed would be 1250 rpm for four
propulsor to assure powering through hump speed, versus propulsors with the expectation of requiring 1540 DHP.
designing for desired top speed, was well considered. Based Looking at Figure 3 it can be se,en that the expe,cted top
speed of PSD was 24.6 knots at 1250 rpm in calm water.
upon the results of earlier model experiments it is quite clear
that this type of heavily loaded flat planing surface possesses
a very broad hump regime with a rapid decline in drag TRIAL RESULTS FOR HWSTD
beyond hump. In fact, a propulsor could be designed that is
capable of driving the craft at the desired 22 knots yet During the full scale trial of HWSTD the vehicle
incapable of overcoming the hump drag. In addition there achieved a top speed of 29.4 knots (corrected for current) as
was much uncertainty about the expected correlation between
measured by a Motorola "Falcon" Tracker. Our earlier
both the hull form modeling and the waterjet performance
expectations had been 30.8 knots. On the whole this is
especially considering the shortness of the inlets. To assure reasonable correlation with the model results. However,
that the test craft would exceed hump speed the propulsion there were several significant differences between HWSTD as
system was sized to provide in excess of 20 percent available model tested and built.
margin onthrust at hump.
The first thing to consider in investigating
correlation between model and full scale are the differences in
Given these sets of conditions, it was predicted that
HWSTD would reach a speed of 30.8 knots at 1250 rpm in the geometry, weights and centers. Geometrically HWSTD
calm water. Looking again at Figure 2 several items should was very close to the model tested. The difference as can be
seen in Figure 1 are small. In fact the greatest differences in
be pointed out. First the model experiments for HWSTD
entailed a speed range of 5 to 25 knots. These interpolated geometry is in the bow plate width where there is a
thrustline values are shown as open circles for calm water difference of only 2.0 inches shorter out of 106.0 inches.
and solid circles for a 2 foot significant Pierson-Moskowitz
Weight wise, the test craft was lighter than the model at

392
 35,825 pounds with a LCG further forward than the model at the intercept of the upper thrust available curve (a waterjet at
8.04 feet Both of these differences should have increased 1310 rpm) with the thrust required curve is 28.0 knots,
top speed to 31.5 knots as is seen by looking at the upper which again is excellent correlation between model and full
thrust available curve's intercept of the updated thrust scale measured speed of 28.5 knots.
required curve in Figure 4.
DETAILED COMPARISONS FOR PSD
The largest difference between our expectations and
actual performance was the fact that the test craft was unable The waterjets in both HWSTD and PSD were
to deliver an average of 1250 rpm for all three waterjets. turned by hydraulic motors. This allowed the indirect
Looking again at Figure 4, it can be seen that the intercept measurement of input torque, hence horsepower, by
of the lower thrust available curve (a waterjet at 1106 rpm) measuring the pressure differential across the individual
with the thrust required curve is 29.6 knots which is motors. Computer programs for design optimization and
excellent correlation between model and full scale measured thrust-speed-power estimations were used to define propulsor
speed of 29.4 knots. characteristics. For each steady speed data set the estimated
waterjet thrust, torque, and horsepower can be computed
TRIAL RESULTS FOR PSD utilizing the waterjet program by inputting the mean vehicle
overwater speed and mean waterjet speed. This data was
During the initial full scale trial of PSD, the provided in engineering notes and calculations by John G.
vehicle achieved a top speed of 28.5 lcnots (corrected for Stricker, Alan J. Becnel, and John G. Purnell of
current) and again measured by a Motorola "Falcon" Tracker. CARDEROCKDIV, NSWC, Code 2723. Thus the
Our earlier expectations had been only 24.6 knots, quite a estimated waterjet thrust can be compare,d to the expe,ctecl
difference considering our success with HWSTD. Again we required thrust.
must investigate several significant differences between PSD
as model tested and built The comparison of estimated output thrust to
required thrust for PSD in the light condition is given in
Geometrically, PSD as it was built, is very Figure 6. This plot shows the correlation between the
different from the model that was tested. The main required thrust estimated from the model results and the
difference as can be seen in Figure 1, is in the total chine to estimated waterjet output thrust for a series of steady speed
chine width of model and vehicle. The model was built with runs (30 seconds in duration) where waterjet speed was
a full scale equivalent width of 13.73 feet. This overall increased by 100 rpm increments. At first glance you can
width was changed during final design in order to eliminate see that the estimated output duust correlates well with the
"overlapping" chine covers under the vehicle when stowed required thrust up to a speed of 9 knots. In the range
and the vehicle is on land with a resultant reduction in beam between 9 to 12 knots the output curve rises above the
to 13 feet. Fortunately, the other appendages were also required curve only to fall back to the model curve at 12
reduced in size by varying percentages allowing for the 'mots. Due to the shape of the thrust required curve, there is
reanalysis of the model data at a scale ratio that would give a large increase in vehicle speed (11.8 kts to 15.4 kts) with
the correct beam. This was done fully understanding that a the next 100 rpm increment in wateijet spe,ed. The increase
geosym between model and full sc,ale had been slightly lost. in speed correlates well with the predicted plateau in the
However, scaling by beam should more closely approximate thrust curve. However, the estimated waterjet output is 15%
the full scale vehicle's performance better than any other higher than the model prediction. This trend of output
characteristic length. thrust being 10 to 15% higher than required continues out to
29 knots with the exception of two points which fall on the
Weight wise the test craft was heavier than the model curve.
model at 55200 pounds with a LCG further aft than the
model at 9.30 feet. Both of these differences should have In order to sort out what is occurring, the history of
decreased top speed to 24.0 knots as is se,en by looking at how each point was collected and in what order should be
the lower thrust available curve's intercept with the updated examined. In Figure 7 it is shown that the first set of data
thrust required curve in Figure 5. was collected, as previous described, by increasing waterjet
speed by 100 rpm increments in the order indicated by the
In addition to the difference in geometry there was arrows up to a average waterjet speed of 1260 rpm. The
also a difference between the expected and actual power variable transom flap was then changed from a setting of 2
delivered to the waterjets. L,earning from HWSTD, PSD degrees down to 9 degrees down. The change in transom
was built with excess power available to turn the waterjets flap, thus reduction in running trim, accounts for the
at a higher than designed speed. Looking again at Figure 5 increase in speed. The unloading in the waterjet accounts for

393
the slight increase in average waterjet speed from 1260 to estimated and measured power) is excellent The data points
1280 rpm. At this point two data collects were made each at speeds of 20, 15 and 11 to 9 knots are still above the
of 30 second duration. The throttle was then reduced to estimated curve.
lower the waterjet speed by 100 rpm. At this point the
estimated output and required thrust fell back to values CONCLUSIONS
which correlate well with the model results.
The thrust correlation of the one-third scale
The second set of data shown in Figure 7 shows the HWSTD model and one-quarter scale PSD model with full
same trend but at a higher spe,ed. In this case the vehicle scale is excellent in the planing regime. There is still the
was brought to full throttle at the 9 degree flap setting. As question of what, if any, hysteresis effect is present in both
vehicle and model due to change in speed, and if this effect is
you can see, four 30 second data collects were made at this
fixed throttle and flap setting. There is an obvious long correctly represented in the model. Further work is required
time constant required in order for the vehicle to reach steady with PSD to attempt to repeat the data presented at the same
state. This was observed in the model tests as well and can GVW and LCG as well as the heavier condition. Additional
be explained by the time required to drain the entrained water model experiments are necessary to further quantify the
in the track wells and transom flap area. However the observed hysteresis effect due to acceleration.
estimated output thrust is still higher than the required, and
once again with a 100 rpm decrease in waterjet speed the ACKNOWLEDGMENTS
craft's output and required thrust falls back to the expected
The authors would like to thank the test crews from
value.
CARDEROCKDIV, NSWC, Special Trials Unit and AAI
The data in Figure 7 indicates some sort of Corp. Without their de,dication neither of these vehicles
hysteresis effect due to acceleration and deacceleration. This would have been able to prove what they were c,apable of.
correlates with the model data in an anecdotal way. During We would like to thank Mr. Richard Swanek and Mr. Steve
the model experiments it was the practice to hold the model Ouimette, Program Managers for HWSTD and PSD who
fixed during carriage accelerations, and due to the length of held the programs together especially through the rough
the towing basins at CARDEROCK, NSWC, obtain the spots. We would like to especially thank Mr. Walter
complete speed set in a single pass of the carriage. It was Zeitfuss, Head of the Marine Corps Prograrns Office, who
observed that the scatter and settling time was improved by encouraged us to try a radical idea eight years ago and
this method versus allowing the model to be free during allowed us to stay with the program from model to actual
acceleration. Perhaps this has biased the model data to the vehicle.
low side, which is indicated by the scatter. However, further
model experiments are required before this can be said with FtEFERENCES
any certainty.
Hoyt, John G. and Dipper, Martin J., "A Database
Another explanation (other than the model Matrix Approach to Planing Craft Resistance Model
Experiments", American Towing Tank Conference
estimated thrust is higher than expected) can be that the
output of the waterjet is lower than expected. This might be (ATTC), August 1989.
seen in the comparison between estimated delivered
horsepower versus what was measured. The correlation Blount, Donald L. and Hubble, Nadine E., "Sizing
Segmental Section Commercially Available
between measured and estimated horsepower shown in
Propellers for Small Craft", Society of Naval
Figure 8 indicates that there was a decrease in absorbed
Architects and Marine Engineers 1981 Propellers
power at the higher power levels. This could be due to
Symposium, May 1981.
cavitation, or air ingestion or both. If it is assumed that
with the presence of air, the required torque to spin the
waterjet will decrease, it is also safe to assume for a first
order approximation that the output thrust will decrease by
the same proportion as the ratio of the powers.

The thrust deduction due to an assumed air

ingestion problem has been applie,d to the data in Figure 9.
All points have fallen closer to the model estimated required
thrust line. In fact the correlation beyond 22 knots (the
power level where there was the greatest variation between

394
 Amenuau Puff Scat inches u Tend by liodel
EXPer11.11.1
Cbicuured Fun Scale ltubesif
different dun model)

High Water Speed Technology Demonstrator

62.0

27.0

48 deg

200.0 16.0
12 dcg
Side View
950

1111=1
66.0
106.0 (66.25)
(104.0)

-4-- 44.8
2095
(210.88)
415.0

Expanded Bottom View

Propulsion System Demonstrator

74.26

30.00

74439 18.82

Side View

19.07
(15.42)

111:1
165.30
(156.0)
82.65
(75.0)
127 16
(125 16)

81.89

48.83 111. 77-82 .40.69 1663 .4-19.32

17.80 22634
72.

Expanded Bottom View

Fig. 1. Vehicle Schematics

395
 15000

U)'
10000
.0

7/)

GVW: 36,530 lbs

5000 LCG: 7.73 ft fwd transom
Transom Flap Set @ 6 deg
I e Calm Water
Calm Water with 20% Margin
ts 2 ft Sig Sea
4.-- 3 16.1 in Jets @ 1250 rpm

o
o 5 10 15 20 25 30 35
Speed (kts)

Fig. 2. HWSTD Performance Estimate (As Designed)

20000 111,1111
EXPECTED PERFORMANCE
24.6 kts Calm Water
@ 1250 rpm & 1540 DHP
1
15000

Ti; 10000
:

GVW: 54,500 lbs : .

LCG: 10.00 ft fwd transom

Transom Flap Set @ 8 deg
5000 : 4
---- --- - .
e Calm Water
Calm Water with 20% Margin

e4 2 ft Sig Sea
16.1 In Jets @ 1250 rpm

O 10 15 20 25 30
Speed (kts)

Fig. 3. PSD Performance Estimate (As Designed)

396
 15000

MEASURED PERFORMANCE
; 29.4 kts Calm Water
:
g 1106 rpm & 1020 DHP
Test Craft Unable :
--- --

10000 To Provide Maximum

Power To Waterjets
.0 _;

GVW: 35,825 lbs

5000 LCG: 8.04 ft fwd transom
:
Transom Flap Set
6 deg Model / 4.8 deg Vehicle
e HWSTD 35825 lbs / 8.04 ft CW
- - - - HWSTD 35825 lbs / 8.04 ft CW 20%
o 3-16.1 in Jets @ 1250 rpm &1155 SHP ? ---

---
3 16.1 in Jets @ 1106 rpm & 1020 SHP

5 10 15 20 25 30 35
Speed (kts)
Fig. 4. HWSTD Performance (As Tested)

20000
: . . : : : : :

MEASURED PERFORMANCE
4 4 285 kts Calm Water
- Test Craft Provided @ 1310 rpm & 1615 DHP
-- Excess Power
-

15000 To Waterjets
"
: : I I . . . . . . .

.--i----i-----i. ---- i ---- i --- -i---- . ---.. ---- :4-

,---4 ---- 4-4+ ---- : ---- : ---- 4-4-4-4 ---- :,-- . ---4 ---- : ---- : ---- 7----,----r ---- i ---- i-------- ---- ---- ., ----
10000 . . . . . . . . .
. . . . . . . . . . . .

.,:
. 4.
. i- -i i : 4. i.- -i i- -4
:
:
- - i ---- -i.----? ----i ---- i ---- i-----.-----i- GVW: 55,200 lbs
7-
.
4 ---- ; ---- ' ---- 4----; ----- - -;'' LCG: 9.30 ft fwd transom -- i ---- .:.......". ---- '.. ---- : ---- .......
' ---- 1 ---- 1 ---- i: : ----
. ' ' -- Transom Flap Set . :

: : 8 deg Model / 9.6 deg Vehicle ;

5000
;
. I
; 4 ;
;. e Calm Water
Calm Water with 20% Margin
4 - 16.1 in Jets @ 1250 rpm
: --------------
: : : ---- 4 16.1 in Jets @ 1300 rpm :
--
:
---- ---
: : .

5 10 15 20 25 30
Speed (kts)

Fig. 5. PSD Performance (As Tested)

397
 20000
. . . .
.:

. . .
.
.
' 1300 rpm

. .

1250 rpm

15000

10000
. : :

: GVW. 55,200 lbs :

:
LCG: 9.30 ft fwd transom - .... : :
.... : ....

----o--- Model Data Calm Water

5000 stk.

Trial Data @ TF 2 deg

.
o.. .
.
Trial Data @ TF 9 deg
4 16.1 in Jets @ 1250 rpm
: . . :

: -15 .
4 16.1 in Jets @ 1300 rpm
11111=1211111M111111111111111111111111=1111

15 20 25 30

Speed (kts)
Fig. 6. PSD Raw Trial Data

20000

: .NO Ti-HRO. TI'LE. CHANGE.

128D rpm ir .. .... .... .... ....

15000
i ;
Hysteresis
:

. .... : .... 1193 rpm 1310 rpm

:

10000 :

1226 rpm .1

:
.... .... : ....
.... ..
GVW: 55,200 lbs . .... ... .... ..
LCG: 9.30 ft fwd transom .... .. ..

...
- - Model Data Calm Water
5000 -

Trial Data @ TF 2 deg

Trial Data @ TF 9 deg
......
4 16.1 in Jets @ 1250 rpm
. ....
4 16.1 in Jets @ 1300 rpm

5 15 20 25 30

Speed (kts)

Fig. 7. PSD Raw Trial Data History

398
 1800

1500 GVW. 55,200 lbs

LCG: 9.30 ft fwd transom

1200

900

600

300

o I

300 600 900 1200 1500 1800

Computed Delivered Horsepower

Fig. 8. PSD Waterjet Performance

20000

i
15000

.0

z 10000

e Model Data Calm Water

Trial Data @ TF 2 deg Corrected
5000
Trial Data @ TF 9 deg Corrected
4 4 16.1 In Jets @ 1250 rpm
..
4 16.1 in Jets @ 1300 rpm ...

lIti
.. ....

.I;;;f;;; ;---
.. .... .... ... .... ...

I I

5 10 15 20 25 30
Speed (kts)

Fig. 9. PSD Correlated Trial Data

399
 Scaling SES [Surface Effect Ship] Motions
D.R. Lavis (Band, Lavis & Associates, USA)

ABSTRACT ing motions for an SES operating hull-

borne, in irregular head seas, are
There has been renewed interest concern- generally more precisely predicted from
ing the subject of scaling, to full- model tests than are the motions for an
scale, the seakeeping motions of SES SES operating on cushion.
models. This paper has been prepared in
response to this interest. Experience nas shown, however, that in
general, the standard deviation of the
The paper reviews the historical back- motion amplitudes of a subscale SES
ground to the subject as treated in the model operating on cushion can duplicate
United States and the United Kingdom and full-scale behavior with reasonable
important items of difficulty are dis- accuracy, whereas the scaling of the
cussed. Future studies which can fur- corresponding vertical accelerations,
ther our understanding of the problem unfortunately, are less well understood,
are suggested. principally because of a lack of compa-
rable data.
NOMENCLATURE
Experience with fully amphibious ACVs at
ACV Air Cushion Vehicle the linear towing tanks at EEL in
EEL Experimental & Electronic England, where the opportunity for such
Laboratories A
correlation is, perhaps, the greatest,
- Division of
has shown that accelerations are very
Westland Aerospace
dependent on the structural stiffness of
Vent-Valve Controller Gain the model and full-scale craft. EEL
Atmospheric Pressure experience is that models tend to be
Pa
very much stiffer than the full-scale
Cushion Pressure craft. Therefore, it is difficult to
establish a precise correlation. How-
SES Surface Effect Ship ever, EEL comparisons of SR.N4 model-
V Cushion Volume scale and full-scale accelerations have
suggested that their models produce
WHCF Wave Height Correlation Factor higher accelerations than the full-scale
craft such that there is a wave height-
Linear Scale Ratio correlation factor (WHCF) for accelera-
Y Ratio of Specific Heats tions in the same way that there is for
rough-water resistance, where WHCF is
INTRODUCT ION the ratio of full-scale wave height to
model Froude scaled wave height neces-
There are numerous factors which nor- sary to obtain the correlation. EEL
mally prevent a subscale SES model from uses a value of 1.67 for the WHCF when
precisely duplicating the seakeeping interpreting model-scale rough-water
motions of
a full-scale SES. It has resistance data, and a value of 1.2 for
been found experimentally that seakeep- interpreting model-scale acceleration

401
 full-scale accelerations
Thus, tests, which employ a fixed-in-surge
data.
measured in a sea with a significant set-up in either regular or irregular
wave height of 6 feet, for example, are head seas. They have, however, tested
expected to be equivalent to model ACV models in irregular head seas using
acceleration measured in a scaled sig- both fixed-in-surge and free-to-surge
nificant wave height of 5 feet. EEL has techniques. Although EEL has no direct
comparison of ACV test results using the
indicated that the value of 1.2 has
resulted in good correlation between two different test techniques, it is
model-scale and full-scale SR.N4 verti- their belief that ACV behavior is not
locations realistic when testing in irregular head
cal accelerations for some
within the craft; however, the correla- seas using fixed-in-surge techniques.
tion in other craft locations was very This is due to the belief that the model
poor, even with the WHCF applied. This is "being driven (unrealistically) into"
is primarily the occasional large waves when fixed in
variation, they believe,
result of the differences in surge. This, they believe, results in
the EEL
structural stiffness which exist between unrealistic drag, motions, etc.
also believes that the degree of un-
the model and full-scale craft.
realistic behavior experienced when
value used for WHCF would also
using a fixed-in-surge set-up might not
The
naturally depend to some extent on the be as bad in regular waves as it is in
irregular waves. The majority of the
relative size of the model and full-
The value of 1.2 used by evidence, however, suggests that the
scale craft.
EEL would, therefore, be appropriate for
difference between the two test tech-
and these niques is small.
the models tested at EEL,
models have linear scale ratios gener-
Also, for Numerous attempts have also been made to
ally between 10 and 14.
rough-water resistance, WHCF has been explain the scaling process theoreti-
cally (References 1 through 5, for
found to vary slightly with the location
example). These attempts have concen-
of the full-scale trials, due presumably
to a change in the type of sea state
trated on the fact that atmospheric
For craft motions, this pressure is not scaled down when testing
experienced. Since the ther-
dependency is unknown, but is believed subscale models.
Also, ACV tests conducted modynamic compression and expansion of
to be small.
the cushion volume which effects changes
by EEL for Band, Lavis & Associates, pressure is governed by
for in cushion
Inc. in 1987 have shown that WHCF
varies with absolute rather than gage pressure, the
rough-water resistance
wave height and forward model and full-scale craft do not behave
significant
A similar effect may also influ- in the same way unless the atmospheric
speed.
correlation of seakeeping pressure is correctly scaled.
ence the
motions.
All of these theoretical studies con-
Tests at the EEL linear towing tanks are cluded that, because of this lack of
usually conducted by towing the model correct scaling, models tend to under-
Normal prac- predict the vertical accelerations at
with a free-to-surge rig.
tice in the USA is to test models fixed full-scale for the range of encounter
frequencies of normal interest. The
in surge. Experience from testing SES
linear seakeeping simulation of Refer-
both ways at the Davidson Laboratory of achieved good
at DTRC have ence 1, for example,
Stevens Institute and
difference is correlation by simulating, in both the
indicated that the
time and frequency domains, the motions
negligible.
of a model in pitch, heave and roll
using model parameters as input. Using
EEL has not made a similar comparison full-scale parameters as input the same
between the two techniques because EEL that an
seakeeping simulation revealed
does not have any experience with SES
402
increase in vertical acceleration would generated waves seen by a full-scale
occur for the same scaled sea spectra. craft are short-crested, i.e., multi-
The full-scale craft, however, when directional. A significant contribution
tested several years later, revealed to the heave acceleration of an SES,
that the acceleration response was (in however, is the sudden closure of
general) not too dissimilar from that cushion flow from beneath the two-
originally predicted by the towing tank dimensional end seals. When operating
model, although the comparison was again in long-crested waves, the rates of
confused somewhat by differences in closure of cushion flow may be even more
model and full-scale sea spectra. sudden and may result in larger excur-
sions of cushion pressure and hence
This experience and that of EEL suggests larger heave acceleration.
that there are many factors which, if
not represented adequately, will distort This phenomena is believed to contribute
the results obtained from either physi- significantly to the reason why EEL must
cal or mathematical models. use a wave-height correlation factor
when interpreting model seakeeping
Since a model is subjected to the same results. It is interesting to note
gravitational constant as a full-scale that, in a typical multi-directional
craft, it is necessary to scale model sea, about 80% of the wave energy is
parameters such that the ratio of iner- contained by those components which have
tial forces to gravitational forces is directions within 45° of the main direc-
equal for both. This ratio is termed tion of the sea. It is believed that
the Froude Number VAg and simple dimen- this 80% factor is closely related to
sional analysis yields the relationship the wave height correction factor of
for other parameters in terms of the 1.2.
linear scale factor X.
Also, differences in wave shape and wave
The imperfections in model tests arise spectra between model and full-scale
due to physical limitations of the always occur to confuse the correlation.
models and the representation of the Most often wave buoys do not measure the
operating conditions. A summary of the waves encountered by the full-scale
ideal scale factors and of the achiev- craft, nor are bow-mounted wave-height
able representations of the most sig- sensors adequately reliable. These
nificant parameters is given in Table 1. factors have made it particularly diffi-
Of these the most influential and most cult to make truly scientific compari-
difficult to handle include: sons between available full-scale and
model-scale measurements.
The Sea State
Dynamic Properties of End Seals
Cushion & Seal Fan Characteristics DYNAMIC PROPERTIES OF END SEALS
Atmospheric Pressure.
Fluctuations in cushion pressure which
The principal difficulties encountered contribute significantly to SES heave
in each of these are discussed below. accelerations result from (1) a rate of
change of cushion volume, due to craft
motion and the motion of waves between
SEA STATE the sidehulls and (2) a rate of change
of air leakage area from beneath the end
One most important feature of the sea seals and sidehulls. It is the second
state generated in a linear towing tank of these two effects which is in-
is that waves are long crested, i.e., fluenced, to a large extent, by the
unidirectional. Most often the wind- dynamic properties of the end seals.

403
 Table 1. Effects of Model Scale

Scale Factor

Ideal Achleved Cause and Effect of Incorrect Scaling

Parameter

Model

?t, -X Manufacturing tolerances at small scale limit accuracy of representation. For example,
Characteristic Length
the inflated geometry of the seals are often at least 5% in error.

-X2 Errors in the seal geometry give hse to errors in cushion area. This is only a small effect.
Area X?

X X' Mass and weight can be scaled precisely but with increasing difficulty as the scale of the
Ma. or Weight
model becomes small.

X.5
Inertia can be modeled preci.ly if the full scale value is known with accuracy. This is not
Moment-of-Inertia A.
generally the case for development tests before the craft is built.

Model Operating
Characteristics
ki.s -X15 For tests with a restrained model only an operating point on the fan curve need be repre-
Lift System Fan Slope
sentad. This can be achieved with accuracy if means of me.urement and adjustment
are provided. For dynamic tests the corrplete lift system of the full-scale craft should be
faithfully modeled. However, flows are subject to Reynold's Nurnber effects and often a
correctly scaed cilaracteristic can only be achieved by utilizing non-scaJed components.
There is also difficutty in accurately characterizing a complex distribution system, statically
and dynamically, particularly when there are complex interactions with the supplies to the
seals.

X The gauge value of cushion pressure can be correctly scaled as this depends primarily
Cushion Pressure X.
upon the model weight, sidehull geometry, and cushion area.

1 Where structuraJ effects are significant, such as flexible seals, utilizing the .me basic
Stiffness (Young's X
materials at model scaJe as full scale misrepre.nts the deformations and loads Mod-
Modulus of Materials)
ifying material thickness to compensate then misrepre.nts inertia and strength. Overall
hull structural dynamics can be successfully represented by properly scaling the hull
flexural stiffne..

Model Test Condition

1
Ideally, atmospheric pressure should be scaled while retaining a non-scaled value of
Atmospheric Pressure X
density. This is impo.ible. This causes the compressibility effects of the lift system and
particularly the cushion to be misrepresented and affects the dynamic response in a co,
plex manner. To overcome this problem, a cushion characteristic synthesizer can be
u.d. To predict full-scale motions, mathematical models can include the effect but the
mathematical representation must be proven through model test correlation.

k X. Random seas can be generated that have spectra corresponding reasonably closely to
Sea State (Wave Length mathemati.11y described seas such as the Pierson-Moskowitz. However, the seas have
and Height)
a uniform transver. shape (long crested waves) which is not a true representation of real
seas. A stationary or rreving point spectra does not corrpletely descrbe the surface of a
sea and, therefore, the accuracy of the representation .nnot be confirmed.

Measured Character-
istics

-X? Forces and Moments are subject to the errors, introduced directly or indirectly, by not
Force and Moments X' scaling the environment of the test such as atmospheric pressure. Other errors are due to
(Drag, Lift, Sidefor., Etc.)
non-scaled viscosity, (Reynold's Number) non-aled surface tension (Weber's Number)
non-scaled pressures (Cavitation Number) which affect both aerodynamic and hydro-
dynamic components of force. Other errors are introduced due to interference of the
towing rig, channel width and channel depth. Also, the restraints applied to the model.
These errors are in addition to the errors due to the limitations of instrumentation.

-X'
These are modified as discu.ed above, if supply and other characteristics are not
Stiffness (Pitch and Roll) X' function of rnany
properly represented. 11 is difficutt to a.ess the error as stiffness s a
variables including operation conditions such as speed as well as configuration.
Stiffness (Heave) X' -X2

x.4.5 -k4,5 These are aJso modified as di.ussed above and the errors are even more difficult to
Damping (Pitch and Roll) are involved in a very
estimate as many parameters that determine dynamic response

X.2.5 ...X.25 complex manner.

Damping (Heave)

404
Thus, it is important that model and are higher at model scale than what
full-scale end seals have dynamic would normally be indicated by Froude
similarity. To achieve this the mass scaling. This disparity is accentuated
and stiffness of the skirt material and as the scale ratio increases, or as
the characteristics of the fans that larger full-scale craft are considered
supply air to the seals should faith- relative to the limited size of towing-
fully represent a true scale of the tank models.
full-scale craft. This is often not
achieved. This difficulty can be overcome, to some
extent, by following the procedures used
in Reference 1, which first establishes
CUSHION AND SEAL FAN CHARACTERISTICS correlation with the theoretical sea-
keeping prediction at model scale, and
It has been demonstrated by simulation then predicts the full-scale motions
(Reference 1) and by model tests that using the same math model. However, as
variation in the slope of the pressure- mentioned earlier, predictions of full-
flow curve (ap/aQ) has a significant scale vertical accelerations, made in
influence on craft dynamic behavior. this way, were much higher than those
Reference 1 has shown that variation in subsequently measured in full-scale
total air-flow rate has a significant trials.
affect. The variations in the spatial
distribution of flow is also important. Thus, based on what currently exists (in
The dynamic behavior of the air in the the U.S. and at EEL) in the way of
ducting and fan volute, which results in correlation between model and full-scale
a frequency dependent hysteresis of the data, it is concluded that the distor-
output pressure and flow, will also tions in model-scale damping and fre-
influence the dynamics of the cushion. quency are not as dramatic as suggested
Thus, it is vital that both physical and by the analyses of References 1 through
theoretical models represent, as closely 5.
as possible, the intended full-scale
characteristics. Because of Reynolds
Number effects, a geometrically-scaled SUGGESTIONS FOR FUTURE STUDY
model fan will always underperform.
Although these effects are well under- Examination of Experimental Data
stood, calibration of the flow charac-
teristics of a new model should always It is believed that a significant con-
be made. tribution to understanding the motion
scaling issue will result from examining
model and full-scale data that already
ATMOSPHERIC PRESSURE exists. There have been numerous SES
and ACV craft built and tested in the
Analyses (References 1 through 5) con- U.S., the development of which were
cerned with the scaling of the motions supported by model-test programs. No
of subscale models have demonstrated concerted effort, to date, has been made
that a distortion exists in the fre- to examine the correlation between the
quency and damping characteristics of motion behavior of these models and
models that do not conform to the normal full-scale craft.
(i.e., Froude) scaling process. This is
due to the inability to correctly scale Theoretical Analysis
the atmospheric pressure at non-scaled
density for the subscale model. Refer- References 1 5 provide simpli-
through
ences 1 through 5 all agree that the fied analytical representations of the
natural frequency and damping of the cushion heave dynamics of an ACV or SES.
cushion heave-pressure mode of response The application of these equations will

405
show a significant increase in vertical heave attenuation. Additional lift-air
acceleration response at full-scale com- flow rate would be required for the
pared to model-scale at heave-resonant model to compensate for the flow lost
conditions. through the vent valves.

It is believed that this effect is REFERENCES

unrealistic based on the empirical
evidence discussed previously in this 1. Lavis, D.R., Bartholomew, R.J., and
paper. It is suggested that a signifi- Jones, J.C., "On the Prediction of
cant improvement in our understanding of Acceleration Response of Air Cushion
scaling cushion dynamics would result Vehicles to Random Seaways and the
from modifying the equations to include Distortion Effects of the Cushion
a more realistic representation of heave Inherent in Scale Models," AIAA
damping, particularly for the presence Paper No. 72-598, AIAA/SNAME/USN
of sidehulls and end seals. Advanced Marine Vehicles Mtg, An-
napolis, Maryland, July 1972.
Cushion Synthesizer
2. Shenfil, L., Notes on SEV Heave
It is suggested (Reference 2) that the Dynamics, Aerojet General Liquid
installation, on a subscale model, of a Rocket Company (ALRC), Sacramento,
controllable cushion-vent valve will California, March 1974.
satisfy the requirements for Froude
This would involve active 3. Kaplan, P. and Davis, S., "A
scaling.
of the vent valve via a Simplified Representation of the
modulation
cushion pressure feedback system, with Vertical Plane Dynamics of SES
K, the overall gain of the system, fixed Craft," AIAA/SNAME Advanced Marine
at a constant value, as given by the Vehicle Conference, 1974.
following equation:
4. Lavis, D.R., "Dynamic Stability and
Control of Surface Effect Vehicles
Vc 1
K -y-
P + PI (1) (SES and ACVs)," Chapter 4 of Book
a
Pc + Pa on SES and ACV Design, 1975.

can then be 5. Band, E.G.0 and Lavis, D.R., "Heave

Full-scale representation
obtained by the model. Stability of Air-Cushion Supported
Ships," PRADS' 89 Conference, Varna,
The system installed on a model would be Bulgaria, October 1989.
similar to that used for full-scale

406
 A Simplified Power Estimation Method for
Monohull High-Speed Small Craft
C.M. Lee (NKF Engineering Inc., USA)

ABSTRACT DISCUSSION

This paper presents a summary of The technical approach consisted

the results of a study of the parametric of three primary activities:
interrelationships between size, weight,
hull proportions, propulsion power, and Data Reduction
craft speed of modern, high performance,
monohull marine craft. The objective of Parameter Derivation
the study was to derive an empirical Results Evaluation
method for: i) expeditious definition
and optimization of hull geometry, and
The raw data which provided the
propulsion plant of candidate craft
concepts, and foundation for the study was derived
ii) assessment of
from a U.S. Navy survey of modern,
performance claims in boat design
patrol boat type, marine craft which was
proposals and similar technical
conducted in the Summer of 1991
literature.
(Reference 1). Preliminary parametric
Parametric relationships were
derived to describe the performance relationships were adjusted to
characteristics of high speed craft in correlation with specific full-scale
test data. Finally, performance estimate
terms related to lifting surface theory
(i.e aspect ratio and pressure loading based on parametric data were derived
of the planing surface) rather than in for a number of known craft and
more conventional "naval architectural" specifically compared with data derived
such from reports on the full-scale,
terms as hull geometry
coefficients, slenderness ratio etc. The
instrumented, testing of several craft
(References 2 through 7).
data provide a fast and convenient
method for estimation of propulsion It should be noted that the initial
power/speed relationships which requires
database was constructed from
only limited craft design information descriptive information voluntarily
and yields results consistent with the provided by commercial boat builders.
precision necessary for the intended The commercial nature of much of the
usage. material introduced issues regarding
The scope of the paper includes a consistency in interpretation of
brief review of the technical approach displacement conditions and the veracity
used to accomplish the work, typical of performance statements.
data trends and the charts which provide Interpretation of displacement
the basis for the simplified power conditions, ("Full Load", "Half Load",
etc.) as reported in the literature,
estimation method. Performance estimates
derived by means of the method are were considered to be beyond the
compared with data derived from the interest of this work. Issues regarding
careful and thorough testing of full- displacement conditions were therefore
scale craft. avoided by use of the term "Gross

407
Weight" rather than "Displacement which the craft becomes sufficiently
Condition" to describe the mass value as supported by hydrodynamic pressure that
specifically related to performance the hull may be considered to be a
considerations. Mass data given in the lifting surface. The objective of the
descriptive material were adjusted, as initial analysis was to derive a
necessary, to give values in Pounds (lb) description of the relationships between
and Long Tons (LT) of 22401b. the five boat design descriptors
principally related to craft
The issue regarding performance performance:
claims was principally a concern
regarding the accuracy of weight/power/ Length Le(ft)
performance relationships calculated Breadth Be(ft)
from commercial data. This issue was Weight W(lb or LT)
resolved through the course of the study Speed V(Knots)
in that raw performance data were used Power P(BHP)
only for the purposes of establishing
the fundamental characteristics of The descriptors listed above were
"trend lines" in the initial analyses. addressed in terms of several
These initial trend lines were later proportional relationships:
calibrated to test data in the
derivation of specific parametric Inverse Aspect Ratio Le/Be
relationships. Speed Ratio VPiLe
Additionally, consolidation of Power Density P/W
performance information related to a Static Press.Loading (p) W/Le*Be
large number of craft designs raised the Dynamic Pressure Coeff. W/(Le*Be*V2)
issue regarding consistency in the
definition of the hydrodynamically Analysis addressed parametric
effective hull dimensions. Detailed relationships between:
analysis of each craft to accurately
define hydrodynamic geometry was not Gross Weight
feasible within the constraints of the
Power Density
information available. Approximate,
performance related, hull dimensions Pressure Coefficient
were therefore derived on the basis of Static Pressure Loading
overall hull dimensions as follows:
Pressure Coefficient
Effective Length (Le) = 0.80*L0A
Speed Ratio
Effective Breadth (Be) = 0.90*BOA Aspect Ratio
The results of the work are
For the above assessments overall hull summarized in the following paragraphs.
length and breadth are considered to be
the hull dimensions remaining after Data Reduction
deduction for bow and stern platforms, Typical results of the data reduction
transverse outriggers, or similar type are shown in Figures 1 through 3. Note
appendage structures. that the scatter of data around the
Analysis was constrained to address general trend lines is indicative of the
only the parameters fundamentally influence of: a) other parameters, and
related to performance of boats in the b) the presence of spurious data.
planing mode. That is, the performance
at speeds above the minimum speed at

408
Parameter Derivation general, follow the trend line, but
Parametric relationships were derived appear relatively "heavily" loaded.
through analysis of the data scatter Figure 3 therefore provides some
shown in the general trend charts to guidance on the practical upper and
approximately define the influence of lower limits in Dynamic Loading relative
secondary parameters. Some winnowing of to Speed Ratio. Note that low values of
data, curve fitting and cross-fairing dynamic loading are indicative of low
were accomplished to smooth the system weight (particularly hull
parametric surfaces. The characteristics structure), high speed (high power
of the approximate parametric surfaces density) or some combination of these
were then refined by correlation with variables. It should also be noted that
test data derived through carefully the "Gentry Eagle" and the "CPIC" and
planned and executed tests of several the 56ft.Target Boat are very light,
full-sized, craft. It should be noted high power, craft. Proposed craft
that the characteristics of the tested designs which reflect significantly
craft were not included in the data set lower dynamic loading justify careful
from which the trend lines were derived. evaluation of their technical viability.
Two methods were used in the Figure 5 shows the basic
correlation activity: interrelationships between Power
Density, Speed Ratio and Pressure
i.Pressure Loading.Pressure loading based upon a reference
coefficient data, derived from analysis Length/Breadth ratio of 3.91. Figure 6
of test results, were merely overlaid on shows Power Density correction factor
the trend display to provide a measure for Length/Breadth ratios different from
of the practical accuracy of the the "standard" Length/Breadth ratio of
predicted trend. 3.91.
Figures 5 and 6 provide the means for
ii.Power Density. Power Density the rapid estimation of the power/speed
correlation was converged by iterative relationship for high-speed (V/-NiL > 3)
adjustment of the approximate parametric marine craft using only weight,
surface characteristics to progressively principal dimensions, and speed
reduce the difference between the information.
approximate trend characteristics and
the collective body of test data. ¡Results Evaluation
Power/Speed relationships derived
The results of this work are shown in through the application of Figures 5 and
Figures 4 through 6. Figure 4 shows the 6 were compared with actual performance
relationship between Pressure relationships accomplished in full-scale
Coefficient and Speed Ratio, as derived tests. Typical results are shown in
from Figure 3, compared to upper and Figures 7 through 12 and discussed in
lower practical boundaries derived from the following paragraphs.
full-scale test data. As shown the
curves derived from test data, in Coastal Patrol (Interdiction) Craft
general, follow the shape of the trend (CPIC). Comparison of calculated and
line. Note that the "Gentry Eagle", CPIC test performance characteristics for the
and 56ft. Target Boat correlate very CPIC is shown in Figure 7. Note that the
closely and slightly below the calculated and test power lines are
"Reference" trend line and, therefore, essentially parallel with the calculated
may be considered to be relatively power line separated from the test power
"lightly" loaded. The "Seafox", 110ft. line only by about 1% SHP (or about 0.5
Patrol Boat, and HSB "Harley" also, in Knot) at 36 Knots (V/'Le= 4.00) and the

409
lines intersect at a speed of 39 Knots "Seafox" are shown in Figure 11. Note
(VPVLe=4.36). that the calculated power line and test
power line are in close agreement for
"Gentry Eagle". Calculated per- speed ratio from about 25 Knots to 35
formance characteristics and test data Knots (V/,'Le values about 4.00 to 6.00)
for the "Gentry Eagle" are shown in except that the power line for one test
Figure 8. Note that the calculated and (with 22X30 propellers) demonstrated a
test power lines are essentially powerlevel slightly above the
identical at craft speed up to a speed calculated values and other test data.
of. approximately 42 Knots (V/4Le=4.00).
Above 42 Knots the predicted power line 56ft.Target Boat. Calculated and test
diverges upward from the test line. At a performance characteristics for a 56ft.
speed in the order of 60 Knots the Target Boat are shown in Figure 12. Note
divergence is such as the calculated that the predicted line is close to and
power exceeds the test power by 5 or 10 slightly above the test line at low
percent. It is believed that the speeds but intersects the test line at a
divergence demonstrated in this study is speed in the order of 31 Knots
the result of the improved overall (v/,,h,e=4.68). The discrepancy between
propulsive efficiency provided by the calculated power and test power
waterjet propulsors at speeds above 35 lines was not resolved within the course
Knots. of this study but is believed to be, at
least in part, the result of the unusual
110ft. Patrol Boat. Calculated per- number of propellers installed on this
formance characteristics and test data craft and, hence, the relatively large
for a 110ft. Patrol Boat are shown in influence of appendage drag on
Figure 9. Note that the calculated power speed/power characteristics.
line essentially parallels the test
power line but appears to underestimate CONCLUS ION
the power level. The precise source of
the discrepancy was not resolved within The results of the power calculations
the course of this study. However, it is demonstrated good agreement with the
believed to be, at least in substantial power/speed lines derived from full-
part, the result of drag produced by the scale test data. Discrepancies between
permanently deployed, active-fin, roll calculated and test values may be
stabilizers with which the craft was attributed to the influence of
equipped. particular underwater appendages and/or
propulsor configurations. On this basis,
High Speed Boat (HSB)"Harley". Test it is believed that the work
data and calculated performance accomplished to date offers an approach
characteristics for the HSB "Harley" are toward the expeditious assessment of
shown in Figure 10. Note that the size, speed and power relationships of
calculated power line shows good modern high speed marine craft.
agreement with test data. Also note Refinement is expected to define
that, in this case, the calculated power additional correction factors such as to
was adjusted to account for the account for particular configuration
reduction in appendage drag resulting details and operational conditions.
from the installation of "Arneson" type Additionally it is anticipated that the
surface-propeller drives. next stage in development will
incorporate weight distribution, and
"Seafox". Calculated performance Breguet range considerations such as to
characteristics and test data for the provide an integrated boat-performance

410
200 30
Average Length/Breadth . 3.91
Average PressloadIng .117.63 psf
160

20
Ave age Reference Emil
120

1114111( 111011
ingimmanumm
1111111111111

pills,
80

ummagimm
10

40
1111111tio
o
2 8 10
o mosimilmmos
2 6 8 10
V /iTe
Figure 1 General Plot of Power Density vs Speed Ratio Figure 3 General Plot of Pressure Coefficient vs Speed Ratio

250 30
Data for craft under 100LT. only "Reference" from Figure 3
(Extrapolated) 110ft.Patrol Boat
200
....----.
Seafox
Gentry Eagle
20 thkAverage (93.15LT
150

MOM Gentry Eagle

(82.27LT)
100
10 Illbc
conelai HSB Harley
50
Pl?..1 1.111PPIIII._
'****1
56ft.Target Boat
200 400 600 800 2 4 6 8 10
Gross Weight (LT) V irCe
Figure 2 General Plot of Pressure Loading vs Gross Weight Figure 4 Pressure Coefficient vs Speed Ratio
 6000
200 Note: Data are for a
Coastal Patrol/Interdiction Craft (CPIC)
standard Lc/Bc = 3.91. For
= 100.00ft.
other Le/Be ratios apply
correction factor. /lie = 7.00 5000
LOA
BOA = 18.25ft
Le/Be = 4.87
150 = 91.32psf

111VOVIMI
p
4000
6.00 o
0.
.....iamefilIMI1111111111111 1111111 3000
. Test SHP
100 A Calculated SHP
1111111111111111111111MC 5.00
1111111.111111111111M 4.00 2000

50
11.11111111111111.111IIIIIIIIIIIIIIIIIIMEMNr-M1
IIIMIIIIIIIIMli 2.00 1000
Main Propulsion:
3-TF25A,2000HP GT, 7:1 Red.Gear
3-F.P Bronze Propellers
_ -
o o
60 70 80 90 100 110 120 15 20 25 30 35 40
Pressure Loading (psf) Speed (Knots)

Figure 5 Power Density vs Pressure Loading Figure 7 CPIC at GW=120,0001b.

ill
V/ile 15000
1.50
Gentry Eaglq
AV/ Le =2 5
o
1.25 V
aV
Le = 3
Le = 4 alliiiiiMM diiimm..111116
=6

=7
LOA = 109.90f t.
BOA = 24.00 ft.
Le/Be = 4.07

rd
=4 p = 97.04psf.

oil
EV Le = 5
o 1.00 0 V 10000
o Le = 6
V fEe = 7
Test SHP @ Fwd LCG
Test SHP @ Aft LCG
0.7 =3 o

lik Calculated SHP

C3-

0.5
a....e. =2
o
5000

o
o. 0.2
Wrilm."----Miiiiiiiiiii
2.0 2.5 3.0 3.5
iiiMili
4.0 4.5 5.0
o
15
Note: Waterjet propulsors

30
Speed (Knots)
45 60
Le/Be Ratio
Figure 8 "Gentry Eagle" at GW=82.27LT.
Figure 6 Le/Be Ratio Correction
 10000 1200
110ft. Patrol Boat Seafox
LOA = 110.00ft.
1000
LOA =
BOA =
39,50ft.
my
ill
8000 BOA = 21.08ft. - 9.75ft.
Le/Be = 4.64 Le/B - 3.24

II p
=-98.71 psf.
p =197.23psf

.41rili 800
P
1111111r
6000
o
o. . Calculated SHP .._ .
e...
co
600
4000

2000 A11411111
400

200
,..... ..
-# -- r eTTeesstt sSHHpP 2224xx3208 pPrroop

,...4 piNote: Testidraotiai inuildizeesrsdrag

_., Test SHP NR22x31 4)rop
Calculated SHP
I. . .
o
10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Speed (Knots) Speed (Knots)

I. Figure 9 110 ft. Patrol Boat at GW=329,2601b. Figure 11 "Seafox" at GW=24,6001b.

1200 2000
HSB Harley
56ft.Target Boat
LOA = 39.50 h. LOA = 54.73ft.
1000 BOA = 9.25 ft. BOA = 13.60ft.
Le/Be = 3.80 Le/Be = 3.58
p = 70.63 psf. 1500
p = 62.30 psf
800 ,....

600
Test SHP (LCG @ Mldpo nt)
Calculated SHP
A
Test SHP
Calculated SHP
111101
400 ---......!.1
500 1111
200
includes correctionfor ARNESON drive.
Note. Five engines and propellers
. o
10 20 30 40 50 10 15 20 25 30 35 40
Speed (Knots) Speed (Knots)
Figure 10 HSB "Harley" at GW=19,1171b. Figure 12 56ft. Target Boat at GW = 33,3901b.
model for expeditious sizing and Evaluation of Coastal Patrol and
optimization studies. However, it should Interdiction Craft (CPIC-X) (U) Part 1"
be noted that the material presented in
this paper is intended only to NAVSEACOMBATSYSENGSTA Report No.
contribute to the understanding of the 60-227 Dated February 1990. "Calm Water
principles related to high speed boat Performance Measurements on the Gentry
performance and, possibly, complement Eagle Trans-Atlantic Racer"
but not replace existing analytical
methods related to boat performance. NAVSEACOMBATSYSENGSTA Report No.
60-230 Dated July 1990 "Test and
Evaluation of Two Candidates for the
ACKNOWLEDGEMENTS 40ft. High Speed Boat (HSB) Design".

The author wishes to express NAVSEACOMBATSYSENGSTA Report No.

appreciation to Mr. Mark Hoggard of 60-108 Dated April 1983. "Development
NSWC, Carderock (Norfolk detachment) for Test and Evaluation of the 11 Meter
support of the work and permission to Special Warfare Craft Light: SEAFOX"
publish this paper and to Messrs Scott
Cummings, Mike Jones and Dave Rowland of
NSWC Carderock (Norfolk detachment) for
their critical review of the manuscript
and constructive suggestions. The author
also wishes to express appreciation to
the many boatbuilding companies which
supported the US Navy survey and, hence,
made this study possible.
Finally it should be noted that any
opinions expressed within the paper are
the opinions of the author alone and are
not necessarily shared by NKF
Engineering Inc., the sponsor of the
study or by the reviewers of the paper.

REFERENCE

1. "Industry Survey of Fast Patrol Boat

Builders" NSCSES Report No. 60-255
(Rev.A) dated September 1991.

2.NAVSEACOMBATSYSENGSTA Report No.

60-211 Dated September 1988. "USCoast
Guard 110ft. WPB Island Class "B" Lead
Ship Final Documentation. Trial Results
Lead Ship: ATTU (WPB-1317)"

NAVSEACOMBATSYSENGSTA Technical Note

No. 63-59 Dated August 1989. "Summary of
Propeller Trials"

NAVSEACOMBATSYSENGSTA Report No.

6660-018 Dated 03-76. "Technical

414
23RD ATTC GROUP PHOTOGRAPH, JUNE 12, 1992, AT UNO
 23RD ATTC BUSINESS MEETING MINUTES

The Business Meeting was called to order by Dr. Latorre, Chairman, at 3:15 p.m., Friday,
June 12, 1992.

Plans for 20th (1993) ITTC

Dr. R. Paulling presented a brief summary of the 20th ITTC which will be held in the
Berkeley-San Francisco area in 1993. The plans for the conference program are being
finalized and the arrangements will be detailed in the conference announcement.

Election of Representatives of North and South America on the ITTC Executive

Committee

Dr. Latorre requested Dr. Morgan and Dr. Murdey to come forward. Dr. Morgan
nominated Dr. David Murdey for the three-year period ending in 1996. The nomination was
seconded and Dr. Murdey was unanimously elected as the ITTC regional representative for
a three-year term ending 1996. Dr. Murdey thanked the members.

Location of the 24th (1995) ATTC

The invitation of the Institute of Pan American Naval Engineers (IPEN) was discussed. A
motion was made to hold the 24th ATTC in South America in a country to be decided. As
a back-up, the Offshore Technology Research Center, Texas A&M University, College
Station, Texas, and Webb Institute were named second and third alternate sites. It was
indicated that the ATTC executive committee would assist in setting up the technical
committees for the 24th ATTC and handling other conference matters. The motion was
seconded and unanimously approved.

There being no other business, the meeting was closed by Dr. Latorre at 3:30 p.m.

416
 AIMS AND ORGANIZATION
OF THE AMERICAN TOVVING TANK CONFERENCE

The American Towing Tank The conference assembles from time to

Conference is a voluntary association of time in different countries of the
establishments having a responsibility in American continents. Since 1950 it has
the prediction of hydromechanic assembled at three-year intervals.
performance of ships and other The details of organization of any
waterborne craft and their appendages particular conference are the responsibility
from tests on scaled models. of the host establishment.
The objective of the conference is to The conference is a pure
promote exchange of knowledge between communicative body; it has no authority of
tank staffs for the purpose of improving financial sponsorship; its membership is
methods and techniques. This includes an voluntary and self-supporting.
exchange of knowledge on the design of Each establishment may be represented
facilities, equipment, and instrumentation, by one or more members at triennial
on experimental and construction formal meetings of the conference.
techniques, and on scaling laws. As a Although representation is not limited,
means to this end, the conference seeks to it is the intent that it be kept reasonably
correlate testing among the various small so that fruitful discussions can be
member establishments in order to obtained at the working level. The
facilitate the interpretation of experiments conference intends to meet formally at
and to issue standards. each of the member establishments in
Membership in the conference is by rotation, the sequence being decided by
establishment and is open, upon invitation, the membership.
to all establishments in the Western The scope of the conference is set by
Hemisphere. All the member the Executive Committee, based on the
establishments agree to the free and full recommendations of the previous general
exchange of all information on the meetings.
foregoing subjects which is neither The Executive Committee shall
proprietary nor of a classified military comprise the nominated representatives of
nature. (To that end mutual agreements the institutions at which the last three
on exchange of publications and data of conferences were held plus the chairman
interest will be entered into.) of the SNAME Hydrodynamics
The members or delegates will be Committee, plus the representative of the
persons holding positions of primary institution at which the next conference is
responsibility in towing tanks or water to be held.
tunnels, shipbuilding research associations In the event that the chairman of the
or Departments of Naval Architecture of Hydrodynamics Committee of SNAME
a University in which prominence is given does not serve as a member of the
to the subjects pertaining to the objectives Executive Committee of the ATTC, the
of the conference. A few delegates can be Executive Committee may invite another
invited who are not qualified as above but official to serve on the Executive
who have rendered services to the aim of Committee.
the conference, as well as observers.
417
 If a member of the Executive the state-of-the-art reports and the
Committee resigns, a replacement shall be soliciting of papers to be presented to the
selected by his institution. conference.
In addition, the United States of The reports accepted by the staff of the
America's representative on the ITTC conference are discussed in the technical
Standing Committee shall, ex officio, be a sessions during the meetings, first formally
member of the ATTC Executive and then informally. To this end, those
Committee. reports are previously forwarded, well in
The ATTC membership of the advance of the dates of the conference, to
Standing Committee of the ITTC will be the chairman of the Conference
rotated. The representative will normally Organizing Committee, who arranges for
serve through two ITTC conferences. copies to be transmitted to each member.
The ATTC representative on the ITTC The same applies to the formal
Standing Committee shall: contributions which any member may
make to the conference. if the chairman
hold a senior position in a model of the Technical Committee considers that
basin; any contribution is unacceptable in view of
be acquainted with the operation of the objectives of the conference, he should
the ITTC; and reject it.
have means for financing The chairman of each Technical
attendance at an annual meeting of Session, in association with the chairman
the ITTC Executive Committee. of the Technical Committee concerned,
formulates decisions and recommendations
A newly appointed ATTC arising from the session. The decisions
representative on the ITTC Standing and recommendations will then be
Committee shall take office after the considered and agreed upon at the
ITTC Conference subsequent to the concluding session of the conference.
ATTC Conference at which he is The chairman of the Conference
appointed. Organizing Committee will arrange for the
The chairman of the Executive publication of the proceedings of the
Committee shall be the nominated conference.
representative of the institution at which The venue of the next conference is
the next conference is to be, viz, he is the subject to the invitation of an institution,
chairman of the next conference. and such invitation shall be accepted by
The secretary of the ATTC shall be the Executive Committee after a vote of
any person so nominated by the the delegates present at the business
conference chairman. meeting.
The Executive Committee shall appoint
chairmen of such technical committees as
are felt necessary for the effective conduct
of the conference.
The chairmen of such technical
committees will appoint such persons as
they consider necessary for the effective
conduct of such committees, and further
will be responsible for the production of
418
 PAST MEETINGS OF THE AMERICAN TOWING TANK CONFERENCE

1st ATTC 14-15 Apr 1938 Experimental Towing Tank, Hoboken NJ

2nd ATTC 19-20 Sep 1939 1st day, Receiving Ship Building Navy Yard, Washington D.C.
2nd day, David Taylor Model Basin, Washington, D.C.
3rd ATTC 14 Nov 1940 Waldorf Astoria, New York, NY
4th ATTC 14 Nov 1941 Waldorf Astoria, New York, NY
5th ATTC 29-30 Sep 1943 David Taylor Model Basin, Washington, D.C.
6th ATTC 12-13 Nov 1946 Experimental Towing Tank, Hoboken, NJ
7th ATTC 7-8 Oct 1947 Newport News Shipbuilding & Dry Dock Co., Newport News,
VA
8th ATTC 11-13 Oct 1948 University of Michigan, Ann Arbor, MI
9th ATTC 11-14 Sep 1950 National Research Council of C,anada, Ottawa, Canada
10th ATTC 4-6 May 1953 Massachusetts Institute of Technology, Cambridge, MA
llth ATFC 12-14- Sep 1956 David Taylor Model Basin, Washington, D.C.
12th ATTC 31 Aug-2 Sep 1959 University of California, Berkeley, CA
13th ATTC 5-7 Sep 1962 University of Michigan, Ann Arbor, MI
14th ATTC 9-11 Sep 1965 Webb Institute of Naval Architecture, Glen C,ove, NY
15th ATTC 25-28 Jun 1968 National Research Council of Canada, Ottawa, Canada
16th ATTC 9-13 Aug 1971 Instituto de Pesquisas Tecnologicas, Sao Paulo, Brazil
17th ATTC 18-20 Jun 1974 Navy Undersea C,enter, California Institute of Technolog,
Pasadena, CA
18th ATTC 23-25 Aug 1977 U.S. Naval Academy, Annapolis, MD
19th ATTC 9-11 Jul 1980 University of Michigan, Ann Arbor, MI
20th ATTC 2-4 Aug 1983 Davidson Laboratory, Hoboken, NJ
21st ATTC 5-7 Aug 1986 National Academy of Sciences, Washington, D.C.
22nd ATTC 8-11 Aug 1989 Institute for Marine Dynamics, St. John's, Newfoundland
23rd ATTC 11-12 Jun 1992 University of New Orleans, New Orleans, LA

419
 PROGRAM WAV El School of Naval Architecture
and Marine Engineering
THIS IS A WAVE FORM F POGR AM University of New Orleans
SC.01. Of MAVAL 4.1.01TECTIAI
MO NANO! 11101NLVIMO

New Orleans, Louisiana 70148

REAL A, C, X , Y, X0 K
.4111P (504) 286-7180
DO 10 K=1.0,10.0

A=(0.5/SQRT (K))- 0.000/

C A IS MA XIM,UM VALUE XO CAN RAYE

DO 20 X0=0, A, 0.01

C=K*CX0**2)

X= X0*( 1-2*C )

Y=X0-2,--(SQRT ( C-4*(C**2)))

WRITE(*,*) X9Y
20 CONTINUE

DO 30 X0=A, 0,-0. 01

C=K..*(X04 2)

X=X0*(1-24--C)

Y=-X0*(SQRT
WRITE(*,*) X, Y

30 CONTINUE

10 CONTINUE

END

-,
.,?1
1 a

:
,: ,-
/7
4 .,--- -,-
.
i105 -
1
, .,--
Ei. A. -
;
,
,¡;;.'7V-:-_;^:. "I- --r -_ - -'' - - 1

, ac,?.
...,.-.1. ....- - , i
, i

L -r------..V_,,,,- _: : k , i ,

_'--,,,,. k , l
-4Cii.:f:.'-;--,,i--,'c- ,"----- --...1.. ss
't
-
1i;-.."-§-..--:',:k--- ...1 - - - -

Fl -
\\.\
. r r 7
,
.

420