Feed-Pump Hydraulic Performance and Design Improvement, Phase I: J2esearch Program Design
Feed-Pump Hydraulic Performance and Design Improvement, Phase I: J2esearch Program Design
Feed-Pump Hydraulic Performance and Design Improvement, Phase I: J2esearch Program Design
Prepared by
Borg-Warner Corporation
(Byron Jackson Pump Division and
Borg-Warner Research Center)
Carson, California
and
Massa^ f Technology
Cambri__ . s
" I E CT R I C P 0 W E R R E S E A R C H I N ST ITO T E
DISCLAIMER
Prepared by
BORG-WARNER CORPORATION
17929 Adria Maru Lane
Carson, California 90746
Principal Investigators
W. H. Brown, Project Leader
8. Gopalakrishnan EPRI-CS~2332 Vol.
R. Fehlau
Principal Investigator
W. E. Thompson
Principal Investigator
D. G. Wilson
Prepared for
NOTICE
This report was prepared by the organization(s) named below as an account of work sponsored by the Electric
Power Research Institute, Inc (EPRI) Neither EPRI, members of EPRI, the organization(s) named below, nor any
person acting on behalf of any of them (a) makes any warranty, express or implied, with respect to the use of
any information, apparatus, method, or process disclosed in this report or that such use may not infringe private-
ly owned rights, or (b) assumes any liabilities with respect to the use of, or for damages resulting from the use
of, any information, apparatus, method, or process disclosed in this report
Prepared by
Borg-Warner Corporation
(Byron Jackson Pump Division and
Borg-Warner Research Center)
Carson, California
and
Massachusetts Institute of Technology
Cambridge, Massachusetts
ABSTRACT
ill
EPRI PERSPECTIVE
PROJECT DESCRIPTION
RP1884-6 is part of an EPRI program to improve the reliability of
feedwater pumps for large power-generating units. Problems of feed-
water pumps are a leading cause of unscheduled outages, and they are
estimated to have cost more than $408 million for replacement power
alone during 1981. The EPRI-sponsored program will be a multiyear
effort to provide an improved basis for the design, procurement,
operation, and testing of large feedwater pumps with increased reli-
ability and stability over the full range of operation. Phase I of
this effort was completed by two independent teamsj the result of one
team is reported here. The Phase I work was the preparation of a
research program design for the remainder of the overall EPRI effort,
including both theoretical and experimental research. The work was
based on surveys of architect-engineers, utilities, independent
research laboratories, and pump manufacturers as well as on the
resources of the team that performed the work.
v
feedwater pump problems, design methods, theory, and research
requirements.
PROJECT OBJECTIVES
The general objective of this project was to prepare a research plan
for a multiyear effort to improve feedwater pump reliability and
stability. The plan was to include resource and facility requirements
together with schedules and work descriptions.
PROJECT RESULTS
A comprehensive research program was formulated to develop a new
generation of advanced design feedwater pumps that would be highly
reliable and stable throughout their operating range. The proposed
research effort would require a five-year period and would cost
approximately $5.2 million (1981 dollars).
vi
entire program, including monitoring and troubleshooting, materials
selection, and code modification.
vxi
ACKNOWLEDGMENTS
IX
TABLE OF CONTENTS
Volume 1
Page
Section 1
Feed Pump Procurement and Design Methods
1,0 Byron Jackson/EPRI Survey of Architect/Engineers 1-1
2.0 Pro Forma Specification 1-9
3.0 Review of the Current State of the Art of Feed Pump 1-10
Performance Prediction
4.0 Review of Feed-Pump Research and Development and of 1-11
Manufacturers of Boiler-Feed Pumps
Section 2
Problems of Feed Pumps
1.0 Byron Jackson/EPRI Survey of Utilities 2-1
2.0 Major Outage-Producing Failure Causes Identified in 2-3
EPRI FP-754, "Survey of Feed Pump Outages"
3.0 Problems Identified in EPRI NP-1571, "Repetitive 2-3
Failure Causes for Feedwater Pumps"
Section 3
Assessment of Feed Pump Experimental Research
1.0 Rotordynamics Testing 3-1
2.0 Advanced Instrumentation in Hydraulics 3-6
3.0 Cavitation Damage Testing 3-9
4.0 Summary of Existing Pump Test Facilities, Research 3-14
Facilities, and Instrumentation
Section 4
Review of Pump Theory
xi
Page
4.0 Two-Phase Flow in Pumps 4-14
5.0 Impeller Forces 4-15
6.0 Flow Instability 4-22
Section 5
Feed Pump Research Requirements
1.0 Feed Pump Problems 5-1
2.0 Phase IIi Research Program 5-3
3.0 Phase Ills Development of Design Methods 5-9
Section 6
Phase III Research Program
HYDRAULICS
1.0 Off-Design Flow Studies 6-1
2.0 Velocity Field Measurements in a Pump Stage 6-3
3.0 Impeller/Volute Hydraulic Analysis 6-4
4.0 Hydraulic Instability 6-6
5.0 Analytical Study of Cavitation Performance 6-8
6.0 Analytical Study of Cavitation Damage 6-10
7.0 Correlation Between Noise and Cavitation Damage 6-12
8.0 Correlation Between Type of Cavitation and Damage Rate 6-13
9.0 Direct Measurement of Cavitation Damage in Impellers 6-14
ROTOR DYNAMICS
10.0 Development of Advanced Rotor-Bearing Systems for 6-16
Feed Water Pumps
11.0 Investigation of Axial Balancing Devices 6-17
12.0 Seal Ring Coefficients 6-18
13.0 Wear Ring Pressure Losses 6-20
PUMP SYSTEMS
14.0 System Interactions 6-21
15.0 Suction Effects on Pump Performance 6-23
Xll
Page
FULL-SCALE VERIFICATION
16.0 Full-scale Tests 6-24
17.0 Shaft Seals 6-29
TROUBLESHOOTING AIDS
30.0 Troubleshooting Manual 6-45
Section 7
Determination of Benefits
1.0 Objective 7-1
2.0 Approach 7-1
3.0 Research Program Benefit/Cost Ratio 7-2
4.0 Tabulation of Importance Factors 7-3
5,0 Estimation of Cost for Each Failure Mode 7-3
xiii
Pag
6,0 Weighted Benefit/Cost Ratio 7-6
7,0 Comments and Conclusions 7-6
Volume 2
Appendixes
Appendix
A Tabulated and Summarized Responses to Byron Jackson/EPRI A-1
Survey of Architect/Engineers
B Pro Forma Technical Specification, Turbine-Driven Boiler B-1
Feed Pumps for Large Fossil-Fired Power Plants
C Responses From Independent Research Laboratories and Pump C-1
Manufacturers Relating to Feed Pump Research and Develop-
ment and to Current Pump Designs
D Tabulated and Summarized Responses to Byron Jackson/EPRI D-1
Survey of Utilities
E Rotor Dynamics Testing E-1
F Advanced Instrumentation in Hydraulics F-1
G Cavitation Damage Testing G-1
H Summary of Existing Byron Jackson Test Facilities and H-1
Instrumentation
I Computer-Aided Literature Search Methodology I-l
J Hydraulics j-1
K Vibration in Pumps K-1
L Cavitation L-1
M Two-Phase Flow in Pumps M-1
N Impeller Forces N-1
0 Flow Instability 0-1
xiv
LIST OF FIGURES
Volume 1
Figure Page
3-1 Byron Jackson Test Rig 3--2
Volume 2
E-1 Byron Jackson Test Rig E-2
E-2 Air Test E-4
E~3 Water Test E-4
E-4 Critical Speed as a Function of Clearance for Smooth and E-5
Grooved Wear Rings
E-5 Test Rig Concepts E-6
E-6 University of Louisville Test Section Assembly E-8
XV
Figure Page
E-7 Experimental Apparatus for Annular Seal Dynamic Properties E-8
F-1 Reference Beam Mode F-3
F-2 Differential Doppler Mode F-3
F-3 Schematic of Dual-Beam Laser-Doppler Anemometer F-4
F-4 Details of Crossover Region F-5
F-5 Schematic Representation of the Photodetector Signal F-7
F-6 Radial Velocity Fields at Near Design Flow F-8
F-7 Radial Velocity Fields at Low Flow F-9
G-l Schematic of Beta Back-Scatter System G-2
G-2 Relationship Between Noise, Velocity, and Cavitation G-4
Erosion
G-3 Bubble Energy Spectra G-5
G-4 Cavitation Erosion Weight Loss and Spectrum Area G-6
H-1 Spectral Dynamics Model SD-2001D Digital Signal H-4
Processor
H-2 Vibration Amplitude vs. Frequency H-5
H-3 Order Tracking H-7
H-4 Example of Structural Resonance Excited by the Impeller H-8
Vane Frequency
H-5 Example of Discharge Pipe Acoustic Resonance Excited by H-9
the Impeller Vane Frequency
H-6 Campbell Diagram H-10
H-7 Campbell Diagram H-11
J-1 Reverse Flow and Recirculation in a Pump Impeller J-4
J-2 Static Pressure Distribution J-5
J-3 Comparison of Calculated Radial Forces With Test Data J-7
J-4 Flow Separation and Inlet Recirculation J-8
J-5 Reverse Impeller Discharge at 10% Flow Rate J-10
K-1 Rotor Vibratory Mode Shapes K-5
K-2 Pump Vibration Before and After Installation of K-7
Long-Taper Transition Pieces
xvi
Figure Page
L-1 Cavitation Conditions at Various Cavitation Levels L-2
L-2 Calculated Inception Point vs. Measured Inception for L-4
Standard Mixed-Flow Pump
L-3 Cavitation Damage Rate L-7
M-1 Two-Phase Flow Head Loss Function M-3
xvii
APPENDIX A
TABULATED AND SUMMARIZED RESPONSES TO BYRON JACKSON/EPRI SURVEY
OF ARCHITECT/ENGINEERS
TVA and AEP were mentioned as exceptions to the above. They may
build more 1,300-MWE plants. Many responders believe that the
trend is toward smaller (400-MWE to 500-MWE) plants.
2. What will be the maximum boiler feed pump size required for
fossil-fired power plants built in the next 10 years?
The above figures were given as representing the size of the feed
pumps used by TVA and AEP in their existing 1,300-MWE plants.
A-1
Most responders see very few requirements for pumps larger than
16,000 to 24,000 horsepower.
What will the ultimately required boiler feed pump sizes be?
Responder Response
a. 2-50% and 1-33%
b. 2-50% and 1-25% to 35%
c. 2-50% and 1-50%, limited to 8,000 to 9,000
horsepower
A-2
d. 2-50% or 2-50% and 1 smaller
e. 2-60% and 1-20%
f. 3-50% or 4-33%
What will be the range of sizes for most nuclear power plants
built in the next 10 years?
What will be the maximum feed pump size required for nuclear
power plants built in the next 10 years?
What will the ultimately required nuclear feed pump sizes be?
A-3
Minimum Flow Maximum Flow
Responder % of Design Flow % of Design
Flow
a. 25 160
b. 40 140
c. 50 130
d. 30 125
e. 25 115
f. 10 to 25 Depends on the
system.
Responses were the same as for question 10, except that responder
c. gave a range of 10% to 140% or 160%, accomplished with speed
control.
Responder Response
a. 2-50% and 1-33%, or 3-50%
b. 2-50% and 1-25% to 35%
c. 2-50% and 1-50%, limited to 8,000 to 9,000
horsepower
d. 3-50% o r 4 - 3 3 %
e. 2-50% (PWR), 3-50% (BWR)
f. 2-50%
A-4
There is generally a preference that the 20% to 33% capacity
motor-driven boiler feed pumps should take suction from the
deaerator and should not have booster pumps,
Two responders try to design the piping not to exceed the allow-
able nozzle loads quoted by the pump manufacturer. Four respond-
ers specify the nozzle loads to the pump manufacturer but, in
all cases, will work with the manufacturer and make some effort
to reduce loads that are greater than recommended.
A-5
elbows at a right angle to the plane of rotation of the first-
stage impeller unless extra straight-run pipe is provided. More
information on this subject would be welcomed by many responders.
Two responders said that even in nuclear plants, feed pumps are
not classed as "safety related" and are not subject to special
seismic analysis. Two responders said that seismic require-
ments apply to nuclear feed pumps, and one said that they apply
only to feedwater systems in PWR plants.
A-6
No responders are presently requiring dry run of boiler feed
pumps. One requires a severe low NPSH test, and others are
considering similar requirements.
a. At design point?
One will not give credit for any quoted efficiency above
85%. Several others would like to do something similar.
When efficiency evaluations are made and the client does not
A-7
specify evaluation at off-design flow rates, half of the re-
sponders evaluate off-design and half do not.
Responder Response
a. A formal analysis of EEI/NERC data is under
way.
b. We will probably do a formal reliability study
on the next job.
c. We evaluate EEI data, vendor installation list,
and vendor outage estimates.
d. Based on knowledge of industry specialists and
EEI information.
e. Based on EEI data and internal data.
f. Based on past experience, including field
experience.
A-8
Question Responder
a. b. c. d. e. f.
a.Maximum head per No No No No No Yes
stage?
b.Maximum shutoff Some Yes Yes No No Yes
head?
c.Maximum number of No Some No No No No
stages?
d.Maximum RPM? Yes No Yes Yes Yes Yes
e.Maximum suction No Yes No No No No
specific speed?
f.Other hydraulic No Yes Yes No No Inter-
design parameters? nal
clear-
ances
per
FP-754
Question Responder
a. b. c. d. e. f.
a.Vibration? Yes Yes Yes Yes No Yes
b.Radial shaft motion No Yes Yes Yes No Yes
(orbit)?
c.Axial shaft motion Nuc. Yes No Yes No No
(end float)? only
d.Discharge pressure No Yes No No No No
pulsations?
e.Suction pressure No Yes No No No No
pulsations?
f.Fluid-borne noise? No No No No No No
g.Airborne noise? No OSHA OSHA No No No
only only
h,Other? No Hot No No No Hot
align' align-
ment ment
check check
A-9
What instrumentation is needed to detect the onset of cavitation
in various parts of a pump?
Do you specify a certain type of shaft seal? (If so, please give
details.)
Responder Response
a. Throttle bushings, temperature controlled
b. Fixed-injection type, other temperature
controlled
c. Labyrinth type, temperature controlled
d. Packless leak-off type, pressure controlled
e. Fossil, no. Nuclear plants, bleed-off type
f. Generally not
Responder
a. b. c. d. e. f.
Test @ operating temp. Yes Yes Yes No Some Some
Test @ full speed Yes Yes Yes Yes Yes Yes
A-10
Lube system Yes - - -
Airborne noise No - - Yes
Fluid-borne noise Some _ - Some
Special Q.C, Yes - - -
Suction transient _ Yes - -
Seal leak rate - _ Yes -
Vibration _ _ - Yes
Post-test exam. - - - Yes
Loss of injection „
» — ^
36. What boiler feed pump field tests do you specify during plant
start-up and plant acceptance tests?
38. Are existing test codes adequate? (If not, please give details.
For example, what performance standards are needed for vibration,
shaft orbit, shaft end float, and pressure pulsation levels?)
There was, however, general agreement that the best way to re-
solve existing problems is by revision of the Hydraulic Test
A-11
Code, not by having different requirements for each A/E.
Three responders had not seen this publication, and two had
generally favorable remarks. One responder feels that the manu-
facturers are asked for too much information and that there is
too much specification of pump design.
Three responders had not seen this publication and three had
generally favorable comments.
Two responders had not seen this publication and three had
generally favorable comments. One had no comment.
A-12
APPENDIX B
PRO FORMA TECHNICAL SPECIFICATION
TURBINE-DRIVEN BOILER FEED PUMPS
FOR
LARGE FOSSIL-FIRED POWER PLANTS
1.0 SCOPE
2.0 GENERAL
The pump manufacturer shall be responsible for all items in the above
list that are covered by this section of the specification, except as
otherwise indicated in Table 1, "Project Requirements For Boiler Feed
Pumps. "
B-1
2. 2 Applicable Documents
The feed piimps will form a part of the feedwater system for the power
plant described in Table 1 under "Description of Service."
3-2
3.2 Design Conditions and Performance Requirements
3.2.1 The feed pumps shall be designed for the conditions specified
in Table 1 under "Design Conditions and Performance."
3.2.5 When one pump in the parallel operation trips, the remaining
feed purap(s) running at rated speed shall be capable of operating at
the intersection of the feedwater system resistance curve (attached).
This is known as pump runout condition.
3.2.9 Pump head curves shall rise continuously from design flow to
shutoff when determined at constant pump speed.
B-3
3.2.10 The pumps shall be identical in design. All replacement
parts shall be interchangeable.
3.2.11 Wetted pump parts shall be suitable for the water handled by
pumps. The chemical analysis of feedwater (condensate) is given in
Table 1.
B-4
bearing temperature or wear, high turbine-drive exhaust-hood pressure,
turbine-drive overspeed, or any other signal from equipment pro-
tection devices,
3.2.20 The pump shall be designed for frequent startup and shutdown
(either initially cyclic, or becoming cyclic when dictated by system
requirements). It is anticipated that the pump can be subjected to
daily start-up and shutdown, temperature changes, and frequent daily
load cycles.
3.3 Materials
Materials for the feed pumps shall be as specified below. The pump
manufacturer may propose alternative materials that in his experience
are more suited to the specified service. The alternative materials
shall be subject to the owner's approval. When two or more materials
are specified for a component or part, the pump manufacturer shall
choose among them.
B-6
3.4.3 Rotation. Pumps shall rotate clockwise when viewed from the
pump coupling end.
Areas around the discharge, and areas at the seals that are in
contact with the feed water, shall be surfaced, per paragraph 3.3,
with an overlay of a metal alloy to resist both corrosion and the
erosion resulting from high fluid velocity and turbulence.
B-7
prevent oil leakage. The housings shall be of steel and shall
provide for easy bearing inspection.
3.4.7 Stuffing Boxes and Injection Seal System. Pump shafts shall
be provided with injection-type throttle bushing stuffing boxes to
meet the conditions caused by shaft peripheral speed and pump suction
temperature and pressure. Seal drains will be to the condenser.
The complete shaft seal control system (except filter) shall be assem-
bled by the pump manufacturer.
All piping and conventional shut-off valves for connection from the
condensate discharge to the feed pumps and leak-off systems will be
provided by others.
B-8
The system furnished shall include all controls necessary for automat-
ic operation.
The pump seal design shall be such as to permit the pumping unit and
turbine to coast to a stop safely immediately following emergency
plant trip-out when no injection water is available.
3.4.8 Nozzle Loads, The pump manufacturer shall state the allowable
forces and moments at the face of each nozzle connection (suction
and discharge) on the feed pump.
B-9
unless otherwise specified in Table 1 for specific connections or
where necessary for assembly and dismantling of the pumping units.
3.4.12 Thermal Insulation and Metal Lagging. Each pump barrel shall
be provided by the pump manufacturer with thermal insulating
material completely enclosed in a rigid steel jacket and fastened to
the barrel. Insulation thickness shall limit surface temperature to
120° F with normal ambient temperature (see Table 1 under "Design
Conditions and Performance") when operating at pump design tempera-
ture. The insulation and jacket shall be arranged to allow for pump
disassembly without damage to the insulation or jacket. When the
pump installation utilizes concrete pedestals, the owner will
provide and install thermal insulation and lagging in the field.
B-10
bottom of the outer casing. The connections shall be provided with
double valves if pressure exceeds pump suction pressure during any
operating condition. The location of connections and valves shall be
such that the valves will be easily accessible for operation and
maintenance. The make and type of all valves shall be approved by
the owner,
3.5.5 Oil for lubrication of feed pumps will be taken from the
auxiliary turbine driver oil system. The pump manufacturer shall be
responsible for coordinating the lubrication system requirements of
the pump with the auxiliary turbine driver manufacturer.
B-11
Oil piping requirements are specified in paragraph 3.4.10.
The pump manufacturer shall provide one complete set of all the
necessary special tools required for installation and maintenance of
the feed pumps and accessories, including a dismantling carriage
designed to support the pump inner assembly during dismantling.
The pump manufacturer shall provide one complete inner assembly con-
sisting of pump shaft, impellers, diffusers, and other rotating and
stationary parts subject to wear, including all required keys, pins,
screws, fasteners, gaskets, and seals.
The pump manufacturer shall also provide one complete set of journal
bearings and one set of thrust shoes for thrust bearings.
B-12
Abrasive blasting shall not be used on surfaces containing areas on
which grit can accumulate and not be readily removed by subsequent
cleaning, rinsing, or flushing operations.
B-13
as are necessary to insure that the work conforms to this specifica-
tion.
Whenever size permits, the pump and any accessories shall be packed
securely in weatherproof crates. If the pump or any component is too
large for packing in a crate, it shall be placed on skids and
protected with an overwrap.
3.12 Marking
a. Manufacturer's name
b. Manufacturer's model number
c. Manufacturer's shop order number or serial number
d. Owner's equipment tag number
e. Service
f. Design capacity, gpm
g. Design temperature, F
B-14
h. Design head, ft
i. Design speed, rpm
j. Suction pressure, psig
k. Discharge pressure, psig
4.0 INSTALLATION
B-15
5.0 TESTING
The inner casing (including spare) and all crossover passages shall
be hydrostatically tested in accordance with the pump manufacturer's
standard procedure.
If the owner reserves the right to witness the hydrostatic tests, the
pump manufacturer, as indicated in Table 1, shall notify the owner of
the tests at least three days in advance of the time of the tests.
Test readings shall be made for at least eight points over the range
between minimum flow to pump runout flow.
B-16
If a speed lower than the rated speed is used during testing, the
efficiency shall not be corrected when extrapolating the performance
to plant operating conditions.
Br-17
Each piamp shall be dismantled and inspected after the performance
test is completed. Any clearance adjustment or any dimensional
change performed after the test shall require a retest of the pump,
unless the retest is waived in writing by the owner.
B-18
charge head and flow (if applicable), required mini-
mum flow, and runout flow plotted versus discharge
capacity in gpm for each 200-rpm increment in speed.
b. Detailed procedure for determining interstage bleed
discharge heads and flows (only if applicable) at
any pump operating condition.
c. Test procedures, calibration data for instruments and
equipment, and complete set of test calculations.
d. Efficiency adjustment calculations from a cold test to
hot-water operating conditions.
e. Verification of interchangeability tests for spare
internal assemblies in each pump.
f. Verification of journal bearing stability, hydraulic
stability at mimimum flow, and radial and axial
vibration limits.
B-19
6.0 INFORMATION TO BE SUBMITTED
The pump manufacturer shall submit information with the bid in suffi-
cient detail to permit the owner to fully understand and evaluate
the equipment being offered.
The pump manufacturer shall complete the Proposal Bid Data Sheets
included in the Bid Form of this specification for submittal with
the bid. After acceptance by the owner, the data will become part
of this specification for the feed pumps and accessories.
The information and data below shall be submitted to the owner, for
approval or for information, in accordance with the timing specified
under "Information to be Submitted" in Table 1.
B-20
e. Installation drawings that may be required for
the complete equipment installation.
f. Injection seal system drawings.
g. Complete pump performance curves.
h. Torque-speed2curves from zero to rated pump speed,
and rotor WR .
i. Pump warm-up orifice drawings.
j. Nozzle load drawings.
For Information.
B-21
TABLE 1, Sheet 1
PROJECT REQUIREMENTS FOR BOILER FEED PUMP (TURBINE DRIVEN)
SPEC.
PARA. ITEM
Not
2.1 Work to be Provided Req'd Req'd
Interstage Bleed
2.2 State and Local Codes
B-22
TABLE 1, Sheet 2
PROJECT REQUIREMENTS FOR BOILER FEED PUMPS (TURBINE DRIVEN)
SPEC.
PARA. ITEM
•XT- Feedwater (or Condensate) Analysis:
Total Solids ppb
Dissolved Oxygen cc/LTR
Silica as SiOo
Iron as Fe ppb
Copper as Cu - ppb
Nickel ppb
Hydrazine (Residual) ppm
Sodium ppb
Conductivity (cation) 25°C
pH @25°C to
Injection Sealing Water From Condensate Pump
System Discharge:
Max. Pressure, psig
Min. Pressure, psig
Normal Pressure, psig
Max. Temperature, °F
Normal Temperature, °F
Performance Tests
NPSH Test
Hydrotests
B-23
TABLE 1, Sheet 3
PROJECT REQUIREMENTS FOR BOILER FEED PUMPS (TURBINE DRIVEN)
j SPEC.
PARA. ITEM
6. Information To Be Submitted Days After Award
Overall Schedule 30
Document Submittal Schedule 30
Equipment Delivery Schedule 30
Outline Drawing 60
Welding and Repair Welding Procedures 60
Foundation Drawing T,
Lubrication Piping Drawing 60
Assembly Piping Drawings 60
Injection Seal Drawings 60
Pump Performance With bid
Torque Speed Curves With bid
Warm-Up Orifice Drawing 60
Nozzle Loading Drawings With bid
B-24
TABLE 1, Sheet 4
PROJECT REQUIREMENTS FOR BOILER FEED PUMPS (TURBINE DRIVEN)(EXAMPLE)
Spec. Item
Para.
B-25
TABLE 1, Sheet 5
PROJECT REQUIREMENTS FOR BOILER FEED PUMPS (TURBINE DRIVEN)(EXAMPLE)
Spec,
Para. Item
B-26
APPENDIX B
ADDENDUM 1
DEFINITION OF TERMS
The following abbreviations are used throughout this specification
for the convenience of the reader:
B-2 7
APPENDIX B
ADDENDUM 2
B-2e
APPENDIX B
ADDENDUM 2
Seller
RECIRCULATION DATA
1 Type of seal
2 Diametral clearance at seal, in
(if applicable)
3 Injection water flow required per seal
gpm ...,...,,
4 Injection water temperature, pressure
°F/psig .....
5 Maximum particle size in injection
water
6 Leak-off flow per seal, gpm .......•,.
7 Leak-off temperature, °F
8 Temperature or pressure control ......
9 Single or double injection
B-29
APPENDIX B
ADDENDUM 2
B I D F O R M - TECHNICAL
Seller
B-30
APPENDIX B
ADDENDUM 2
BID FORM - TECHNICAL
Seller
2.6.5 Suction connection:
2.6.5.1 Size/schedule _
2.6.5.2 Weld end details .......... .
2.6.5.3 Location top/bottom .......... .....
2.6.6 Discharge connection:
2.6.6.1 OD _
2.6.6.2 Weld end details ....... *
2.6.6.3 Location top/bottom *
2.6.7 Warm-up connections:
2.6.7.1 Size
2.6.7.2 Quantity and location
2.6.7.3 Warm-up flow, gpm
2.6.8 Superheat attemperator connection:
2.6-8.1 Size and rating ... (if applicable) . /_
2.6.8.2 Flange details .....................
2.6.9 Reheat attemperator connection:
(if applicable)
2.6.9.1 Size and rating /_
2.6.9.2 Flange details ................... *
2.6.10 Balancing device leak-off connection:
(if applicable)
2.6.10.1 Type
2.6.10.2 S i z e - schedule /
2.6.11 Direction of rotation from driver end,
CW or CCW . .****
2.6.12 Developed head per stage at best
efficiency point:
2.6.12.1 First stage, ft
2.6.12.2 Succeeding stages, ft
2.6.13.1 Shaft diameter at bearings (nom.) in.
2.6.13.2 Shaft diameter at impeller (nom.), in.
2.6.13.3 Shrink fit at impellers, mils .....
2.6.13.6 Shrink fit at bal.disk/drum, mils ..
2.6.14 Shaft static deflection (at rest), mils __
2.6.15 Shaft span between bearing centers, in. _
2.6.16 L/D ratio at bearing (L=brg. length, in.,
D=shaft dia., in. ) ...........
2.6.17.1 Journal bearings - type
2.6.17.2 Journal bearings - size ______
2.6.17.3 Journal bearing length, in.
2.6.17.4 Journal bearing clearance, mils ...
Specified clockwise
B-31
APPENDIX B
ADDENDUM 2
BIDFORM - TECHNICAL
Seller
2.6.18.1 Thrust bearing type
2.6.18.2 Thrust bearing Size ^
2.6.19 Percent of axial thrust carried by the
thrust bearing, % .
2.5.20 At what loads is thrust brg. loaded?
2.6.21 Type of axial balancing device
Cdrum, disk, opposed impeller) .....
2.6.22 Balancing device leak-off flow, gpm
(min/max)
2.6.23 Can balancing device handle axial
thrust at all loads and conditions?
2.6.24 Method of indicating shaft axial
position (displacement)
2.6.25 Balancing device radial clearance
(.if opposed impellers, give center
bushing)
2.6.26 Pump designed with diffuser or volute?
2.6.27 Diameter, clearances at wear ring -
first stage, mils
2.6.28 Diameter clearances at wear ring -
series stages, mils
2.6.29 Diameter clearances at impeller hub, mils
2.6.30 Impeller eye area first stage, sq.in.
2.6.31 Number of impeller vanes first stage
2.6.32 Number of impeller vanes normal stages '_
2.6.33 Impeller vane mounting arrangement
between impellers (in-line, staggered)
2. 6. 34 Permissible maximum unbalance in any
plane, oz. in
2. 6. 35 First critical speed of rotor, rpm
2. 6. 36 Second critical speed of rotor, rpm
2. 6. 37 Foundation:
2. 6. 37. 1 Stiffness vertical, lb/in ^
2. 6. 37. 2 Stiffness horizontal, lb/in ^
2. 6. 38 Bearing lubrication:
2. 6. 38, 1 Type.
2. 6. 38. 2 Required flow per journal bearing,
gpm. .... ........
2. 6. 3 8 . 3 Required flow to thrust bearing,
gpm • -
2. 6, 38. 4 Total oil flow, gpm ..............
2. 6. 38. 5 Oil type recommended '_
2. 6. 38. 6 Heat to be removed from oil,BTU/hr
2. 6. 38. 7 Bearing temperature (recom/max), °F
2. 6. 38. 8 Oil pressure required (min/recom/max),
psig
B-32
APPENDIX B
ADDENDUM 2
BID FORM - TECHNICAL
Seller
2.6.39 Pump shaft at coupling:
2.6.39.1 Diameter, in __
2.6.39.2 Type of fit .
2.6.39.3 Space between shafts, in
2.6.39.4 End float required, in (motor
drive only
2.6.41 Space envelope required to remove
rotating element, in. ........
2.6.42 Rotor removal cradle (number
required
2.6.43 Rotor handling device (number
required)
2.6.44 End covers for pump flushing, when
internals are removed (each
pump)
2.7 PUMP MATERIALS (ASTM DESIGNATION)
2. 7. 1 Outer casing
2.7.2 Discharge head (barrel end
head) _____
2.7.3 Inner casing
2.7.4 Diffusers or volutes
2.7.5 Casing rings
2. 7. 6 Impellers
2.7.7 Impeller wear rings ( if applicable)
2.7.8 Pump shaft
2.7.9 Shaft sleeves (if applicable)..
2.7.10 Rotating balancing disk/drum..
2.7.11 Stationary balancing disk/driim
or centerpiece
2.7.12 Internal gaskets
2.7.13 Interstage packing (if applicable)
2.7.14 Recirculation breakdown orifice
iif applicable) ..............
2.7.15 Casing bolts, studs, nuts (inner
and outer casing).............
2.7.16 Pump suction nozzle ..........
2.7.17 Pump discharge nozzle
NOTE: Refer to paragraph 3.3 for specified materials.
B-33
APPENDIX B
ADDENDUM 2
Seller
2.8 WEIGHTS (LB)
2.8.1 Pump (empty without water)....
2.8.2 Foundation type (cast, fabricated,
etc. )
2.8.3 Foundation weight
2.8.4 Casing
2.8.5 Inner assembly (spare element).
2.8.6 Rotor (all rotating parts without
coupling half)
2.8.17 Largest weight to be handled
(individual piece or assembly)
during service
B-34
APPENDIX C
RESPONSES FROM INDEPENDENT RESEARCH LABORATORIES AND PUMP
MANUFACTURERS RELATING TO FEED-PUMP RESEARCH AND DEVELOPMENT AND
TO CURRENT PUMP DESIGNS
2.0 CREARE
Two responses were received from Penn State: one by Dr. Blaine R.
Parkin, director of the Garfield Thomas water tunnel, and Professor
C'-2
Robert D. Henderson of the same facility, and a second by Dr. B.
Lakshminarayana of the fluid-mechanics group. The Applied Research
Laboratory is an organization and a group of facilities in which
faculty, graduate students, and a substantial research staff perform
research work, 90 percent of which is for the U.S. government. The
water tunnel itself is old but impressive, used mainly for the
testing of torpedoes, model submarines, and propellers. There are
many other loops available for tests in water and other liquids,
extensive data-acquistion systems, and extensive instrumentation. An
impressive feature of the visit was the wide variety of activities
being carried on under the direction of highly regarded people.
There is also some element of competition in that, for instance, two
groups were examining the computational fluid mechanics of pump and
compressor rotors with different approaches. Although there was no
work specifically being undertaken in centrifugal pumps, such work
could be carried out in the facilities available.
7.0 INGERSOLL-RAND
C'-3
8.0 KSB
Dr. Robert Dernedde wrote that KSB declined to respond to the ques-
tionnaire, but subsequently he, with Walter Wilhelm, a representative
of KSB in this country, visited M.I.T. for a discussion.
Dr. Dernedde was cordial and stated that he was somewhat embarrassed
at not returning the questionnaire, but it was the view of KSB that
the monetary exchange rate, coupled with the high quality built into
KSB pumps, put the price above that at which any sales could be
contemplated in the United States. Therefore, to complete the
questionnaire would simply be potentially to give away information.
He said that one point in efficiency of the pump would, during its
life, be equivalent to an energy cost equal to the new cost of the
pump. Therefore pump quality was important.
9.0 SULZER
C-4
dynamics was still necessary, however.
A visit was made to this location, and a full response was received
from Dr. S, K. (Jimmy) Chow. This division makes nuclear pumps for
submarines and other naval vessels. It has about thirteen test loops
for pumps of various sizes. Most are used for acceptance tests
rather than for research. However, there is a cavitation test loop,
for visual observation of cavitation, and an air-test rig. Much of
the experimental development work is for the liquid-metal fast-
breeder reactor. Rotordynamics and bearing analysis are taken
to considerable depth. They have laser-doppler anemometers (the
reviewer did not see them applied to pump flow) and would like to
have the capability of measuring radial loads.
11.0 WORTHINGTON
C~5
APPENDIX D
TABULATED AND SUMMARIZED RESPONSES TO BYRON JACKSON/EPRI
SURVEY OF UTILITIES
1. Name the major problem with boiler feed pumps in large (600 MWE
or larger) fossil-fired plants. If you had to name one problem
only, what would it be? Please state both the symptom and the
root cause.
2. What other problems exist? Please state both the symptoms and
the root causes. List in order of seriousness.
D-1
b. 1. First-stage impeller cavitation
2. Volute-to-barrel fit gasket washout on nonmodified
pumps
3. Gear-type couplings not sufficiently flexible
4. Shaft breakage
5. Hot alignment, especially with pumps having
long horizontal pipe runs
c. 1. Volute/diffuser impeller washout
2. Seizures occurring mostly at start-up
3. Seals
4. Balancing devices
5. Vibration
d. 1. Water intrusion into bearings due to seal failures
2. Seizures due to change of materials by vendor
3. Leakage at gasket pump element to barrel
4. Washout at volute faces on some early units
5. Shaft breakage
6. Equalizing pipe breakage
e. 1. Loss of performance, vibration. Failure of inter-
stage partitions
2. Loss of performance. Failure of the suction-to-
discharge joint between the inner case and the
outer casing
3. Failure of piping'—instrumentation and lub€;-oil
piping attached to the pump case. Vibration (align-
ment, coupling/pump imbalance
4. Loss of performance. Lack of fasteners on the
suction and wall of the inner casing
5. Vibration and loss of performance. Cracking of
impeller vanes both on inlet and outlet. (Not
due to cavitation damage.)
List in order of frequency:
a. 1, hot alignment
2. Cavitation
3. Gear-type coupling
4. Volute fit washout
5. Shaft breakage
b. 1. Bearing lubrication problem due to cooling-water
fouling
2. Pxjmp s e i z u r e s
3. Seals
4. Performance degradation due to volute/diffuser
washout
5. Vibration, mostly after start-up after service,
indicating hot misalignment
6. Thrust balancing devices
c. 1. Balancing devices
2. Seal water control problems
D~2
d. We do not have the frequency data requested. We started
our supercritical units between 1966 and 1973. This was
early in the high-energy pump game and we had pump prob-
lems. However, we worked with the pump manufacturer and
now feel that we have solutions for these problems.
e. No comment.
How many boiler feed pump problems and/or failures have you had
during each of the past six years. Please classify as to time of
occurrence.
D-3
b. Total Within 90 days Within 90 days Midlife
failures after start-up before sched- problem or
of new or re- uled pump failure
paired pump maintenance
1980 4 3 0 1
1979 10 4
1978 18 9
1977 15 Unknown
1976 3
1975 1 -
c. Between 1972 and 1978 there have been 129 pump failures,
2 5 of which were seizures.
d. Total Within 90 days Within 90 days Midlife
failures after startup before sched- problem or
of new or re- uled pump failure
paired pump maintenance
1980 9(19)
1979 10(34)
1978 8(23)
1977 3(15)
1976 8(22)
( ) Includes partial shutdowns
1. Separate from pump outages, a total of 8 6 B.F.P.
drive problems were experienced, 16 of which re-
quired full shutdown
2. Information not available in this form
We do not have the data arranged in the requested format.
D-4
c. Improved specifications. Improved minimum flow system,
including increased minimuir- flow, as initial specified
flows by vendors were too low.
d. Rerouting balance disc leak-off piping to pump suction.
Monitor disc leak rate to observe its condition. Im-
proved temperature control of seals.
e. These actions include installation of very heavy-duty
auxiliary piping, installation of more modern impeller
design when replacing impellers (after about 10 years of
service on the old-design impellers), stronger interstage
partitions to overcome what we believe was a water hammer
problem, modification of the high-pressure internal joint
to prevent washout, and modified thrust collar mounting.
D-5
a. We do not have the statistics to analyze and report fail-
ures in nuclear feed pumps in a systematic fashion. Our
reliability and design group has been established to an-
alyze and classify forced outages, deratings, and ex-
cessive maintenance of pumps at nuclear plants.
What other problems exist? Please state both the symptoms and
the root causes. List in order of seriousness.
D-5
How many nuclear feed pump problems and/or failures have you had
during each of the past six years. Please classify as to time
of occurrence.
D-7
b. Nothing to report
c. None
d. No comment
e. No comment
Parameter Instrument
Discharge pressure Local gage plus control room
indications
Suction pressure Local gage
Discharge pressure No
fluctuations
Suction pressure No
fluctuations
Pump flow Indicated and recorded in con-
trol room
Input power Ammeter plus watt-hour meter
Vibration at nozzles No
Vibration at bearings See question 16
Vibration at No
Axial shaft motion No
Radial shaft motion No
Airborne noise Now reviewing per OSHA standards
Fluid-borne noise No
Water temperature In and out, recorded in control
room
Parameter Instrument
Discharge pressure Gage
Suction pressure Gage
Discharge pressure
fluctuations
Suction pressure
fluctuations
Pump flow Indicator at pump and in con-
trol room
Input power
Vibration at nozzles
Vibration at bearings Yes
Vibration at
D-8
Parameter Instrument
Discharge pressure Yes
Suction pressure Yes
Discharge pressure No
fluctuations
Suction pressure Yes
fluctuations
Pump flow Yes
Input power Yes
Vibration at nozzles Yes
Vibration at bearings Yes
Vibration at -
Axial shaft motion No
Radial shaft motion On newer pumps only
Airborne noise No
Fluid-borne noise No
Water temperature Yes
RPM Yes
Parameter Instrument
Discharge pressure Yes
Suction pressure Yes
Discharge pressure No
fluctuations
Suction pressure No
fluctuations
Pump flow Ye s (only minimum flow)
Input power No (turbine drivers)
Vibration at nozzles No
Vibration at bearings No
Vibration at No
Axial shaft motion No
Radial shaft motion No
Airborne noise No
Fluid-borne noise No
Water temperature Ye
RPM Ye
Parameter Instrument
Discharge pressure Yes
Suction pressure Yes
Discharge pressure No
fluctuations
Suction pressure No
fluctuations
Pump flow Yes
Input power No
Vibration at nozzles No
Vibration at bearings No
Vibration at -
Axial shaft motion No
Radial shaft motion No
Fluid-borne noise No
Water temperature Yes
D-9
What other instrumentation has been used for diagnosis of prob-
lems?
d. No comment.
e. Do not know.
D-10
and we try to do pump repairs during these scheduled out-
ages. The swing load plants are somewhat less critical,
and we may take a pump out of service for repairs.
d. Yearly check of bearings, seals, and balancing device
only.
e. At turbine inspection shutdown if indication of trouble.
Otherwise service only when failed.
c. No comment
d. No comment
e. No comment
D-11
c. We have not seen this report.
d. No comment
e. No coiranent
21. Do you have any comments on EPRI FP-754, "Survey of Feed Pump
Outages"?
a. Informative
b. The data were good; especially, the discussion of root
causes was valuable. This was a good report from the
standpoint of information as reference for maintenance
people.
c. Quite good. Some recommendations should be included in
specifications.
d. We cannot directly correlate the findings of this report
to our experience.
e. No comment
D-12
APPENDIX E
ROTOR DYNAMICS TESTING
A great deal of work has been done during the last few years to
understand these characteristics both analytically and experimen-
tally. Offered here is an overview of some of the experimental
work that has been done recently to demonstrate and to identify
these effects. So far, experimental work can be broadly classified
in two major categories. In the first the shaft is allowed to
undergo whirling motion. In the second it is made very stiff, and
the forces in the wear rings, etc., are directly measured. The
advantage of the first type of test is that one can immediately
detect the presence of a critical speed, which is a matter of great
practical significance. The difficulty associated with this type of
test is that understanding why critical speed occurs is dependent
upon the modeling made for the entire system. Consequently, there is
no direct substantiation of the effect of the annular clearances. In
the second set of tests, which uses a stiff shaft, the forces are
directly measured and, therefore, the sealing ring properties are
immediately and directly established.
E-1
PRESSURE GAUGE
PROXIMITY PROBE
E-2
quency electric drive. The pressure difference across the wear rings
was changed independently of the speed, and the vibration of the shaft
was measured by using two proximity probes. When the tests were first
conducted in air, the critical speed was clearly observed as shown in
Figure E-2, which plots the once-per-revolution amplitude versus the
speed. It also shows that the phase shift of nearly 180° is observed
as well. When water was introduced into the pump, the amplitude
response characteristics changed dramatically. For example. Figure E-3
shows that when smooth wear rings were used with a nominal clearance
of .016 inch (diametral clearance) and there was a pressure difference
across the rings of 250 psi, no discernable critical speed was ev-
ident; also there was scarcely any phase shift. In this case, the
critical speed appears to be totally suppressed; the fact that it is
can be easily be demonstrated by calculations. A paper (1) to be
published in the next year's ASME meeting in London contains these
calculations. However, when the wear ring clearance is opened up or
when the wear ring is deeply serrated, critical speed phenomena begin
to appear again. It is also interesting to note that the effect of
wear on the rings is to reduce the critical speed, particularly for
rings that are smooth; this can be seen in Figure E-4. It shows that
as the diametral clearance of the seal rings increases, the stiff-
ening effect is reduced and, therefore, the critical speed decreases.
The effect, however, is much weaker for grooved rings, because they
produce less stiffness.
E-3
DISPLACEMENT PHASE LAG DISPLACEMENT PHASE LAG
CyiLS. P K - P K ) (DEGREES) (MILS PK~PK) (DEGREES)
; 1
' 1'! i1
' •! 1
"i ..II
•nti !,•• -!^
.), I TT—
•.:u 1 • 1'!
-t" . .'.
= ^ if—
H
tn
a o
G
[____.
3 M
I
M 33
e -4+*^ r-.t, 1
I
©, : ~
s:
ri- H- O,
t-i
H
1^
: -
— Z2JL1. tn
cn -^ ri-
rt
iSi= " r
^ 7 }
^-^-t - *: plh
•J=^:zt±
*' 1' •lit
rrA^r .-; .
••ijT*-r;L-. ' ' ' t
?9@8
ises \
sses S,r,
s
mm \
- ^ \ s ^
4I0B ^^SB«^-€J
^*«%,Jg
^ ^ ^ ^ ^
-
4ooa U—»-
..., — _ _
ssso HMMiMMBlLWMIUJHBaHUUMj
42 ^3 J«
D«
glAHETML @LIAMN£E
E-5
FIGURE E-5. Test Rig Concepts
E-6
In the second type of experiment the journal is offset and, at the
same time, is rotated to make a centered circular orbit. (See Figure
E-5.) These are the types of tests done by Childs (12_, 1_3) at Louis-
ville. (See Figure E-6.) Childs determined the forces by measuring
pressure along the axis and around the circumference. He measured
five coefficients, because he assumed that all the matrices are
symmetrical and the diagonal elements equal and opposite. Further,
he assumed that the cross-coupled inertia terms are zero. Therefore,
the total of twelve coefficients is reduced to five. It is noted
also that in this particular setup the whirling was at synchronous
speed. In the third type of test the journal is offset and then
vibrated translationally. (See Figure E-5.) This type of testing
is being done by Childs now at Texas A&M. Initial results will be
forthcoming soon. Similar tests have been done at Hitachi in Japan
by lino and Kaneko (1_4) . (See Figure E-7.) With this kind of testing
all twelve coefficients can be measured. Further, the conicity of the
axis can be included, and by doing static tests it is possible to
determine the pressure loss coefficients and the friction loss
coefficients. Finally, in the fourth type of stiff-shaft testing the
journal is set with an offset and then vibrated by orbiting it
accurately. In this case it is possible to vary the precession speed
with respect to the rotating speed. (See Figure E-5.) This kind of
testing is being done by M. L. Adams and E. Makay (15_, 15.) • All twelve
coefficients can be determined accurately in this setup.
The recent theoretical work done by Childs has shown that it will be
possible to determine the sealing ring coefficients by means of
theoretical calculations. In the long run it will be important to be
able to do this type of calculations. Childs has pointed out that for
them to be accurate, the primary requirement is a full understanding
of the pressure loss coefficients. They are threefold. One is the
entry loss coefficient, both at the first entrance to the ring and
between the groove and the land for the grooved base. The second set
is caused by the exit loss, or the discharge loss coefficient. The
third set is the friction coefficient caused by the flow of fluid on
the surface. Evidently the Reynolds numbers both for axial flow and
circumferential flow play a part. Whether the vibration itself will
have an effect on the pressure loss coefficients is a moot point.
However, it will probably be important in the near future to run tests
E-7
prob® th®r!?s©c!©upl®
prsssur®
fransducsr
DISS=LACEIVIENT SENSOR
WATER
AXIAL THRUST
HYDROSTATIC JCHJRNAL ^'^'^
BALANCING DISK
jn -™_--H"«'DROSTATIC JOURNAL
0"i Jm _ BEARING
V!8RAT1CW
MECHAP«CAL
SEAL
HYDRAUUC
VIBRATOR
AXIAL THRUST
BALANCING SHEET
UNIT IN mm
E-8
statically to determine all of them and to use them in conjunction
with the theory to predict the coefficients of these rings. These
predictions then can be compared with accurate test data taken by
Adams and Makay or by Childs.
REFERENCES
3. H.F. Black and J.L. Murray, "The Hydrostatic and Hybrid Bearing
Properties of Annular Pressure Seals in Centrifugal Pumps,"
BHRA Report RR1026, October 19 69.
E-9
Federal Republic, Feb. 27-March 2, 1973.
E-10
APPENDIX F
ADVANCED INSTRUMENTATION IN HYDRAULICS
Two types of laser systems are used for determining the velocity
field. The first is called the laser two-focus system. It is quite
popular in Europe and has been used to determine the velocity field
in advanced high-performance centrifugal compressors. In this method
two laser beams are focused in a very small volume. As a particle
travels between the two light beams, it produces two pulses of scat-
tered light. The elapsed time between the pulses and the known beam
separation yields the velocity component perpendicular to the optical
axis. The principle of this method is quite simple. However, in
practice, it becomes complicated because the two pulses can be gener-
ated by two different particles; therefore, many data points must be
accumulated and analyzed at one probe location in order to avoid
ambiguity. The problem is further compounded when a measurement of a
second velocity component is required, making this technique tedious
and time consuming. However, the advantage is that it is capable of
measuring very high velocities of very small particles in the back-
scatter mode, especially near solid boundaries. In some specialized
applications, such as the studies of flows around the tips of axial
compressor blading, the laser two-focus technique can be superior to
F-1
conventional laser-doppler systems. Useful references on the princi-
ples of operation and application can be found in Schodl (4), Eckardt
(5_) , and Weyer and Dunker (6_) .
f
^ T~- w/c
F-2
FIGURE F-1. Reference Beam Mode
^°->, , ,^
F-3
• Fbcusing lens
^erture
splitter Photcdetectjor
F-4
Gaussian beam
Intensity distxibuticm
F-5
Figure F-5. The low-frequency component can be eliminated by using
a high-pass filter. Then the frequency of the pulses will be propor-
tional to the component of particle velocity normal to the fringes
divided by the fringe spacing. The fringe spacing depends only on
the wavelength of the laser light and the beam crossing angle.
Therefore, a fixed linear calibration relationship results, regard-
less of the angle at which the detecting device is located with re-
spect to the laser beam. The advantages of this system are its non-
invasive quality and its ability to measure velocity components in-
stantaneously. Measurement of two components is quite straightfor-
ward, achieved by rotating the fringe pattern by 90 degrees. The
disadvantages of this system include the necessity for particles to
be in the flow and the essentially intermittent nature of the signal
that has to be processed. The fact that the particle velocity is
measured, and not the fluid velocity, may be important in some in-
stances.
Figures F-6 and F-7 show the radial velocity profiles measured at the
discharge from the impeller in the vicinity of the volute. Near the
design mass flow (Figure F-6) the front station clearly shows a jet
on the pressure side, followed by the wake of the blade, and finally
high velocity on the suction side. The blade passage therefore ex-
hibits, at least in part of the passage, an isentropic type of flow
field with a gradient of velocity equal to twice the angular veloc-
ity. Some stations show a much larger wake, and there is even a
small portion of reverse flow. Thus, even at design point, we find
that for this particular impeller the flow field is extremely com-
plex. At a mass flow of approximately 30% of the best efficiency
F-6
K magnitude
visibility V =
K iresgnitude
F-7
lAL VELOCITY PROFILES
F-8
VEIXCITY PROPIUIS
- 250 GPM
3.0
2.5
a.o
X.5
1.0--
URING POSITIONS
-.05_.
F-9
point flow (Figure F~7), it can be seen that the highest velocity
actually occurs at the corner formed by pressure side of the blade
and the front shroud. Normally, by potential flow calculation one
would expect that the highest would occur on the suction surface.
As a matter of fact, behind the blade on the suction surface there is
a strong wake that even shows negative velocity. In other words, the
flow is actually going inward. The position nearest the back shroud
has a high negative velocity throughout. Therefore, at low flows the
complexity increases tremendously and, as a consequence, the deter-
mination of blade forces, etc., would also be a difficult undertaking.
REFERENCES
F-10
D. Adler and T. Levy. "Laser Doppler Investigation of the Flow
Inside a Backswept Closed Centrifugal Impeller." J. Mech. Eng.
Sci., Vol. 21, No. 1, 1979.
F-11
APPENDIX G
CAVITATION DAMAGE TESTING
G-1
rates; therefore, the propensity to damage at any particular point in
the flow field can be accurately ascertained.
HIGH-
TENSION
SOPPLir
plLTER I
PHOTOMULTIPLIER
OR GEI(S;R TUBE
DETECTOR COONTE;
D
3ATA LOGGER
STRONTIUM 90 , __
6 SOURCE • ~-*n COMPUTER
SUBSTRATE
G-2
There is one other method for the determination of cavitation damage,
somewhat older than the methods mentioned above. It is based on the
measurement of noise emitted by cavitation. A great deal of work was
done in this area in the middle 1960's, when it appeared that the
measurement of noise could give a quick and direct assessment of the
damage experienced by the impeller. A 1968 paper by Pearsall and
McNulty (3_) , of the National Engineering Laboratory, showed that at
least qualitatively there is a relationship between erosion rate and
the amount of noise, as is shown in Figure G-2. Recent work, however,
has indicated that this measurement is perhaps misleading, because
noise is emitted whenever a cavitation bubble collapses. It is con-
ceivable that a bubble could collapse in the free stream far from the
vane surface without causing damage. It would, however, generate
noise. As a consequence, the noise measurement method for erosion
rate assessment is in doubt. A recent measurement, based not on
direct noise measurement but on the measurement of pulses of acoustic
energy situated at very high-frequency bands, has been claimed to
produce erosion rate correlations. For example, determination of
bearing failures using acoustic detection is well known. It has been
suggested that the same method could be applied to the determination
of erosion rates. This work is worthy of further study because, if
it produces useful results, it would be a valuable tool for pump manu-
facturers and users.
G-3
tWOAT VUOCtTY - ft/i
CBMPARlSas! Cf NXE »D KOSON BTEa a VELOCFTY ON NDEE WD EROSION
G-4
FirsI work eavllotion
Stondord eaviSalien
Covitaiton to no$f
Visibit initiolion
NfE, ^ ^ ^ S v
Hon^l of *)mo9t-»~
AN
AE-
5?6!
E,-AE/2 E^-»-AE/2
H¥poth«$i!ed bubbis t n t r g y ip®ctr» for »»rioui cavit»- TyiMcsl $wln spsctrum
tioo cooditioni t t i ©orwtsnt valocltY, 'f*^ ® gfvsn malsHal
G-5
lOOOr
OH - ?00
S stm
t o NH!
oso - m % st« too
700
on -
m &00
ItiR
eio -
1
'500
B
/ '
f m
1
Ij
*• OIS
OH
- k 400
015
OIJ 500
Oil \
0 10
100-
005 m
1 I M1
004 100
00)
0«
001
1 1 1 1 1 »^« 1 -L
250 500 SSO «00 450 500 550 ?50 500 SSO 400 «0 500 550
TfB«*ra(tff8
Tsmpsralurt
-=^^rr— 5c
Wtlght t o « V&. tott^JtfSturs for sodium Spectrum srsa vs. tsmpsraturs
G-6
that this efficiency would remain a constant, but it was found to
vary by about a factor of 5. The authors believe that this work
could be extended. If successful, this method would become a very
reliable indicator of the cavitation damage rate occurring on a pro-
totype machine.
REFERENCES
G-7
APPENDIX H
SUMMARY OF EXISTING BYRON JACKSON TEST FACILITIES AND INSTRUMENTATION
Test programs at Byron Jackson Los Angeles Operations are carried out
in three separate test areas, depending on the type of pump; a)
standard types and sizes, b) large vertical nuclear primary coolant
pumps, and c) large horizontal pumps. A fourth site is reserved for
Navy "quiet pump" noise tests. Additionally, there are the
Hydraulics Research Laboratory, the Mechanical Seal Laboratory, and
the Dynamics Research Test Facility.
The nuclear primary coolant pump test area contains three permanently
installed hot loops with a maximum capability of 250,000-gpm flow at
550°F and 2,500 psig. The smallest loop is designed for 50,000 gpm.
Complete instrumentation of the loops provides for remote readout and
recording and/or control of temperature, pressure, flow, vibration,
shaft orbit, shaft position, and other critical test parameters.
A separate boiler-feed pump test bed has been provided. The driver
is the largest of its type in the world, a 25,000-hp variable-speed,
wound-rotor motor. With appropriate gear boxes, speeds are available
up to 6,670 rpm. The pump load is measured using a pedestal-mounted
shaft torquemeter between motor and pump. The test facility handles
flows to 32,000 gpm, pressures to 7,000 psig, and temperatures to
400°F. Smaller boiler-feed pumps are tested in the same test bay.
H-1
with power provided by two 5,000-hp electric motors and gear boxes to
provide speeds up to 9,000 rpm.
H-2
d. Position: Proximity sensor, linear variable
differential transformers, magnetic
transducers, and potentiometers.
H-3
,„ ^^fsmm
^^^^«—^
fc
-€1
H-4
1X« ®.48E4-@1 f .-...«wt,>^p§at.a^,wyuM^^
iV» 7.@4E-@1 £ 80
ax- a.S3E4-@2s
aV" 1.42E-01 iNtoARD mmx^n SHAFT PROBE § 8t©0
P
/
P
X» (HZ)
CHZ) X s@ » RPH
i@.
c vmm }
.li*
^SBffOoiCiH^Bffi^LA stf:s^sa^^^\£:W3L,^:£x<i^.i»i!<a.
10 ae 30 40, 50
P.J.G.
FREQUENCY CHZ)
MAGNITUDE .ASMLS P/P ,Bi ,Sl-MAV-Sl,091»138
8X10X14A 7STG. HOB BOILER FEED PUnP
H-5
b. Order Tracking (Figure H-3)
H-6
1X« 3.74E+03
1Y- i , 9 4 E + 0 0 p
K
10
xl0
H-7
BGN- 5386.00 (RPM) END- 3708.28 CRPM)
I ! I I I I I t I
BGN
'1 8 3 4
90 Hz 504 'Hz 3
FREQUENCY CHZ) xW
MAGNITUDE ,At ,BsB-a 9JUN81 ,12-JUN-8ia3i45s 16
1axtax14 5STG, HDB BYRON JACKSON PUMP DIU DYNAMICS RES LAB DG-10063
H"8
P BGN- 5386.00 (RPM) EN0- 3708.88 (RPM)
S
I I I I I I I
20.
P Pump has 3-vane i m p e l l e r e x c i t i n g a c o u s t i c
/ resonance i n d i s c h a r g e pipe a t a p p r o x i m a t e l y *
P 10 (5277/60)x3=264 Hz
5277 rpm
\d^. ^ % ^ ^ « ^ '^
FREQUENCY CHZ) HIO
MAGNITUDE ,AS ,BSD-1 gJUNSl ^lS-JUN-81,15s03t19
l E X i a X 1 4 5STG, HDB BYRON JACKSON PUMP D I U DYNAMICS RES LAB DG-10063
H-9
•.r-^r-r~^--r~4---rl-r^-Y-^~tr^^
R
E
Q 90
80
70.
6@
S®L ,•1- ^ + * t f . - -^ --
40
. - * • *
30
nl®
r . , * *•
1©
•»&.,.,A—A-«.flii..ft if.Jll.i«A„ I. ^ I ift . iTi fl ..iff B>..i.iift ...fll ,fli-ii«ift. 8 fti .iiiili fl •iiJ.i.i.i.tti I^ th Ti .Hi il„„.
40
RPM 0. K 1 ® @5S 10@H»a7 M 1 0
CAMPBELL DIAGRAM , A f , B t B - a a4JUN8i ,14-JUL-81,10l47ti3
iaXiaX14 SSTG. HDB BYRON JACKSON PUMP DIU DYNAMICS RES LAB DG-10063
H-10
F •,--T--.-,-r--,-J--T-tV-H--4 ! i '' .'-.^•r-V-V-^^-Atr-r-
R -r-^rr-
E
0 90
8®.
*. *"
•*.
7®.
6®.
4, . * 4.
50.- ,5--
40
A-
30.
* . * - • « • •
, * - • « •
.*>**-^*
scie
10
.4.*. - - - - - - "
,A.-A.,nm(t,,—,ft ..A—,A.-.,fl ^ I' • I fl H f •• ?i Ti II T I
4®
I® ^mm -®i
RPM
0. Kl® 40X l®®K-a7 Kl@
CW1PBELL DIAGRAM ,At ,B8B-S a4JUN81 ,14-JUL-81*l@s51183
18X13X14 5STG. HDB BYRON JACKSON PUMP DIU DYNAMICS RES LAB DG-10063
H-11
its noninvasive characteristic. The data thus developed will require
special processing in order to assure maximum signal-to-noise ratio.
The available digital signal processing system is ideally suited to
this type of task.
H-12
APPENDIX I
COMPUTER-AIDED LITERATURE SEARCH METHODOLOGY
1. Hydraulics
2. Vibration in pumps
3. Cavitation
4. Two-phase flow in pumps
5. Impeller forces
6. Flow instability
Each citation found, comprising author, title, reference, and
abstract, was included in the computer printout.
I-l
British Hydromechanics Research Association, Cranfield,
Bedford MK4 3 OAJ, UK
Search strategy: (1 or 2)
and (3 or 4 or 5 or 6 or 7)
and (8 or 9 or 10 or 11)
1-2
APPENDIX J
HYDRAULICS
1.0 INTRODUCTION
J-1
Busemans (8^), Sorensen (9^) , Uchimaru (.10) , and Acosta (1) used the
method of conformal transformation for analysis of flow through cen-
trifugal pump impellers fitted with thin logarithmic spiral blades.
In order to obtain the distribution of the flow components, confor-
mal mapping was employed. This transformed the given physical plan
into one on which the circular system of blades was mapped into a unit
circle. Acosta (1) applied this method for calculation of the pres-
sure distribution along the blade surfaces. The calculations were
compared with experimental results and showed insufficient agreement.
A number of subsequent theories have extended the earlier analysis to
include thick blades (11, 12), flow in a centrifugal impeller
(13-20), and several related topics (21-57).
J-2
furnished by the researchers of Hosei University (7_0) in Tokyo,
Japan. The computation procedures included finite element, singulari-
ty, and quasi-three-dimensional methods. The 2-D methods predicted
performance of the pump just as well as the 3D methods as long as
the flow in the impeller was not separated. The inclusion of viscous
effects did not improve the calculation accuracy of either method.
However, when considering the case of an impeller in which the flow
field was separated, neither the 2-D nor the 3-D method predicted
the performance of the impeller very well.
When separation was taken into account, a 2-D method showed good
accuracy. The limitation of the separation method (7^1) is that it
can predict the velocity field only after the location of separation
on the impeller is assumed. The flow separation method (TJ^) , there-
fore, is not predictive at all. Thus we can conclude that for non-
separated flow excellent methods are available because the as-
sumption of axisymmetric flow in front of and after the impeller and
steady relative flow inside the impeller is valid.
The impeller pumps against a certain head and, if the head is high.
J-3
^
%^
^-
J-4
\s^ Bz y
^X^« jT
H
CM ^^Ni2''>C^^
^ Hs ^ ^ \
1 v^^^'^^-^'V^-^,
/ \J\/X>;
<T ! ^^V\
A^o^ !' \ \ \. V
\^.
/ y^
. ^ 11 \ X\. \ ^
a6o»
Eck. 1^57
FIGURE J-2. Static Pressure Distribution
J-5
the flow is low in typical centrifugal machines. Therefore, as an
impeller passage rotates within the casing, the impeller passage
experiences different static pressures, and it will have to pump
different flows as it goes around the periphery. This suggests that
the relative flow field within the impeller is unsteady. Recent
work by Lorett indicates that some of these effects can be predicted.
J-6
Hg : S t a t i c Pressure
3(»o'
IveasEN. I 9 6 0
50
30
LBS t Test Data
- Calc.
20'
F : Radial Force
10
300
'00 S P M " °
J-?
FIGURE J-4. Flow Separation and Inlet Recirculation
J-8
suction, i.e., the flow will be casing driven. At this point, the
velocity field at suction will not be governed merely by the flow
rate and the area but also by the negative flow driven from the
casing. It is interesting to note that separation is actually fixed
in space, because the separation will be determined by the point at
which the highest static pressure is reached around the casing. As
far as the impeller is concerned, this separation is moving backwards
but at the same speed as the impeller rotation. If this hypothesis
is correct, the prediction of reverse flow will be connected with
the prediction of onset of separation in the various stream surfaces
that constitute the impeller. This work has been continued, and a
paper by B. Schiavello (76) attempts to calculate the separation
point by including the effects of blade loading and of variation
of slip factor as a function of flow rate. A further extension of
this work may lead to a better predictive method for the onset of
recirculation at the impeller eye.
J-9
Hs : Static
Pressure
2T
CM : Meridional
Velocity
SPECIFIC SPEED
Region of Reverse Flow
J-10
Reference 15:
Five impellers were tested and the slip factors were correlated.
The unusual and useful aspect of this work is that the slip factors
were found over a wide range of flow coefficients, from zero flow to
well past the design point.
In our M. I.T. pixonp work we had been looking in vain for such
correlations and measured the flow deviations ourselves. Then Noor-
bakhsh brought us his work. We found to our pleasure that the de-
viation predicted by correlation of the pump with the nearest
design specifications with our pump gave the same curve, within the
limits of accuracy expected, and stopped our work forthwith. Only
within the last year have we had reason to use other impeller
correlations, and we found unexplainable inconsistencies. We re-
sumed our deviation measurements. Therefore, unless these incon-
sistencies can be explained, these correlations should be regarded
as only very approximate.
Reference 24:
From the test results, shown graphically, the authors observe that, as
the tongue is shortened, the head developed by the pump increases at
lower flow and falls at higher flow, and the best efficiency point
shifts from higher flows for longer tongues to lower flows as the
tongue length is reduced. As the tongue position is widened, the
same results as from shortening the tongue are observed.
J-11
widened tongue position (flatter angle), the effect would be to peak
at a flatter angle. That would mean a lower discharge flow rate or,
more precisely, an increase in incidence losses with higher discharge
flow rate.
Reference 30:
A statement was made that vanes in the return channel between pump
stages had more effect on the efficiency than did vanes in the dif-
fuser. This statement gains meaning once the previously reviewed
work has been looked at. The study seemed essentially a "try and
test" investigation. The comments made seemed somewhat naive, and
there was no attempt to produce correlations of use to pump designers
Reference 76:
J'-12
"volute characteristic" (increasing with flow) and the "impeller
characteristic" (decreasing with flow) defines the operating
condition of the pump. Although based on an ideal fluid, this paper
demonstrates from other published experiments the acceptable
accuracy of the method.
From the test results the author surmised that a wide volute throat
has both constructional and hydraulic advantage in comparison to
volute cross sections that narrow down to fit the impeller outlet
closely. The cutwater for volute B was cut back in stages, a pro-
cedure that resulted in a broad, crested efficiency and a best
efficiency point at higher flow rate as the cutback increased.
The author also made a study of the effects of skin friction and
found them to have secondary effects with regard to the hydraulic
losses incurred in the volute flow. The information contained in
this paper should prove useful to those interested in volute-impeller
interaction as well as to those concerned with improving centrifugal
impellers in volute casing performance.
Reference 77:
J-IS
the deviation. Many symbols are undefined, e.g., the units for the
specific speed and the conditions for many of the steps are not given
(i.e., inviscid, isentropic, and so forth). The material is presented
as if to a friendly gathering of old colleagues (e.g., "In the usual
formulation, the left-hand side of the principal equation is . . . .,"
which is hardly helpful nor precise)•
Before this method could be recommended over others for pump design,
favorable comparisons of predicted and measured velocities and losses
would be required.
REFERENCES
J-14
T. C. M. Kumar and Y. V. N. Rao. "Quasi-Two-Dimensional Anal-
ysis of Flow Through a Centrifugal Pump Impeller." Journal of
Fluids Engineering, Trans. ASME, Series I, Vol. 99, No. 3, Sept.
1977, pp. 531-542.
J-15
G. Kamimoto and Y. Matsuoka. "One Method of Calculating the
Flow in a Centrifugal-Type Impeller, Especially With a Large
Number of Vanes and With Splitter Vanes." Paper No. B-12, Pro-
ceedings of lAHR Symposium, Sendai, Japan, 1962, pp. 377-395.
J-16
W. S. Gearhart, R. E. Henderson, J. F. McMahon, and J. H.
Horlock. "The Quasi-Steady Design of a Compressor or Pump Stage
for Minimum Fluctuating Lift." Journal of Engineering for Power,
Trans. ASME, Series A, Vol. 91, No. 1, January 1969, pp. 31-36.
J-17
D. P. Mohammed and A. Ramaraju. "Experimental Studies on the
Flow Within the Side Recesses of a Centrifugal Pump Impeller and
Casing." ASME Paper 30-WA/FE 7 for meeting Nov. 16-21, 1980.
J-18
A. U. Salaspini. "Experimental and Theoretical Study on Centri-
fugal Pump Impeller-Scroll Matching," Univ. Libre de Bruxelles,
Belgium, 1975.
J-19
p. C. Tramm and R. C. Dean. "Centrifugal Compressor and Pimip
Stability, Stall and Surge." 1976 Joint Gas Turbine and Fluids
Engng Div, Conf., ASME.
J-20
Y. R. Reddy and S. Kar. "Study of Flow Phenomena in the Impel-
ler Passage by Using a Singularity Method." ASME Paper No. 71-
WA/FE-2 3.
J-21
S. Mikuzi, I. Ariga, and I. Watanabe, "Description of Compressor
Geometries." ASME Symposium, New Orleans, La. 1981.
J-22
BIBLIOGRAPHY
J-23
APPENDIX K
VIBRATION IN PUMPS
1.0 INTRODUCTION
K-1
1.2 Theory
^x ^xx ^xy
X cXX c x y X M M
XX x y
X
= + +
F K K Y C C Y My x M y y Y
yx yy
[_y
yx yy
Black (3_6, 32_, 54, also see bibliography) developed a procedure for
calculating the dynamic force coefficients for small rotary and
lateral motion of the shaft and proposed it to be:
F K k X C c X m 0 X
X
4 +
F -k K y -c C y o m y
L^
K-2
that the capacity of these clearance spaces for promoting dynamic
instability is increased. It is also observed that the fluid inertia
terms that are normally neglected in the Reynolds equation approach
(Black) are shown to be essential for the correct prediction of
cross-coupled stiffness and direct damping terms.
Adams and Makay (£6) and Childs (6, _38f 39' iJ? il' li^ ^^® currently
continuing experiments to gather more data to support any conclusions
about the adequacy of the theory presented above.
K-3
not been adequately determined. Brennen, Acosta, et al. (50) are
presently conducting an experiment to measure these forces and
determine their role in rotor dynamics.
1.4 Miscellaneous
K-4
•Rigid Besriafs ra Eigii Foundation
e l , f « 15 H8
iesriag
K-5
Very often the frequency could be the same as the blade-passing
frequency. In some installations the frequency could be a small
fraction of the operating speed. A typical example is the ap-
pearance of pressure pulsations in the discharge pipe at the vane-
passing frequency. The existence of reinforced longitudinal waves
can be attributed to the resonant length of the piping. In more
complex instances, as has been demonstrated by Sparks and Wachel (52
the pump can actively take part in this phenomenon by influencing
the acoustic response, and it can also serve as an amplifying
element. Corely (53) demonstrated that strong vibrations were
experienced when the inlet and discharge piping had abrupt
transitions. Apparently some low-level vortices were generated at
the piping discontinuities. These problems were solved by using a
long, smoothly reducing cross section for piping. (See Figure K-2.)
Reference 39:
K-6
(At inboard bearing)
11
/ m
>
,r^
— i y ...e^ yt^—t
ISMI / ^
/ ^
^
/ y ^
^
/ L—__
% t t
K-7
force coefficients. The authors compare measured values and
theoretical values for different boundary conditions. Future
tests are discussed and are necessary to support any conclusions
about the adequacy of the theory.
Reference 54:
REFERENCES
K-8
Y. N. Chen, D. Florjancic. "Vortex Induced Resonance in a
Pipe System Due to Branching." Conf. Proc. Univ. of
Southhampton, England, Sept. 16-18, 1975, pp. 79-86,
K"9
T, lino and T. Kaenko. "Hydraulic Forces Caused by Annular
Pressure Seals in Centrifugal Pumps." NASA Conference
Publication 2133. Proceedings of a Workshop held at Texas
A&M University, College Station, May 12-14, 1980.
K-10
J. H. Spurk and R. Keiper. "Self-Excited Vibration in Turbo-
machines Resulting from Flow Through Labyrinth Glands."
General Electricity Generating Board, Vol. 156, 1976.
K-11
Y. Yamada, K. Nakagayashi, and Y, Suzuki, "Viscous Frictional
Moment Between Eccentric Rotating Cylinders When Outer
Cylinder Rotates." Bulletin of the JSME, Vol, 12, No. 53,
1969, pp. 1024-1031.
K-12
R, J, Fritz and E. Kiss. "The Vibration Response of a Canti-
levered Cylinder Surrounded by an Annular Fluid." KAPL-M-5439.
Clearinghouse for Federal and Scientific Information, 1966.
K-13
52. C. R. Sparks and J. C. Wachel. "Pulsation in Centrifugal Pump
and Piping Systems." Hydrocarbon Processing, Vol. 56, No. 7,
July 1977, pp. 183-189.
BIBLIOGRAPHY
K-14
Black, H. F., and Dos Santos, L. "Stability of Oscillations
in Boiler Feed Pimip Pipeline Systems." Conference on Vibrations
and Noise in Pumps, Paper Clll/75, pp. 99-106.
K-15
APPENDIX L
CAVITATION
1.0 INTRODUCTION
L-1
CAVITATION DAMAGE Q = CONST.
N = CONST.
• ^
INCIPIENT CAVITATION
CAVITATION BREAK-OFF
NPSH
1.2 Theory
Over the years, a great deal of research has been done with regard
to cavitation inception. Despite much research there is still
uncertainty, in a number of areas, e.g., interaction of scale
effects, different liquids, and hydrodynamics. There are so many
combinations of scale effects, liquid condition, and flow regime
that it is obviously difficult, or perhaps impossible, to provide a
comprehensive theory for prediction of cavitation inception.
However, if the calculations are applied within fairly narrow
limits, it is believed that good results can be obtained even with
the most simplistic theory. For example, Figure L-2 shows the
calculated inception point versus the measured inception for a
standard mixed-flow pump. It can be seen that the prediction is
quite good as far as the levels are concerned. The point at which
the minimum predicted inception occurs is slightly different.
However, from a practical point of view in the boiler feed pump
industry, this kind of prediction accuracy would be more than
satisfactory.
L-3
60--
50
40--
30--
10-
6 7 8 9 10 11
1000 GALLONS PER MINUTE
L-4
Although designers and users must be concerned with the problem of
incipient cavitation to know when and where it first appears, it is
more important to understand the cavitation regime, during which
serious erosion, vibration, and deterioration of hydraulic perform-
ance may occur. Prediction of the onset and extent of cavitation
erosion has been attempted using theoretical and experimental tech-
niques, A comprehensive study of existing literature can be found
in the papers written by Thiruvengadam (2_7, 2^) . One of the classi-
cal results of this study is the erosion equation that can be written
as follows:
I
1
= !<*">'' -p[-|^
2 P^o"
In this equation:
I = erosion intensity
p = fluid density
V = flow velocity
o -^
a = cavitation number
Y = surface tension
L-5
By suitable manipulation of the above equation, one can deduce that
the primary parameters involved in cavitation damage are the
following:
a. Degree of cavitation
b. The mean cavitation number, i.e., the general pressure
level at which implosions are taking place
c. Velocity
d. The length scale
We may also identify a few secondary parameters:
e. Surface tension
f. Fluid density
g. Nuclei size
h. Fluid temperature
The effect .of the mean cavitation number, however, is not well
known. It is clear that pump manufacturers generally provide a
greater safety margin when the pump suction pressure is high. That
is typical of boiler feed pumps using booster pumps to provide high
levels of NPSH at the first stage of the main boiler feed pump.
The effect of velocity on the damage rate has also been researched
widely. Several references can be found that demonstrate that the
damage race increases at some high exponent of the velocity. The
typical number used is 6, originally used by Knapp (2^9^) at California
Institute of Technology, The velocity exponent of 6 can also be
deduced from the previously mentioned equation of Thiruvengadam.
L~6
Q = CONST.
N = CONST.
NPSH
L-7
The effect of length on cavitation damage has also been widely
observed, but there appears to be no standard way to describe it.
For example, the result of Thiruvengadam shows that the rate of dam-
age might simply be proportional to the size of the model. However,
very recent work reported by Rao and Rao (30) indicates that the
damage rate could increase much faster than the size of the cavi-
tation model. For example, in aluminum they found that the exponent
of this rate of increase could be as high as 15. If that is true,
the prediction of the damage rate in a full-scale prototype could
not easily be made from measured observation (31-34) or from measure-
ments made in a small-scale model. This area requires further study.
L-8
1.3 Byron Jackson/EPRI/M.I.T. Literature Review
Reference 6:
Reference 36:
L-9
were lumped into a secondary-flow loss coefficient proportional
(through a factor k) to the square of the lift coefficient (Stanitz
and Prian). The value of k that gave best agreement was found to
be 0.004, above Vavra's recommendation of 0.014-0.018. This seems
to be the only adjustable constant in the prediction method.
The agreement shown is better than one has seen for any other stream-
line-curvature predictions, even without cavitation.
REFERENCES
L-10
Guy E. Habercom. "Cavitation Part 3. Corrosion and Erosion."
(A bibliography with abstracts). National Technical Information
Serv., Springfield, Virginia.
L-11
W. 0. Keefe. "Will Zig-Zag Flow Be the Straightest Way to Solve
Your Pump-Cavitation Problem?" Power, Vol. 124, No. 4, April
1980, pp. 70-71.
L~12
A. Thiruvengadam. Intensity of Cavitation Damage Encountered In
Field Installations. Hydronautics, Inc. Laurel, Md., Feb. 1965.
L-13
Flow in Steam- and Gas-Turbine Plants. Inst. Mech. Engrs.,
London, at Univ. of Warwick, April 1973.
BIBLIOGRAPHY
L'-14
APPENDIX M
TWO-PHASE FLOW IN PUMPS
1.0 INTRODUCTION
1.1 Definition
Up to the present time the problem has been to correlate, and hence
to predict, the head rise that a specified pump will produce when
operating at a given flow coefficient with a given inlet flow
stream of liquid and vapor (or gas). This implies operation in the
first quadrant. For pressurized-water nuclear-reactor systems, the
head or pressure loss experienced in the second and third quadrants
(reversed flow, forward rotation, and reversed flow, reversed
rotation) is also required.
Although the forces and torques on the pump rotor and the overall
system during operation on two-phase flow are also of considerable
interest, these do not appear to have been the subject of research
in the true sense.
M-1
1. 3 Past Work
M-2
Theoretical line
1.0
= void fraction (flow rate)
^Single-phase characteristic
'*Two-phase points
M-3
Reference 1:
Reference 2%
Reference 3:
Torques on the tongue axis were measured. The author found that
maximum efficiency occurred at zero torque. He therefore pro-
posed an automatic control system for pumps in which the position
of the tongues could be varied with operating point.
Three references were quoted, all Polish work. The reviewer assumes
that many similar test programs have been carried out previously.
It is assumed that the relationship between the net thrust on the
tongue and the efficiency would depend critically on where the
tongue axis was located.
M-4
Reference 4;
Reference 5:
In general, these pumps were developed for conditions of very low net
positive suction pressure and were fitted with large, separate screw-
type inducers, usually running at a lower speed than the main pump.
"The basic limit to pumping two-phase hydrogen occurs when the com-
bined effects of high-flow-coefficient operation and blade blockage
cause the flow area within the inducer blade passage to be less than
the inlet flow area. When this occurs, both two-phase and pure
saturated liquid flows will choke. At design liquid flow coefficient,
allowable pump inlet vapor-volume fractions of 10% have been obtained
experimentally. By designing at the maximum leading-edge incidence
angle that will permit two-phase pumping, this value can be doubled
and, by also using a low-speed preinducer, this value can be
increased nearly three times."
Reference 6:
This article is not very useful for design, because it concerns the
test of a pump designed elsewhere. However, the tests showed that
M-5
the pump operated satisfactorily with zero net positive suction
pressure, had good transient response, and seemed to have a wide
range of stable operation. This conclusion is drawn solely from the
plotted test results; nothing was written about flow oscillations or
instabilities.
Reference 7:
This pump has a separate screw-type inducer and, therefore, may not
be of interest for boiler-feed pumps. The principal finding seemed
to be that the angle of attack of the inducers "decisively determines
the cavitation performance." The authors carried out two main
experiments. The inducer in the second pump was designed from the
results of the preliminary experiment and gave good suction
performance. Low-flow oscillations were experienced, and radial and
axial thrusts were measured.
Reference 8:
M-6
It was found that the impeller with three blades showed an increase
in head when a small quantity of air was admitted (void fraction less
than 0.03). This increase has been frequently observed in a small
proportion of other pumps. The quoted supposition has been that the
bubbles fill up a pressure minimiim and reduce losses. The present
authors seem to be stating (our copy was indistinct, and the authors*
acquired English is not unaitibiguous) that the bubbles accumulate at a
pressure minimum and in so doing change the effective blade angle of
the impeller. The water is thereby given a larger tangential velo-
city, which is responsible for the increased head.
Reference 9:
REFERENCES
M-7
W. Lipski. "The Influence of Shape and Location of the Tongue of
Spiral Casing on the Performance of Single-stage Radial Pumps."
Proceedings of the Sixth Conference on Fluid Machinery,
Hungarian Academy of Sciences, Budapest, 1979.
M-8
APPENDIX N
IMPELLER FORCES
1.0 INTRODUCTION
N-1
(1), pp. 116-123, correlates the steady, radial thrust force with the
capacity or volume flow rate of the pump in the following way:
F, = f [1- (Q/Q^)]'
Since Q is the normal or design point flow rate, zero radial thrust
is predicted as design capacity and a maximum thrust at shutoff.
Force increases rapidly at capacities larger than design, as well.
The correlation yields the magnitude but not the direction of the
radial thrust. In a discussion of reference (5^) , Stepanoff admits
this correlation was suggested in a California Institute of Technology
report (circa 1935).
Potential flow theory was used by Domm and Hergt in 1970 (6^) to
derive a theoretical prediction of radial force magnitude and
direction. In this apparently first analytical approach, the
effect of an eccentric location of the impeller with respect to
the volute was included. Measurements confirmed the trends. In
1970, Hergt and Krieger (2) made additional experimental measure-
ments, using a vaned diffuser, and observed a radial force that
increased with impeller eccentricity. A t very low flow, what was
taken to be rotating stall of the diffuser blades resulted in a
N'-2
rotating radial force on the impeller and nonsynchronous vibration
of the pump shaft. The significance of the components of the
rotating force was not yet recognized, A further exposition of
these measurements was made in reference (8).
N-3
pressure distribution in the spaces between the impeller outside
surfaces and the stationary casing walls. More modern work has
simply served to provide more realistic and complex analysis and
measurements of these influences.
Most subsequent work, like that of Gyulai (12_) , is concerned with the
influence of the clearance-space flows on the pressure distribution
over the impeller outside surfaces. Using both numerical predictions
and experimental measurements, he documents the effect of (1) the
impeller throughflow on the clearance pressure distribution when there
is zero radial flow in the clearance space, (2) misalignment of the
impeller with respect to the volute, as well as (3) axial spacing of
fixed wall to moving impeller surface, (4) radial throughflow in the
clearance spaces, and (5) pump-out vanes as a technique for equalizing
pressure distributions on hub and shroud surfaces. Grabow (13)
provides an investigation of design parameters for the pump-out vanes.
N-4
throughflow, in two similar pumps. Two important effects were
observed. Wear ring clearance yielded an outward flow in the back
space and an inward flow in the front space that produced greatly
different pressure distributions on the impeller surfaces, as is
suggested in Senoo and Hayami's work. Further, such effects were
quite sensitive to axial displacement.
N-5
damping as that effect which occurs in the bearing supports and
rotating damping as that occurring in rotating elements of the
shaft. At the time of his writing Smith was aware of the work of
Newkirk and Kimball, which identified internal friction, or
hysteresis, as an example of rotating damping. Smith's linear
analysis of a symmetrical rotor/bearing system that included both
forms of damping led him to realize that rotating damping in a
shaft spinning at a frequency above the undamped natural frequency of
the system would "urge on" the motion of the free natural vibration,
tending to increase its amplitude without limit. Smith's equations
of motion of the shaft included a contribution to the resulting
force from the velocity of the vibratory motion and were cross-
coupled in the displacement terms.
N-6
mechanisms pertinent to radial, centrifugal turbomachines.
As distinct from steady radial and axial thrust, which has already
been considered, the derivation of unsteady impeller forces from
unsteady channel flow will be considered in this section. As des-
cribed by Smith, Kapitsa, and Crandall, rotating damping, when com-
bined with the velocity of the vibratory motion, is capable of
enhancing or exciting the vibratory shaft motion. Since the velocity
of the vibratory motion may increase and decrease as the shaft orbit
is traversed, so the resulting exciting force may fluctuate or be
unsteady as well. Crandall discusses the frequency dependence of
the rotating damping.
N-7
gated. Thus the impeller forces were predicted to have a
stabilizing influence. The impeller force was found to be a function
of the blade tip angle, the ratio of the whirl angular velocity to
rotational angular velocity, and the blade loading, and to be
independent of the number of blades. In the experimental work it
was observed that the impeller force varied with both shaft dis-
placement and orbital velocity and that tangential fluid forces
damped the whirl motion in most cases. However, at low flow rate
and low whirl speed ratio, negative damping did occur. The ranges
of parameters studied in the numerical work were not compared with
actual radial impeller operating conditions, so that no judgment
on their realism can be made.
N-8
Several comments can be made. In the compressors evaluated by
Thompson (2 3) floating ring seals far outnuitibered labyrinth seals,
and still the fluid dynamic excitation of the unsteady impeller
flow led to unstable rotor motion. Colding-Jorgensen (30) uses a
potential flow method to predict the impeller force as well as
stiffness and damping coefficients. Finally, in a method that avoids
the limitations of unsteady potential flow analysis, Daiguji and
Shirahata (31) and Daiguji and Sakai (32) use a finite element method
to compute the inviscid rotational unsteady flow through a cascade
in which periodic and nonperiodic variations in the flow rate result
in the time-variation of the lift and drag. This approach opens a
more realistic possibility for the analytical estimation of unsteady
impeller forces.
REFERENCES
N-9
H. W. Iverson, R. E. Rolling, and J. J. Carlson. "Volute
Pressure Distribution, Radial Force on the Impeller, and Volute
Mixing Losses of a Radial Flow Centrifugal Pump." Journal of
Engineering for Power, ASME Trans., Vol. 82, No. 2, pp. 136-144.
April 1960.
N-10
F. Gyulai, V. Anton, and M. Popoviciu. "Researches Regarding
Axial Hydraulic Thrust in Centrifugal Pvimps. " Proceedings 5th
Conference on Fluid Machinery, Akademiai Kiado, Budapest, 1975,
Vol. 1, pp. 379-387.
N-11
C. E. Trans, 4179, British Lending Library).
N-12
R. Jenny. "Labyrinths as a Cause of Self-Excited Rotor Oscilla-
tions in Centrifugal Compressors." Sulzer Technical Review,
No. 4, 1980, pp. 149-156.
N-13
APPENDIX O
FLOW INSTABILITY
1.0 INTRODUCTION
0-1
well as the blade geometry designated by the inlet blade angle, the
discharge blade angle, and the schedule of change of blade angle
between inlet and discharge.
1.3 Theory
0-2
reason for this is that such work is doubly difficult, because (1) all
the phenomena listed by Yididiah are three-dimensional flow effects
and (2) all occur at off-design flow conditions. It is only recently
that Moore and Moore [1981 {5)] report the first, limited success
with computation of viscous, three-dimensional radial impeller flows,
from which may ultimately follow an analysis and description of
Yididiah's factors.
REFERENCES
1. A. J. Stepanoff. Centrifugal and Axial Flow Pumps. John Wiley &
Sons, Inc., 2nd ed., 1957.
0-3