J. HYDRONAUTTCS VOL. 1, NO.

Integrated Hull Design, Boundary-Layer Control, and

Propulsion of Submerged Bodies
F. R. GOLDSCHMIED*
University of Utah, Salt Lake City, Utah

Power requirements of large submerged bodies at high Reynolds numbers are optimized
by the hydrodynamic synthesis of body design, boundary-layer control, and propulsion. Con-
ventional rigid skin, all turbulent boundary layers, and a single suction slot are accepted as
realistic engineering constraints. A 3:1 body has been designed and has been tested in a wind
tunnel at a Reynolds number of 107; the wake drag has been found to be CDW = 0.002, and the
equivalent suction drag CDS = 0.0142 yielding a total equivalent drag CD = 0.0162 (based on
volume). This can be compared to CD = 0.0235 for the best conventional streamlined body
(Akron airship model). A total engine power coefficient has also been determined, Cp* =
0.01585, while a conventional streamlined vehicle with stern wake propeller has a Cp* = 0.0215,
thereby showing a net gain of 26%. There is a possible tradeoff between suction and pro-
pulsion powers allowing the total power coefficient to decrease to CP* = 0.0100 and to reach a
50% power gain.

Nomenclature Cp = total power coefficient

TIP = shrouded impeller's pump efficiency
x = axial distance f]T = propulsive efficiency
y = radial distance Cp* = Cp/rip, total engine power coefficient
c = airfoil chord BLC = boundary-layer control
L = body length
X = x/c or x/Lj dimensionless axial length Subscripts
Y = y/c or y/L, dimensionless radial length 0 = freestream condition
g = suction slot width 1 = upstream of discontinuity
5 = boundary-layer thickness 2 = downstream of discontinuity
8* = boundary-layer displacement thickness 3 = body trailing edge
0 = boundary-layer momentum thickness x = lengthwise position on body or airfoil
d = body diameter s = pertaining to boundary-layer-control suction
V = useful enclosed body volume w = pertaining to the wake
D8 = equivalent suction drag
Dw = wake drag 1. Introduction
T = thrust
Ws
Ww
= power charged to boundary-layer-control suction
= power charged to propulsion (counteracting the wake
A ^ investigation of an engine/airframe, propulsion/
boundary-layer-control/body design synthesis was pro-
posed in 1954 by Goldschmied1 to the Office of Naval Re-
drag)
U = fluid velocity outside the boundary layer search (ONR) for airship application (Reynolds number over
u = fluid velocity within boundary layer 108). It was believed from preliminary estimates that there
USl = velocity entering propulsor, relative to body was a good chance to achieve substantial gains in power re-
US2 = velocity leaving propulsor, relative to body quirements, within realistic engineering constraints of naval
p = fluid mass density airships.
ju = fluid absolute viscosity Another power-reduction method was also considered at
v = fluid kinematic viscosity
that time for airships, i.e., the stern propeller as applied to a
RL = UoL/v, body Reynolds number
Re = Uitii/Pj momentum-thickness Reynolds number up- conventional streamlined hull to extract energy from the
stream of discontinuity wake. This was investigated in a large wind tunnel much
ms = total suction mass flow rate later (1962) by McLemore2 and found to offer worthwhile
Cm = ms/pUoL*, total suction mass flow coefficient power gains (as compared to car-mounted or fin-mounted
Q = ms/pTrdi, suction volume flow per unit slot length propellers).
CQ = Q/UiOi, suction flow coefficient The inviscid body design and the boundary-layer-control
C.H = 2AHi/pUi 2 , suction head coefficient analysis were performed in 1954 with ONR funding. This
q = |pf/ 0 2 F 2/3 , denominator of drag coefficient work is reported in Ref. 3. Later a wind-tunnel model was
CDS = Ds/Qj equivalent suction-drag coefficient designed, and in 1956 a series of wind-tunnel tests was per-
CDW = Dw/q, wake-drag coefficient
CD = D/q, total drag coefficient formed at the David Taylor Model Basin, at Reynolds num-
CT = T/q, thrust coefficient bers up to 1.2 X 107, comparing the new design directly with
CPs = Ws/qUo, suction-power coefficient a conventional airship model hull. The experimental results
Cpw = Ww/qUo, propulsion-power coefficient have been reported by Cerreta4 in 1957. No further test
work was carried out because of the vanishing interest in
naval airship development. A final summary report of the
Presented as Paper 66-658 at the AIAA Second Propulsion investigation up to 1957 is given by Ref. 5. It is unfortu-
Joint Specialist Conference, Colorado Springs, Colo., June nate that, despite the excellent results of the first wind-tunnel
13-17, 1966; submitted June 13, 1966; revision received January tests, a self-propelled model was not built and tested in a
3, 1967. This work was funded by the U. S. Navy, Office of
Naval Research, under Contract NOnr 1412(00) LI. [3.09, 4.22] large wind tunnel, such as used by McLemore.2
* Associate Research Professor, Mechanical Engineering De- A review of the wind-tunnel results and a new propulsion
partment and Director, Fluid Control Systems Laboratory. analysis were undertaken by the author in 1965-1966 as
Member AIAA. personal research at the University of Utah. It is the au-
JULY 1967 INTEGRATED HULL DESIGN

thor's belief, however, that this aerodynamic research work ever, the suction quantities required are larger and are a
transcends mere airship application and has basic import to function of the number of slots.
.submerged self-propelled bodies in general. This belief is Then the fluid which has been sucked into the body must
the motivation for the present paper, after due consideration be pumped up to freestream static pressure and to flight ve-
of the hydrodynamic progress made in the last ten years and locity so that it may leave the body ideally at "zero" velocity
the present state-of-the-art in underwater propulsion and drag relative to body (US1 = US2 = 0). This usually takes much
reduction at high Reynolds numbers. power and it is never worthwhile unless it buys a substantial
reduction in propulsion power per unit body volume.
It has been observed, however, that once this boundary-
2. Hydrodynamic Approach
layer-control fluid has been brought up to zero relative ve-
The present approach to the power optimization of self- locity, it will enable a reaction propulsor to produce thrust
propelled submerged bodies at high Reynolds numbers (over in the most economical manner. Thus, only in the combina-
109) is based on a hydrodynamic synthesis of hull design, tion with propulsion has boundary-layer control proved
boundary-layer control, and propulsion, considering the advantageous powerwise.
following well-known and accepted factors.
2.4 Trailing-Edge Full Suction
2.1 Reference Criteria
It is well-known that it is more efficient powerwise to suck
It is important to set forth the reference quantities to be all the turbulent boundary layer on a flat plate at its trailing
used in the definition of classification criteria. Here the edge and to complete mechanically its energization to flight
useful enclosed volume V and the engine's power expenditure velocity than to allow it to pour its kinetic energy into a
(hp) will be used for reference in the definition of a power natural wake requiring thrust to counteract the drag force.
coefficient In the former case, the pump power required to complete
the boundary-layer energization is given by:
(1) w. / 77u r 1 - 7;-w~i
= VP /II nu / /g I IQ I I ^2/ (2)
^y (_y
"Drag" coefficients are semantically detrimental in the pres-
ent investigation, although traditional in many hydrody- In the latter case, the thrust power required to overcome to
namic fields. Control and exploitation of the boundary-layer wake drag is given by
kinetic energy are the cornerstones of the present synthesis.
Ww f* u_ [" _
(3)
2.2 Reynolds Number Effects
" ^ Jo U0 \_ U

Large Reynolds numbers up to and over 109 are considered If the pump efficiency rjP is equal to the propulsive efficiency
in this investigation. The consequences are that the ex- rjT, the theoretical gain for a turbulent boundary laj^er is
tent of the reliable laminar boundary layer on the body will of the order of 10%.
be negligible (over Rx = 107) despite favorable pressure
gradients, and even despite boundary-layer control by dis- 2.5 Boundary-Layer-Intake Propulsion
tributed area suction or damping skins. Also the possibility It is well-known and accepted that it is much more efficient
of severe hull vibrations militates against the chances of powerwise to use boundary-layer fluid (rather than freestream
laminar flow, without mentioning the usual skin protuber-
fluid) to feed the propulsor because this fluid possesses energy
ances, irregularities, and barnacles. Thus as a practical gained from the body. Thrust gains can be made, with ap-
matter, turbulent boundary layers must be accepted through- parent propulsive efficiencies increased (based on freestream
out.
velocity) up to 140%.2 So-called wake propellers, designed
to match the stern boundary-layer profile of a particular
2.3 Boundary-Layer Control body at a particular Reynolds number, have been used suc-
cessfully to regenerate wake energy and so have some pump-
Turbulent boundary-layer control on a conventional jets with boundary-layer intake. However, there are two
streamlined body is concerned with the prevention of flow problems here; first, a hull drag increment is created by these
separation on the aftbody. This job is complicated by the stern propulsors, in the nature of a hull pressure drag, this
high sensitivity of the flow to body angle of attack, creating increment being given as 7-10%,6 12%,7 and 19%2; then
three-dimensional separation contours, and also by the de- such propulsors are quite sensitive to angle of attack.
pendence of the separation point on the Reynolds number.
Thus, a single suction slot which proves adequate in a wind- 2.6 Fineness-Ratio Effect
tunnel test at zero angle can offer no assurances for a proto-
type in actual operation. In general, when the adverse The optimum fineness ratio of streamlined bodies (on the
pressure-gradient area extends over 60-75% of the body basis of enclosed volume) has been found experimentally to
length, a fixed slot will encounter a very wide variation of be between 5:1 and 6:1. Here the sum of skin friction and
boundary-layer thicknesses and profiles. A stern pump- pressure drag is the lowest in relation to volume. These
jet (on conventional body) is a single suction slot well aft results have been determined in wind tunnels and water
(for BLC purposes), subject to all the difficulties and limita- channels at Reynolds numbers in the vicinity of 107. As
tions of the single slot. On the other hand, distributed the Reynolds number increases, the skin friction will decrease
area suction may be used over the aforementioned 60-75% but the pressure drag will increase according to the turbulent
length to control the boundary-layer growth in its entirety. separation criterion of Goldschmied.8 This means that the
However, this is not deemed quite practical because the po- optimum fineness ratio will increase. If the pressure drag
rous skin is structurally weak (prone to failure, especially in could be controlled or eliminated by boundary-layer control,
fatigue) and easy to clog, particularly in seawater. Further- the optimum fineness ratio will decrease toward unity, for
more, the ducting needed over the body under the large suc- minimum surf ace-area/body-volume ratio and consequent
tion area will take up some useful volume and thereby in- minimum skin friction and hull weight. At operational
crease the Cp value. A multiple-ring-slot skin has also been Reynolds numbers of 108 to 109, the boundary layer will
used, obviating the problems of fatigue and clogging. How- always be turbulent even on the sphere.
 F . - R . GOLDSCHMIED J. HYDRONAUTICS

1.8 designing the airfoil so that in inviscid (potential) flow it

1.6 had a stabilizing favorable velocity gradient along the whole
1.4 chord, except at one position (well aft) where a velocity dis-
1.2 continuity occurred. Thus if sufficient suction was applied
1.0 at this one point to prevent separation of the boundary layer
UPPER SURFACE crossing the sudden pressure rise (similar to a shock wave),
.6 <X = 0-* CQ =0.00474 -
<X = 4 - A CQ 0.00503 _
laminar boundary-layers might be expected throughout (if
.4
< X = I O - V C0 =0.00923 the Reynolds number was not too high and the surface was
smooth and clean enough), regardless of airfoil thickness
and of the angle of attack (up to some desired value).
The first airfoil was designed by the method of Goldstein
with a 16% thickness. A great deal of aerodynamic de-
velopment was carried out subsequently both in England
and in Australia with airfoils up to 34% thickness. This
effort culminated in full-scale glider tests in Australia from
1948 to 1951. A brief history of the Griffith airfoil develop-
LOWER SURFACE ment is given by Head11 and by Thwaites.12 The airfoil is
<* = 0 - CQ = 0.00455 ~ also mentioned by Schlichting13 in his well-known book
Oc= 4 - A CQ = 0.00504-
oc = |Q-T C0 = 0.00523 _
(pp. 283, 284).
Figure 1 shows the velocity distribution on a 30% sym-
1.0 metrical Griffith airfoil, as determined in the wind tunnel
by Gregory and Walker14 for three angles of attack 0, 4,
and 10. At a = 0 the velocity-ratio peaks at 1.4 and at
Fig. 1 Griffith 30% airfoil with suction boundary-layer
a = 10 it peaks at only 1.7. The discontinuity is at x/c
control. Note that 30% Griffith airfoil is very similar to
34% Lighthill airfoil. (See Ref. 14).
= 0.80 for all a values and the upstream velocity ratio is
1.4 regardless of a. The suction flow needed is increased
only 55% from a. = 0 to a = 10 (as indicated by the CQ
2.7 Griffith Airfoil Concept coefficient). Thus, it is seen that this airfoil will operate
in the desired manner in an a angle range 10.
The stepwise velocity distribution concept of Griffith, as For ease of comparison, a conventional low-drag 12%
embodied by Ligh thill's9 direct-design-at-incidence in viscid NACA airfoil velocity distribution is depicted in Fig. 2 for
method, yields airfoils with favorable pressure gradients the same values of a. At a = 0 the velocity ratio peaks at
throughout, up to a specified angle of attack, except for a 1.2 (lower than 1.4) at x/c = 0.35, but at a = 10 it peaks
pressure discontinuity. This discontinuity is actually a at 2.8 (much higher than 1.7) at x/c = 0.01. It is the per-
very steep pressure gradient over a small-enough width so formance at a = 10 that really tells the difference between
that a single suction slot is able to encompass the area and to the two airfoils! Of course, the reasons behind the applica-
allow the boundary layer to cross the discontinuity. The tion of the Griffith concept to bodies are quite different from
location of the discontinuity does not change with angle of those which motivated the extensive airfoil development.
attack and also its pressure ratio remains substantially the These reasons will become more apparent later in the present
same. It has been shown experimentally on airfoils up to paper.
34% thick that the Griffith concept is valid over a range of The first problem in the axisymmetric hull design according
angles of attack and that the pressure drag is reduced to to the Griffith concept is the lack of a body method com-
negligible values for any thickness, as long as the boundary- parable to that of Lighthill9 for direct design at incidence of
layer suction is adequate at the discontinuity. inviscid airfoils. Analytical methods for the calculation of
incompressible inviscid velocity distribution over axisymmet-
3. Axisym metric Hull Design ric bodies have been investigated and applied by Gold-
schmied.15 They are of little use for the present design
In 1943 Richards and Burge10 reported the results of small- problem. A thorough discussion of uniform flow past bodies
scale wind-tunnel tests of a new type of airfoil. This scheme of revolution is given by Thwaites16 in his book on incom-
arose from a suggestion by A. A. Griffith and consisted in pressible aerodynamics.

.20 AIRFOIL PROFILE

.15
.10
AIRFOIL NACA 63, -012 V. .05
C 0
-.05
2.8
-.10
2.6 -.15
-20
2.4
CURVES FOR UPPER 0 O.I 0.2 0.3 0.4 0.5 0.6 0.7 0.8 09 1.0
2.2
2.0
SURFACE AT
CX = 0 -

CALCULATED INVISCID UPPER
CX= 4 - T
L8 OC= 10 - A 1.6
l6
u 1.4
1.2
DO 1-4
I0
1.2 u
U0 -8
I.'O .6
.8 .4
.6 .2
.4 0
0 O.I 02 0.3 0.4 0.5 0.6 0.7 OB 0.9 1.0
.2


Fig. 3 Lighthill 34% airfoil designed directly for 7
Fig. 2 Typical conventional low-drag NACA airfoil. angle of attack.
JULY 1967 INTEGRATED HULL DESIGN

It was decided to achieve first a two-dimensional airfoil

shape by LighthilFs method. Appendix IV of Ref. 9 already
presents the complete calculations for a 34% symmetrical
airfoil with the discontinuity at 83%, designed for constant
velocity over the forebody and over the aftbody at a = 7.
The airfoil profile and the velocity distribution for a. = 7
are shown in Fig. 3. Furthermore Appendix V of Ref. 9
presents the complete calculations for a 48% symmetrical
airfoil with the discontinuity at 89%, designed for constant
velocity over the forebody and over the aftbody at a =
12. These airfoil profiles have been calculated for two-
dimensional inviscid flow. The expected boundary-layer
displacement thickness 5* should be subtracted from the
calculated profile.
It was decided to use the 34% profile to yield a 3:1 body
with adequate angle-of-attack range. The next problem
was to convert the two-dimensional profile into the corre- Fig. 4 Inviscid velocity distribution as obtained in
sponding axisymmetric shape. For this purpose the analog electric analog tank.
electric tank lends itself beautifully because the same tank
can either be used horizontally (two-dimensional analog)
Thus, the determination of the inviscid body shape is partly
or slanted at 10 (axisymmetric analog) as shown by Cheers
and Rayner,17 Goldschmied,18 etc. Figure 4 shows the in- analytical by LighthilPs method and partly experimental in
viscid velocity distribution obtained in the electric tank for the electric analog tank. It is to be noted that the stern
both cases at zero angle of attack. now ends into a sharp trailing point, with no allowance for a
finite stern-propulsion jet. Necessary design modifications
are to be considered later in the final integrated design.
Similarly, the inviscid body design gives no information on
Table 1 Step velocity profiles the detailed suction-slot arrangement. For convenient
reference, both the 34% airfoil profile (from Ref. 9, Appendix
Airfoil profile table Body profile table
IV) and the final body profile4 are tabulated in Table 1. A
X Y X Y schematic is provided in Fig. 5 showing nomenclature and
coordinates.
0.000000 0.000000 0.0000 0.00000
0.002063 0.007785 0.0326 0.06462 It is to be noted that the aftbody, past the discontinuity,
0.006973 0.016421 0.1007 0.08215 has not been included in the useful volume, because it will
0.014770 0.025675 0.1348 0.0966 be occupied mainly by the impeller and associated ducting.
0.024319 0.035314 0.1691 0.1094 Typical wind-tunnel test results from Ref. 4 are shown in
0.037017 0.045195 0.2029 0.1204 Fig. 6, where the experimental points (at Reynolds number
0.051396 0.055201 0.2368 0.1302 of 1.1 X 107 and at the best suction-slot width) are super-
0.067808 0.065227 0.2710 0.1384 imposed on the inviscid velocity distribution (as determined
0.086151 0.075182 0.3021 0.1451 in the electric analog tank). The agreement is seen to be
0.106321 0.084979 0.3391 0.1515 excellent, except for the slight sink effect just upstream of the
0.128209 0.094540 0.3735 0.1573
0.151701 0.103791 0.4073 0.1617 discontinuity. This discrepancy could be corrected by a
0.203011 0.121087 0.4409 0.1652 very slight recontouring of the body. The velocity ratio
0.23052 0.129013 0.4751 0.1678 across the discontinuity is 1.2/0.6 = 2.0. This yields a
0.259223 0.136370 0.5093 0.1694 large pressure recovery over a negligible axial span g/L =
0.288824 0.143124 0.5431 0.1698 0.008 or 0.8% of the body length, in full conformance with
0.350300 0.154588 0.5771 0.1688 the Griffith concept of aerodynamic design. Furthermore,
0.381861 0.159221 0.6110 0.1666 the velocity gradient is favorable both on the forebody and
0.413769 0.163073 0.6454 0.1633 on the aftbody, giving "natural" control of the turbulent
0.445866 0.166109 0.6792 0.1585 boundary layer in those areas.
0.509968 0.169658 0.7128 0.1521
0.541652 0.170011 0.7261 0.1490 It is unfortunate that the wind-tunnel tests were not ex-
0.572896 0.169345 0.7346 0.1469 tended to angles of attack other than zero. It is just the
0.603514 0.167752 0.7431 0.1446
STA. l-.rSTA. 2
0.662218 0.161698 0.7516 0.1412
0.689991 0, 157332 0.7600 0.1395 SUCTION SLOT
0.716497 0.152094 0.7684 0.1367 r STA. 3
0.741531 0.145816 0.7770 0.1335 *Us,.m<i
0.764888 0.138449 0.7855 0.1302
0.786312 0.129888 0.7941 0.1265
0.805419 0.119934 0.8026 0.1224
0.821777 0.107971 0.8109 0.1176
0.830155 0.097124 0.8198 0.1113
0.8300542 0.093303 Discontinuity' point
0.831546 0.083391 0.8283 0.09795
0.836909 0.073538 0.8446 0.07690
Discontinuity point 0.8615 0.06132
0.850899 0.058587 0.8786 0.04872
0.880142 0.036137 0.8956 0.03778
0.909930 0.021419 0.9126 0.02709 "- PROPULSION
0.936557 0.011761 0.9295 0.01746 PUMP DISC
0.959021 0.005717 0.9462 0.00889 SUCTION PUMP
0.976821 0.002250 0.9635 0.00318 DISC
0.989670 0.000552 0.9806 0.00047
1.000000 0.000000 1 . 0000 0.00000 Fig. 5 Configuration sketch, including suction and
propulsion.
 F. R. GOLDSCHMIED J. HYDRONAUTICS

negligible extent anyway. (1-10% length). F. W. Boltz22

has explored experimentally the problem of transition on
bodies, flat plates, and wings. In the wind-tunnel tests of
Ref. 4, since the body Reynolds number was only 107, transi-
tion was carefully triggered for all tests at 10% length by a
double row of staggered, discrete, conical turbulence-stimu-
lating elements. A sublimation technique was used to check
on the actual occurrence of transition. In this manner, the
experimental boundary-layer data can be interpreted exactly.
The quadrature method of Truckenbrodt23 has been selected
for the computation of the momentum thickness 0i, upstream
SUCTION SLOT =0.008 of the discontinuity. All the suction coefficients at the dis-
continuity use 0i as the reference length

A =
L
Fig. 6 Wind-tunnel velocity distribution with boundary-
layer-control suction on axisymmetric hull. Data from
Ref. 4.

ability to handle a range of a angles that is one of the major

claims for the present body design with boundary-layer
control. In fact, flow about conventional streamlined bodies (4)
is quite sensitive to angle of attack. This is illustrated quite +W
dramatically by Fig. 7 which portrays the contours of mini-
mum pressure and of turbulent separation at a. = 0, 6,
[E* y^ AT integration constant (5)
Ltfo* L ' L J
12, and 18 as determined experimentally in the wind tunnel
by Freeman19 and analytically by Alien20 on an airship model. 11/2) l+n

Recently this complex flow phenomenon has been studied both | (6)
theoretically and experimentally by Rodgers.21 It can be
readily appreciated that these effects are quite significant 0.036
for both drag and stern propulsion, i.e., for the manner in
cfl = turbulent friction (7)
which energy is put by the body into the boundary layer
and for the manner in which it can be regenerated to useful 1.328
laminar friction (8)
purposes. (t/oLA)1/2
If turbulent flow is assumed from the nose, as it is proper
4. Boundary-Layer Analysis for the case of high Reynolds numbers, then Ci* = 0. For
The boundary-layer analysis comprises the forebody the wind-tunnel tests where transition was fixed at 10%
boundary-layer calculation and the analysis of boundary-layer length, then Ci* must be computed from Eq. (6) from x = 0
control at the discontinuity in regard to suction flow, pres- to x = 0.10L using C/i from Eq. (8). It is seen from Table
sure, and power, and in regard to boundary-layer growth as 2 that H is between 1.43 and 1.52, indicating a good healthy
a function of suction. The aftbody boundary-layer calcu- boundary-layer profile upstream of the velocity discontinuity.
lation and the wake-drag computation are then carried out For the laminar case n = 1.0 and for the turbulent case
to complete the boundary-layer analysis. It is assumed that n = -g-. In Fig. 8 there are represented the experimental
there is no pressure drag at all. momentum-thickness points (taken at 79.5% length), the
The understanding of Reynolds number effects is believed same points corrected theoretically to 83%, and the theoretical
to be adequate to provide exact extrapolation to larger bodies 0i curves at both 79.5 and 83% length. It is seen that the
at higher velocities because there are no problems of turbulent wind-tunnel points are somewhat lower than the curves by
boundary-layer growth under adverse pressure gradients and about 15% at RL = 107. For convenience, the data of
of turbulent separation in three dimensions (see Fig. 7), up Ref. 4 are reproduced in Table 2 for the case of minimum
to a = 7.
Table 2 Boundary-layer wind-tunnel data
4.1 Forebody Boundary Layer
The computation of the axisymmetric boundary layer under RL Si* 0i H = d*/e
favorable velocity gradients on the forebody does not present 4.51 X 10 G
0.1047 0.0705 1.485
any difficulty either for the laminar or the turbulent case. 4.41 0.947 0.622 1.522
The location of laminar/turbulent transition cannot be pre- 4.51 0.0888 0.0609 1.458
dicted exactly, but this is seldom important for large bodies, 4.60 0.0827 0.0553 1.495
because at high Reynolds numbers the laminar region is of 7.15 0.0948 0.0624 1.519
7.20 0.0833 0.0572 1.456
CONTOURS OF BOUNDARY- 7.10 0.0802 0.0562 1.427
LAYER SEPARATION
7.10 0.0878 0.0584 1.503
10.15 0.0848 0.0563 1.506
10.0 0.0793 0.0553 1.434
4.36 0.0833 0.0582 1.431
6.95 0.0880 0.0608 1.447
7.10 0.0846 0.0581 1.456
L |2 H8 7.05 0.0799 0.0558 1.432
CONTOURS OF MINIMUM
6.95 0.0819 0.0562 1.457
PRESSURE 9.8 0.0795 0.0550 1.445
10.2 0.0777 0.0537 1.447
Fig. 7 Contours of minimum pressure and of flow 10.1 0.0791 0.0552 1.433
separation on "Akron" hull at angles of attack. (See 11.3 0.0786 0.0547 1.437
Refs. 19 and 20.)
JULY 1967 INTEGRATED HULL DESIGN

suction with a slot width g/L = 0.008 (boundary-layer

rake at s/L = 0.795 and note that L = 58.80 in.).

4.2 Boundary-Layer Control at the Velocity

Discontinuity

In the body design the location of the velocity disconti-

nuity has been placed at 83% length. Here the velocity ratio
changes abruptly from 1.2 to 0.6 and suction is required to
allow the boundary layer to cross the 2:1 velocity step without
separation. The best suction-slot width has been found in
the wind tunnel to be g/L = 0.008 or g/6i = 6. Since the
ratio of body-radius/momentum-thickness is large at the
slot, 2/i/0i = 75, the flow may be considered two-dimensional
for the purposes of analysis. Thus, a large body of airfoil
theory and experimental knowledge may be exploited for this
present application.
G. I. Taylor suggested a simple argument to explain the
behavior of the boundary-layer in crossing the velocity dis- Fig. 9 Flow parameter for boundary-layer-coritrol suction.
continuity. The argument has two basic assumptions: 1)
There is no change in total head of a streamtube as it crosses
the discontinuity. 2) The static pressure is constant through
are also shown and are seen to have a very steep trend above
the boundary layer. The assumptions have been carefully the theoretical curve. The evidence is not adequate for
examined experimentally by Gregory24 and it is found that firm conclusions but it does appear that at RQ ^> 104 there
the suction-rate estimates are reasonable but that close agree- will be good agreement between theory and experiment.
ment with actual velocity profiles is not obtained. In fact Figure 10 presents the pressure coefficient CH = 2A//i/~
the change in head in a boundary-layer streamline as it
pUi2 plotted against Re. It is seen that the points lie mainly
crosses the discontinuity is small compared with changes in
below the theoretical line. Again it appears that there will
be agreement for Re ?> 104.
head normal to the streamline. However, static pressure
It is to be noted that the theoretical lines for CQ and CH
changes through the boundary layer are observed to be large.
are based on the minimum values for stabilization of the turbu-
Preston, Gregory, and Rawcliffe25 describe a method for
lent boundary layer crossing the discontinuity. Small
assessing the performance of thick suction-slot airfoils, in-
changes in the body and slot contour may have substantial
cluding the suction pump power, on the basis of Taylor's
effects on the actual minimum CQ and CH* This w^ould have
simple criterion. An equivalent total drag is formulated,
to be shown by an organized wind-tunnel development pro-
comprising the actual wake drag and the suction drag (from
gram. It remains only to consider the boundary-layer
the pump power). No attempt is made to include propul-
momentum thickness 02 downstream of the discontinuity.
sion, since it is an airfoil section that is being studied. As
Unfortunately 62 is not given directly in Ref. 4. It is calcu-
it is well-known, suction fluid can be utilized, once aboard
lated from the wake drag. The ratio 0 2 /0i is quite sensitive
the vehicle, to increase the propulsive efficiency by substan-
to the suction coefficient CQ and to the upstream boundary-
tial amounts.
layer profile, as shown in Fig. 11. Two theoretical curves
The suction-flow coefficient CQ is defined as follows, on the
are plotted for N = 5 and N = 7^ where N is the exponent
basis of the total suction mass flow ms given in Ref. 4:
of the turbulent boundary-layer power profile,
1
^Q = (9) u/U, = (11)
The N values were selected for best fit to the experimental
At RL = 107, the approximate relation holds: CQ = 68.5 points. At the experimental Reynolds number, CQ should
ms. On the other hand be less than 3 according to theory and, therefore, 02/0i
should be approximately 2.0. Actually, since the experi-
(10) mental CQ were higher, 02/0i ranged from 0.10 to 1.00.
The boundary-layer-control analysis is concluded with the
knowledge of 0i/L, CQ, CH, and 02/0i for the given body
Figure 9 displays the suction coefficient CQ =
velocity distribution. However, in practice CQ, CH, and
against Re, where Q is taken according to the Taylor criterion
02/0i depend more or less strongly on suction-slot design.
per unit length of slot periphery. The wind-tunnel points
The effect of suction-slot width and slot shape was investi-
gated to some extent in the wind tunnel by Cerreta.4 The
THEORETICAL AT BLC SLOT (83%L)
IO%L TRANSITION drag coefficients at RL = 1.0 X 107 are shown in Table 3.

WIND-TUNNEL DATA AT 79.5%L, IO%L TRANSITION

40 50
R L *IO
Fig. 8 Boundary-layer momentum thickness upstream Fig. 10 Pressure parameter for boundary-layer-control
of boundary-layer-control suction slot. suction.
 F. R. GOLDSCHMIED J. HYDRONAUTICS

2.5
^---^THE
.ORY OF ARC R 8 M 2 577
of the order of only 2! This possibility of tradeoff between
= 0.50 suction power and thrust power opens new ways for the over-
2.0
V\ \ BOUNDARY-LAYER PROF-ILE ^-(f)* all optimization of a self-propelled vehicle.

e2 '5 \
N = 5'
,N.'7
5. Power System Analysis
e,
1.0 \\ \*t S
The successful integration of hull design, boundary-layer
control, and propulsion requires a good understanding of the
EXP RIMEN- AL DA1A power-system analysis. The first concept to note is that the
0.5 - OF DTMB A ERO 91 t^ moving vehicle puts energy into the stationary fluid. The
S
^^
*-*^_j f boundary layer is a reservoir of kinetic energy to be tapped
0
(3 1 2 3 4 3 6 7 8 9 1C and regenerated. The second concept is that power can be
used to control this energy transfer between vehicle and fluid,
given a suitable hull design. The third concept is that
Fig. 11 Ratio of boundary-layer momentum thicknesses propulsion power is minimized when the boundary-layer-
across boundary-layer control suction slot.
control suction pump brings a fluid mass flow ms to zero
relative velocity with the propulsor (US1 = 0). The reactive
It is seen in Table 3 that the g/L = 0.008 slot gives the thrust is given by
lowest suction power and the lowest total equivalent drag T = ms[US2 - USi] = mst70[(J7S2/t7o) - (UJU,)} (12)
coefficient. In most cases the wake drag is small in com-
parison with the equivalent suction drag, in the ratios from Assume that T/msllQ = 0.04 (typical value for present de-
1:7 to 1:12. It is to be noted that the suction-flow coefficient sign), with US2/UQ = 1.04 and USi/U0 = 1.0. The power
CQ is constant for all slot widths g/L, in the vicinity of 5.5. required will be
Ww = (13)
4.3 Aftbody Boundary Layer and Wake Drag

The aftbody is entirely under a favorable pressure gradient,

ww - 0.08 (14)
the velocity ratio rising from 0.6 to 0.8. Therefore, the com-
putation of the momentum thickness 6% should not present Assume now that while T/msU0 = 0.04 still, the velocities are
any difficulty, once 62 has been established. Equations changed to US2/UQ = 0.04 and to USl/Uo = 0. The power
(4, 5, and 7) should be used. The boundary layer will always requirement now is
be turbulent, even at comparatively low Reynolds numbers.
The form factor H = d*/d should remain in the range 1.4-1.5.
In the actual design of a complete self-propelled body, there = 0.0016 (15)
will be a propulsion jet issuing from the stern, which will
entrain the aftbody boundary layer as it leaves station 3 which is smaller by a factor of 50! This means that propul-
(see Fig. 5). This will act to reduce 63 and, therefore, the sion is practically free when the boundary-layer-control
wake drag. Because of the lack of experimental informa- power has been supplied and paid for (i.e., U81 = 0).
tion, this effect has not been included in the analysis. This observation is similar in nature to that of Wislicenus26
Figure 12 shows the relationship between the wake-drag concerning boundary-layer intake for open or shrouded
coefficient CDW and the momentum thickness 6i/L down- stern propulsors. In the latter case the natural boundary
stream of the discontinuity. Since the wake drag has been layer (only partially energized to flight conditions) is ex-
measured in the wind-tunnel tests of Ref. 4, this curve has ploited for propulsion, while in the former the boundary-
been employed to calculate 62 (which was not measured). layer-control discharge, fully energized, is presented to the
Figure 11 shows how #2 can be controlled by CQ. This propulsor.
means that CDW can be controlled by CQ and that the wake It is seen in Table 4 that the boundary-layer-control
drag can be reduced by a factor of 10, for a suction increase suction-power is much larger than the propulsion power for
the data of Ref. 4 (g/L = 0.008). It represents by far the
major portion of the total power requirements.
Therefore, there is available a comparatively large mass
flow ms within the vehicle with zero relative velocity, which
is begging to be used for propulsion and to be discharged with
a finite U8Z from the stern. It would not be very practical
to discharge ms from the vehicle at Usz = 0 in any event!
The reader is referred to Edwards27 for an excellent discussion
CD on the fundamental aspects of propulsion for boundary-
layer-control vehicles, which needs no repetition here.
The pumping system comprises the annular suction slot
and a short annular duct leading into an axial pump stage.
The discharge of this stage is assumed to be just sufficient
to bring the fluid to freestream conditions, since any output

Table 3 Slot-width effects

C Dw CD
0.016 0.0160 0.0013 5.75 0.0173 12.3
0.016 0.0161 0.0014 5.78 0.0175 11.5
0.012 0.0176 0.0054 4.80 0.023 3.26
0.012 0.0153 0.0032 5.55 0.0185 4.80
0.008 0.0149 0.0019 5.35 0.0168 7.85
Fig. 12 Wake-drag coefficient vs boundary-layer mo- 0.008 0.0147 0.0020 5.35 0.0167 7.35
mentum thickness after suction slot.
JULY 1967 INTEGRATED HULL DESIGN

increment above this level is to be charged to propulsion Table 5 Power coefficients

rather than to suction.
For large installations, such as submarines, the over-all RL CpS CPw CP CP* = Cp/Tlp

pumping system efficiency may be expected to be 90%. For s

4. 5 X 10 0. 0179 0. 0000625 0. 017962 0. 01995
small installations, such as torpedoes, the system efficiency 4.41 0. 0197 0. 0000185 0. 019718 0.,0219
probably will not exceed 80%. The propulsion system con- 4. 51 0. 0193 0..000046 0. 019304 0. 0214
ceptually comprises another and separate pump stage and 4.6 0. 0191 0.. 000003 0. 019100 0..0212
an exit nozzle. Actually there will be only one axial impeller 7. 15 0. 0158 0..0000775 0.,015877 0,.0176
for both duties. 7. 20 0. 0166 0,.0000136 0..016613 0..0184
The suction mass-flow coefficient Cm and the suction- 7..10 0. 0167 0. 0000065 0,,016706 0. 01855
7. 10 0. 0177 0.0000048 0.,017704 0.01965
power coefficient CPs = CDS are given in Table 4 from the 10. 15 0. 0149 0.0000321 0.,014932 0..0166
experimental data of Ref. 4. It remains to compute the 10..0 0. 0147 0. 0000350 0.,014735 0,.01635
US2/U0 ratio and the propulsion power Ww (on the assump- 4,.36 0. 0207 0..0000009 0.. 020700 0..02300
tion that US1/UQ = 0.) The computation of Cm is of interest, 6,.95 0,.0157 0.0000491 0.015749 0.0175
given ms from the data of Ref. 4, 2 .10 0..0171 0.0000112 0,.017111 0.. 0190
7,.05 0..0180 0.0000037 0,.018003 0,,0200
1 6,.95 0. 0177 0.0000045 0., 017704 0,,0197
Cm =
RL 18.1 9,.8 0. 0146 0. 0000655 0,.014665 0..0163
10..2 0.,0147 0.0000587 0,,014758 0..0164
and 10.,1 0.,0152 0.0000578 0.,015257 0.,0169
11..3 0..0142 0. 0000388 0.,014238 0..01585
(17)
msUQ 2ms
(18) inversely proportional to the total suction mass-flow coeffi-
v~ cient. The suction-power coefficient Cps, the propulsion
power coefficient CPw, the total power coefficient CP, and the
(19) mechanical power coefficient Cp* (assuming r)P = 90%) are
shown in Table 5 with EL, for the case of g/L = 0.008.
Using a set of experimental values at the highest RL condi-
or
tions,
lCDw
(20) = 1.20 CT = CDw = 0.002
2 Cm
L 2/ 7 2/3 = 6 95

and
CP, = 0.0142 7rdi/L = 0.615
(21) Cn = 0.00386
Then
L2 \ /
(22) 4 X 10~6
2 qUo V^J \ CPw = - 3.71 X 10~5 (25)
3.86 X 10-6 X 4 X 6.95
S!
= Cm
L*\/t/ F/2/3\ 2

iv and
(23) US,/UQ = i [0.002/(0.00386 X 6.95)] = 0.0372 (26)
It is to be noted that the propulsion jet needs to be only
3.7% of the flight velocity. The total power coefficient will
be
In other words, the propulsion power coefficient Cpw is di-
rectly proportional to the square of the thrust coefficient and CP = CPs + CPw = 0.0142 + 0.0000371
(27)
CP = 0.014238
Table 4 Experimental drag coefficients Assuming a pump efficiency rjP = 90% (for a large installa-
tion), the total mechanical power coefficient will be Cp* =
RL Cm CDS CDW CD
Cp/rjp = 0.01585.
6 3
4.51 X 10 4.55 X 10- 0.0179 0.0028 0.0207 There is a tradeoff possible between suction power and
4.41 6.32 0.0197 0.0018 0.0215 propulsion power. Figure 11 indicates (N = 5 curve) that
4.51 6.35 0.0193 0.0009 0.0202 it is theoretically possible (although not achieved in Ref. 4)
4.60 9.45 0.0191 0.0003 0.0194 to reduct CQ from 4.0 to 2.5, thereby doubling 02 from 62/61 =
7.15 4.17 0.0158 0.0030 0.0188 1.0 to 62/61 = 2.0.
7.20 5.18 0.0166 0.0014 0.0180
7.10 5.50 0.0167 0.0010 0.0177 Thus CDS = 0.00885 (assuming constant CH)', CDw =
7.10 6.00 0.0177 0.0009 0.0186 0.004. Now
10.15 4.03 0.0149 0.0019 0.0168 UJUQ = 81.5((7z),/CQ) - 0.072(CW/(7m) = 0.13 (28)
10.0 4.12 0.0147 0.0020 0.0167
4.36 10.1 0.0207 0.0005 0.0212 and Cpw = 0.0003. The total mechanical power coefficient
6.95 4.20 0.0157 0.0024 0.0181 will be
7.10 5.40 0.0171 0.0013 0.0184
7.05 6.23 0.0180 0.0008 0.0188 0.00885 + 0.0003 0.00915
6.95 6.43 0.0177 0.0009 0.0186 - 0.0102
0.90 0.90
9.8 3.98 0.0146 0.0027 0.0173
10.2 4.14 0.0147 0.0026 0.0173 This tradeoff has reduced the total mechanical power by
10.1 4.23 0.0152 0.0026 0.0178
11.3 3.86 0.0142 0.0020 0.0162 35%. It appears then the suction power for boundary-
layer control should be reduced and the wake drag allowed
10 F. R. GOLDSCHMIED J. HYDRONAUTICS

with improved hull and suction-slot design, the power co-

efficient of the new H-BLC-P design may be decreased to
CP* = 0.0102. Comparing now both second-generation
vehicles, a power gain of 45% is made, on the basis of the
conventional body. In other words, the second-generation
situation will favor still more the new H-BLC-P design.
The preceding discussion pertains only the zero angle of
attack. For the new design, it is claimed that conditions
will remain essentially unchanged up to a specified angle
(for instance, 7) and the only penalty would be a peripheral
flow asymmetry to the pump inlet, perhaps affecting the
pump efficiency. It is quite unfortunate that the wind-
tunnel tests of Ref. 4 did not extend to a other than zero,
8 9 10 II 12 13 14 15 16 17
and that inferences must be made on the basis of two-dimen-
R L xicr e
sional tests only. On the other hand, conventional stern-
Fig. 13 Comparative drag of bare hulls without propelled vehicles are quite sensitive to angles of attack, as
propulsion. shown by McLemore.2 For a = 7 the propeller thrust
coefficient decreases 12% while the drag increases 40%.
to increase to an approximate ratio CDS/CDW = 2. To Another comparison may be made on the basis of hydro-
achieve this, however, it will be necessary to change some- dynamic noise, i.e., hull noise and wake noise. The hull
what the hull design in the vicinity of the discontinuity, noise is given by the turbulent boundary-layer pressure
to improve the suction-slot effectiveness. This should fluctuations, which are a strong function of the boundary-
certainly be one of the major objectives of future develop- layer profile parameter H = 5*/d as presented in Fig. 14.
ment. For the conventional hull the pressure fluctuations will
reach 7% to 10% of freestream dynamic pressure, while for
6. Comparison with Conventional Bodies the new design H will always remain between 1.4 and 1.5,
thereby assuring pressure fluctuations only of the order of
The most conventional self-propelled submerged bodies 1%. The wake noise is proportional to the external drag
today are high-fineness-ratio hulls (more or less stream- and, therefore, to the thrust generated. For conventional
lined) with stern propulsion by an open wake propeller. vehicles the entire mechanical power is used to create thrust
Wake propellers have been shown to have substantially by open propellers in the noisiest possible manner. In the
improved propulsive efficiencies as compared to freestream new design the total power is reduced (as much as 50%),
propellers, as demonstrated experimentally by McLemore2 the external thrust is quite small (less than 10% of the con-
and theoretically by Wislicenus26 and others. However, ventional thrust), and a shrouded propeller is used. Thus,
it has been found that stern propellers cause a drag increment the over-all noise situation will be much improved by the new
on the hull, of the nature of a pressure drag. This increment H-BLC-P design.
ranges from 7 to 20%, depending on the body and propeller
design, and it detracts, therefore, directly from the pro-
pulsive efficiency of the hull-propellor combination. 7. Conclusions
A comparison between the integrated hull/boundary-layer-
control/propulsion design (H-BLC-P) and conventional A case has been made for the integration of hull design,
stern-propelled vehicles can be made in several ways, de- boundary-layer control, and propulsion of self-propelled sub-
pending on the viewpoint. A traditional comparison may merged bodies on the basis of partial wind-tunnel tests. It
be made between drag coefficients of bare conventional appears that large power gains (of the order of 50%) can be
bodies and equivalent drag coefficients (suction plus wake) made as compared to the best streamlined body with stern
of boundary-layer-control bodies. Propulsion is not con- wake propellers and that the power gains are much larger at
sidered in either case. Such a comparison is shown in Fig. angles of attack within a specified range and also at higher
13 between the new H-BLC-P design and the Akron model, Reynolds numbers.
a blimp model, and a torpedo model. The drag gain here is
impressive, 32% at RL = 1.7 X 107 on the basis of the
Akron, 38.5% at RL = 1-7 X 107 on the basis of the blimp.
However, the comparison here is not too meaningful because
nothing has been said about propulsion. It is suggested
here that comparison be made only on the basis of the self-
propelled vehicle test, where the mechanical power input
to the propulsor can be measured or at least estimated.
For instance, for a 5:1 streamlined body and stern pro-
pulsion, McLemore2 reports a CP* = 0.020 at RL = 1.75 X
107. This results from a CD = 0.021 and an apparent pro-
pulsive efficiency of 105%, on the basis of enclosed volume.
The incremental drag has been found to be 19%. The na-
ture of this drag is demonstrated by the pressure distribution
on the hull. Correcting the results to RL = 1.1 X 107 it is
obtained CP* = 0.0225/1.05 = 0.0215. For the new H-
BLC-P design, at the same RL it is found CP* = 0.01585
thereby showing a mechanical power gain of 26% on the SEPARATION
basis of the conventional vehicle. -*) RANGE

McLemore2 finds propulsive efficiencies up to 120% at 1.6 1.7 2.0 2.1 2.2
reduced thrust and suggests that improved propellers may BOUNDARY-LAYER PROFILE FORM PARAMETER
be designed to reach such efficiency at full thrust. In this
H=-f*
case the power coefficient is obtained: CP* = 0.0225/1.20 =
0.0188. By the same token, it was shown previously how, Fig. 14 Turbulent boundary-layer pressure fluctuations.
JULY 1967 INTEGRATED HULL DESIGN 11

Complete large-scale wind-tunnel tests of a self-propelled Lachmann (Pergamon Press Inc., New York, 1961), Vol. I,
model (such as carried out by McLemore2 for the conven- pp.106-108.
12
tional streamlined body with stern wake propellers) are Thwaites, B., "Slot suction," p. 232; also "Slot suction on
needed to provide satisfactory experimental evidence for aerofoils specially designed for very low drag," pp. 243-249,
Incompressible Aerodynamics (Clarendon Press, Oxford, England,
this design integration concept. A development program 1960).
should be planned for three fineness ratios, 1.5:1, 2:1, and 13
Schlichting, H., Boundary-Layer Theory (McGraw-Hill
3:1 and for discontinuity locations as far back as 90%. Book Company Inc., New York, 1960), 4th ed.
Particular attention should be paid to the suction-slot de- 14
Gregory, N. and Walker, W. S., "Further wind-tunnel tests
sign and the hull contour in the vicinity of the discontinuity, on a 30% symmetrical suction aerofoil with a movable flap,"
so as to minimize the minimum value of CQ sufficient to British Aeronautical Research Council R&M 2287 (July 1946).
15
stabilize the boundary-layer crossing the pressure rise. Fol- Goldschmied, F. R., "Incompressible potential flow velocity
lowing the aerodynamic development, preliminary design distribution over bodies of revolution," Goodyear Aircraft Rept,
studies should be made for several typical submarine ap- GER-5235 (1953).
16
Thwaites, B., "Uniform flow past bodies of revolution,"
plications in order to explore the new design freedom allowed Incompressible Aerodynamics (Clarendon Press, Oxford, England,
by the present concept of hull/boundary-layer-control/ 1960), Chap. IX, pp. 369-418.
propulsion integration. 17
Cheers, F. and Rayner, W. G., "Tests in the N. P. L. electric
tank on a 4:1 axisymmetrical diffuser having a discontinuity in
the wall velocity," British Aeronautical Research Council Cur-
References rent Paper 2 (March 1946).
18
Goldschmied, F. R., "Calibration of the three-dimensional
1
Goldschmied, F. R., "Proposal for the study of application electric analog tank," Westinghouse Electric Corp., ADR-1 (De-
of boundary-layer control to lighter-than-air craft," Goodyear cember 1954).
19
Aircraft Rept. GER-5796 (1954). Freeman, H. B., "Pressure distribution measurements on
2
McLemore, H. C., "Wind-tunnel tests of a -J^ scale airship the hull and fins of a -^ scale model of the airship Akron," NACA
model with stern propellers," NASA TN D-1026 (January Rept. 443 (1932).
20
1962). Alien, H. J., "Pressure distribution and some effects of vis-
3
Goldschmied, F. R., "A theoretical aerodynamic analysis cosity on slender inclined bodies of revolution," NACA TN 2044
of a boundary-layer controlled airship hull," Goodyear Aircraft (March 1950).
21
Rept. GER-6251 (September 1954). Rodgers, E. J., "Real flow over a body of revolution at
4
Cerreta, P. A., "Wind-tunnel investigation of the drag of a angle of attack," Pennsylvania State Univ., Ordnance Research
proposed boundary-layer-controlled airship," David Taylor Lab. TM 5 2420-13 (March 1963).
22
Model Basin Aero Rept. 914 (March 1957). Boltz, F. W., "Boundary-layer transition characteristics
5
"An investigation of a boundary-layer-controlled airship," on two bodies of revolution, a flat plate and an unswept wing in a
Goodyear Aircraft Rept. GER-8399 (October 1957). low-turbulence wind-tunnel," NASA TN D-309 (March 1960).
6 23
McCormick, B. W., Eisenhuth, J. J., and Lynn, J. E., "A Truckenbrodt, E., "A method of quadrature for calculation
study of torpedo propellersPart I," Pennsylvania State Univ., of the laminar and turbulent boundary-layer in case of plane
Ordnance Research Lab. Rept. NORD 16597-5 (March 30, 1956). and rotationally symmetrical flow," NACA TM 1379 (May
7
Gearhart, W. S. and Henderson, R. E., "Selection of a pro- 1955).
24
ptilsor for a submersible system," J. Aircraft 3, 84-90 (1966). Gregory, N., "Note on G. I. Taylor's criterion for the rate
8
Goldschmied, F. R., "An approach to turbulent incompress- of boundary-layer suction at a velocity discontinuity," British
ible separation under adverse pressure gradients," J. Aircraft Aeronautical Research Council R&M 2496 (May 1947).
25
2,108-115(1965). Preston, J. H., Gregory, N., and Rawcliffe, A. G., "The
9
Lighthill, M. J., "A new method of two-dimensional aero- theoretical estimation of power requirements for slot-suction
dynamic design," British Aeronautical Research Council R&M aerofoils, with numerical results for two thick Griffith sections,"
2112 (April 1945). British Aeronautical Research Council R&M 2577 (June 1948).
10 26
Richards, E. J. and Burge, C. H., "An airfoil designed to Wislicenus, G. F., "Hydrodynamics and propulsion of sub-
give laminar flow over the whole surface with boundary-layer merged bodies," ARS J. 30, 1140-1148 (1960).
27
suction," British Aeronautical Research Council R&M 2263 (June Edwards, J. B., "Fundamental aspects of propulsion for
1943). laminar flow aircraft," Boundary Layer and Flow Control, edited
11
Head, M. R., "Aerofoils with suction applied at the pressure by G. V. Lachmann (Pergamon Press Inc., New York, 1961 \
minimum," Boundary Layer and Flow Control, edited by G. V. Vol. II, pp. 1077-1100.