The Lithgow Propeller
Water Tunnel
BY
A. EMERSON,. M.SC. Wand L. W. BERRY
The rponsibi1ity for the 8ttements and
opinions expressed in papers and discussions reets with the individual authors; the Istitu-ionis a body merely places them on record.
/
/.
Repted from the
Trcznsaction.gof the In8titugjon of Engineers and Shigibuilders in Scotland
Paper No. 108ã
THE LITHGOW PROPELLER WATER TUNNEL
-
By A. EmsoN,* M.Sc. and, L.. W. BEREY*4th March, 1947
INTRODUCTION
Tm propeller water tunnel was gien o the National Physical Laboratory by Sir James Lithgow in 1932 to complete the
essential test facilities at Teddington. Before this date it wa
not possible t' reproduce on the model scale the pressure con-dlitions existing at he ship propeller and it was difficult to make yisual observations on models immersed in water. The variable
pressure water tunnel has, in some- measure, provided both
these facilities.
In 1942 advantage was taken of a lull in the pressure of other. -work to consider modifications to the tunnel design and use.
Some'time was spent in preparing apparatus for measuring forces
on bydrofoils, but it was found that changing fromhydrofoils to
screw testing wasted too much tithe. Apart from this diversion the development programme has been carried out as opportunity allowed with emphasis on the changes which caused least inter-ference with work in the, tunnel and which were of the greatest general improvement in propeller testing. The immediate use of this work has been to make the tunnel available for standard "open water" tests on -model propellers, thus releasing' one of' the tanks for other 'model tests.
-DESCRIPTION or THE TUEL.
-The tunnel consists essentially of a pipe circuit. arranged to give a length of measuring section through which . a uniform
TUE LITEGOW PROPELLER WATER TUNNEL 503
flow of water is obtained. An impeller of suitable design is used to circulate the water; a contaction in the cross sectional area
before the measuring section removes irregularities in the flow; a'fter the measuring section the pipe area is increased t reduce
the velocity of the water and tlrus to reduce the frictional
resistance.
The general arrangement of .the tunnel is shown in Fig. 1.
Starting from the top left-hand corner, above which there is a "free " surface or air space, the wate passes tljrough a 6 to 1 contraction into the measuring length. The cross section here. - is 18 in. square isith corners of 3f in. radius. After themeasuring
section the pipe bends through 90° into the downward expanding
leg. There is an increase in cross section in the bend and the flow is steadied by a central curved vane. Immediately áfter the bend a short length of thr'pipe is used to change thesection
from square to circular and the remainder of the downward
leg is conical (90 angle of cone). At the bottom thereare three, guide vanes round the'90° bend into the impeller section; aftth the impeller the size of the section is ináreased and -the shape
changed to rectangular. The two remaining - coriers of the
circuit are similar; each consists of a 60° bend and a 30° bend and there are three guide vanes in the bends. Underthe "free surface " the top line of the tunnel is maintained by a grid plate.
Most of the tunnel sections are fabricated from i-in. steel
-plate with welded stiffeners. The measuring section and two corners are castings. The sections are bolted together with paper and bitumen joints, and the inside of the tunnel is painted with bitumen. In the measuring section there are three glass
windows carried in frames, which are sqrewed up against packing.
The thickness of the packing is adjusted to give a flush inside
surface to the tunnel.
-The water is circulated by a four-bladed impeller lriven by a D.C. motor developing a maximum of 80 h.p. at 320 r.p.m.
The drive is direct, and the shaft is carried by a cutless rubber
bearing and then .passes through the tunnel wail
in a water
lubricated packing gland. 'The air space above the free surface
is augmented by a large air reservoir.
This minimizes the effects of small air leaks when the pressure is reduced by thevacuum pump. A reserve water tank is used to empty the
upper prt of the tunnel so that changes in apparatus can be'P LITHG0W PROPELLER WATER TuNNEL 505 made without changing the water in th'O circuit. The water is -drawn back from the 'reserve tunk by suction. .
TuNNEL DEVELOPMENT AND PERFORMANCE
Impeller. The original impeller. was obviously causing
con-siderable vibration and unsteadiness. A new impeller was
designed by standard 'methods' and an approximate cakulation showed that its efficiency should be 75 per 'cent. instead of the
25 per cent. efficiency of the existing impeller.
It was not
posible tb obtain a broize impeller 'at the time so a wc)oden
-one was built up from glued boards. With this wooden impller
-the speed was restricted to 12 ft. per sec. in order to avoid damage
to tlie blades;
this restriction did not affect the testing ,ofscrews u,nder reduced pressure. 'The calculated performanoe was obtained And the improvement in tunnel' conditions was
obvious from the decrease in noise and vibration. After, th wooden impeller had hen in use for 18 months te Manganese
Bronze & Brass Co. found time to make the impeller in man-ganese bronze; This gift was a much appreciated encouragement
to proceed with tunnel improvement.
The impeller blde
sections were left in their ,unflnisiied Oondition and though this
Teduced the efficiency. it allowed scope for future experiments on 'roughness or shape of blade section at a higher Reynolds' number
'than is normally available. under laboratory conditions. The
-impeller design is' discussed in Appendix 1.
Tunnel Cirejuit. 'With the wooden impeller the conditions
were sufficiently steady for the changes in velocity' to be observed.
In ihe measuring section the speed records, at constant impeller motor setting, showed a series of steady periods separated by
large fluctuationsup to 2 per cent. variation in speed. This
type of record was obtained at all speeds. To find the -source
of the fluctuations static pressure tappings were taken at the
inner and outer walls of the tunnel at different sections. At the end of the 90° bend after the measuring sections the pressures varied rapidly and there was a steep gradient aCross the section. The presure variations at the lottom of the downward
expan-sion were out.of phase with 'those in the measuring section.
It is believed that the flow round the top bend has ,a steep
. lSee'Biiliography, p. 523.
'2N
506 mnn iirnaow PROPPLT ER WATER TUT1EL
velocity gradient from inside to outside of the bend and that in the expansion the flow becomes unstable
-When running at high vacuum the static pressure in the
measuring sè0tioi is very small and the additional pressure drop occurring at the 90°' bend reduces the preisure to below water vapour pressure. Water vapdur ançl air collect at the top of.the bend and the tunnel "chokes '---any increase in power
input is taken up by the rapid increae in rQsistance of the
corner and there is little increase in, speed.
-A computation Of the resistance or loss of. head in the various
parts of the tunnel showed that 12 per cent. of the nergy loss occurred in the measuring section, 50 per cent. in the top 0° bend and 25 per cent. in the downward expansion. For minimum
resistance2 in the bend, the vaiue of width/depth should bè 24
(instead of 2) and the ratiO of radius/depth hould be large
(preferably 6 instead, of 3). Th alterations required were an expansion between the mesuring section and the beginning of the bend ; an increase in radius of the bend and alterations to
the shape of sectioñ and plaing of splitter vane; a reduction
in shape of the expanding curve and subdivision of the expansion
into compartments to reduce flow breakdown. None of' these changes could be carried' Out without serious interference with other work and it would be necessary to make a model of this part of the tunnel' before acluaUy making alterations. Instead the experimental technique has been modified, and records are
-taken only during the steady speed periods. This method
requires more care and patience but gives good results:
Tunnel C1jaractrstics.
When the water is circulated, the
static pressure in the measuring section fails as the velocity isinèreased. If the pressure at the free surface is reduced the
limiting tunnel velocity is reached when the static pressure p
- '
, equals the water vapour pressure e. This limit is reached first
-in the 90° bendthe corner "chokes" before the measur-ing
section velocity reaches the limiting value.For a given pressure at the free surface, as the impeller speed' is increased the water speed' increases steadily until the limiting speed is approached. Further increase in impeller speed then
produces no material increase in water speed, 'the noise of
cavitation in the corner grows louder and the water level at the free surface rises. 'The impeller motot power input and speed.
t00
THE LITHGOW PROPElLER WATER TUNNEL 507
aoo
0.
I
U.7:
I0 z00
30 a. .300 40 So Go 10. - 20 305PEED IN %PSURIrj& 5UTtOr tN
Fig. 2.Tunnel calibrationcurves.
508 IJI LITHGOW PROPELLER WATER TUNNEL
are plotted in Fig. 2 to a base of speed in the measuring section for different pressures. The velocity distribution across the measuring section varies less than 1 per cent. over the central
12 in. Thre is about 10 obliquity of fiow.
Testing of Propellers. As shown in Fig., 1, th& propeller under
test is carrid on a shaft
passing into the top corner of the tunnel. The shaft is driven by a. 4-h.p. motor with recording thrust meter and torsionmeter and revolution counter in the line of shafting. This recording apparatus is of standard Tankdesign except for a modificationof the arrangement of the bevel
gears in the torque box. The gear is satisfactory for general use
but the friction of the shaft
is rather highabout 0i0
ft.-lb.torqueand has shown variations of 001 ft.-lb. torque. This
-wandering zero becomes very important at low torques. For the normal working range of 3 Lo 6: ft.-lb. torque it is within
the required order of accuracy. The long bearing through
which the propeller shaft enters the tunnpi i carried in a rotating
glnd; the bearing was made long to prevent leakage and the
gland was rotated to reduce friction. Experiments made to
determine the best relative speed-of gland and shaft showed that
the reduction in friction obtained by rotation was small. Since then the gland has not been rotated and the vriation- in shaft friction has lessened. Two furthermodifications are being made, the shaft is to- be carried oh two cutless rubber bearings and the torque measured by balancing the stator reaction of the driving motor, thus eliminating two bearings and the torque box.
The water speed is measured
by the pressure drop iii the
contraction leading to the measuring section. A propeller an&
.rnometer has been hsd to record the speed, but it was found
-that, particularly under reduced pressure, the presence of the anemometer affectd the screwperformance. The flow- into the
screw disc is affected by the constraint of the {unnel walls, and the results are corrected to equivalent open-'ater conditions by a calculated "channel speed correction" depending on the dia-meter of the screw and the thrust coefficient. With this correc-tion the thrust and torque curves obtained in the tunnel agreed with the open-water curves obtained in the Tank for screws of 9 in. thani. Model screws made specially for tunnel testing are usually 7 to 8 hi. diam.
THE LTH0W PROPELLER WATER TUNNEL 509 made depends at present on the strength of the model propeller blades and it can be raised by change in propeller metal. Ob-servatiori of cavitation on screws shows the importance of small defects in making the model propellers. A screw-cutting gear has been designed to help in the manufacture of more accurate
pIopellers, possibly in harder metal. This apparatus is described
in Appendix II by Mr. N. A. Witney who is responsible for the
detailed design. It allows t1e accurate cutting of the designed
radial blade sections. .
-
Further increasedn water speed in the measuring section could be obtained by inserting an "open jet" section, but this would 'give rise to a new set of problems.Illuininatioi 'and Photography. Screws under test are observed
by strob9scopic lighting triggerçd. from the shaft so that the
screw is ifiunlinated oncce per revolution.
At present the
Edgerton type discharge lamps used give"one flash per revolution
of adequate intensity for observation and also give single flashes
pf nluch. higher intensity to allow photographs to be taken.
The lighting arrangement is described inAppéndix III by Mr. A. I. Williamswbo is responsible for the development of lighting and photography. The, possibility of using lamps triggered in series to _give successive pictures showing the formation and
-collapse of cavitation bubbles is being considered and provision
has ben made for bigger windMv space in the measuring section.
CAvITATIoN Txsrs ON SCREW PROPELLERS
To investigate the possibility of cavitation on the ship
pro-pellers it is first assumed that the ship wake in the neighbourhood
of the propellers can be replaced by a thüform mean wake.
For a normal twin-screw ship with well-designed boésings this
first approximation is sufficiently accurate. There is no method of correcting results for a n,on.uriiform wake.; experiments with gauzes of different mesh to produce a velocity variation in the tunnel have been considered.
For a uniform wake, if V1 is the speed of advance of a
ship propeller of diameter 0 at r.p:m. N, and if the static
'pressure at the shaft centre is H (water head) +A (atmospheric pressure) then the conditions to be reproduced in the tunnel are1O TflE LITUGOW PROPELLER WATER TuNNEL
cavitation constant (HfA)fV12 as for the ship propeller. More precisely, the cavitation number used in the presentation of the
results has been- taken as
-=(pe)/pvA2
where p=statie pressure at the shaft centre e =water vapour prcisurep = density of water
vA=relative - speed of blade section at 07 tip radius.
Q(orJ)=v/nd where v=speed of advance
n=evs, per miii.
-- d=diameter of screw.
The model. screw is made approximately 8 in. in diam. and the water speed used is usually about 10 ft. per sec. The speed
of rotation is fixed by the ship value Of .VfN]) -and the piessure in the tunnel is adjusted to give the same cavitation number as
for the ship propeller.
easurements are taken of thrusf,
torque, speed -of rotation, water speed and the pressure at the
shaft centre. Observations or photographs are taken of any
cavitation phenomena on the screw blades.
-The model ,propeller chosen to ifiustrate cavitation tunnei
results is a three-bThde right-hand screw, diameter O.537 ft., constant pitch 0597 ft., nearly effiptical blade shape, B.A.R. of O45, circular back sections, 12° rake arid 0076 blade thickness-ratio. The thickness ratio of the section at 07 radius is 00735 'and there are some results available3 on a- circular-back section
of thi
thickness ratio . at approximately the same Beynolds' number over a range of cavitation numbers.The results of the sôrew experiments in the tunnel are shown.
in Fig. 3 as curves of thrust coefficient KT, and torque coefficient KQ for onstant values of C to a base of cavitation number; in Fig. 4 the KT and KQ curves are plotted to ,a base of CD.
Figs. 6 to 18 show typical appearances of cavitatjon. At the
high slip (low cD value) th section at 0-7 radius is at
approxi-mately i ° incidence ; at the high CD value the incidence is about
3°, both figures referring to the stage before cavitation affects
the performance. (As cavitation reduces the lift of the sections the interference- velocity
is also reduced and the incidence,
increased.) At positive incidence the stagnation point is on the pressure side of the section -and cavitation starts on the back
P°INI
/
Q 0.0735
.lIPJ>
01/ 0
04
l.aFig. 4.Curves of thrust and £oque coefficients to a base of C=v/nd.
N.
K -Q =roRJE; Ov;
StM T
e wru SPaE.o,
Tur,
u-O SE Vj aEIATt. O7r sns: vxxa --_z o.o-N
- -/ IO__
-o 0 70 0-90 t 00 -' ITHE LITHGOW PROPELLER WATER TIThIiEL 51
near the leading edge. With negative incidence the cavitatióñ
occurs behind the maximum thickness- of the section. This
brief account, and the results inserted in Fig. 3 of Waichner's
experiments with a blade section corresponding to the screw
section at O7 tip radius, giye a coherent picture of cavitation on the propeller.
The detailed appliation of screw theory tc
determine local velocities, the comparison of .calculated pressure
distribution with observed cavitation and the effect of teynolds'
number on the results' are beyond the, scope of the present
descriptive paper;.
Two other points of interest have been demonstrated with, the
same propeller
-T]ie ship propeller operates in sea water which is saturated.
with dissolved.' air.
In the tunnel the reduction in prssure
removes mdt of the dissolved, air 'and to. obtain consistentresults the air is removed before starting experiments. The
difference in result on account of tht test procedure is shown in
Fig.5.
Specks of dirt in the water seem inevitably to collect on
the leading edge of the model propeller. Such obstructions andd' any small defects-in the model propeller produce, in the cavitation
region, cavitation streaks and affect- the thrust and torque'
values. A cylindrical projection OO5.iiI. dia. and 005 in. highwas fixed to each blade near the leading edge sat' 07 radius. [hè results obtained with this screw and also with reduced
projections are also shown in Fig: 5. While these experiments ,wére made to show the effect of defects in the model propellers
it is clear-that a ship. propeller working just butside the cavitation
region may suffer local, cavitation erosion if the blade leadiig
edge is damaged:'or defective.
,Aknowledgement. The work described in the papel' has been
carried o' as part of the general Feseareb programme of tb
National Physical Laboratory, and. is published by permission
of the Director ,of the Laboratory. The - Authors desire 'to 'acknowledge the assistance rendered by Mr. C. C. A. Burgess,. whose skifi and care in the construction and maintenance of the--, apparatusrequired for the work contributed much to the uccess-,ful completion of the programme of investigation.,
514 PEE LITEGOW PROPELLER 'WATER T1flEL
CkVLTA1ON liUM3
-Fig. 5.Showing effect and of air in wWer.
., S.
S®
=X R.
oo
/
/4'
07/
C MORL;T °
. PRC1tOMS )C '-JT'4 0015 OCTI3r4.;/.
/
0-15- A 1/
0 I a-a 0-3 0 4- 0 6TBE LITHGOW PBO)?ELLER WATE TU1NL 515
- ,APPENDIX I
DESIGN or TUNNEL Iiw1TL1R
The impeller motor is designed to develop 80 h.p. at 320 r.p.m.
From the results obtained with the original impeller it was
estimated that with this power it ,'vould be possible to obtain a
speed of 30 ft. per sec. through the, measuring section which has
an area A=218sq. ft. This gives a flw of 654 cu. ft per.sec. The impeller section in a circle 414 in. diam. Allowing in.
clearance, the impeller diameter D=40,8 in. It can be shown
:that with a boss diameter less than 15 in. the useful work done by the root sections will be very small. These two diameters determine the area A0 of the water stream at the -impeller and
lience tb velocity !u.
If theimpeller is designed togi-e constant pressure increment
p
t each radius r'
..1 dT.1
p= ,pNcrf22CL
27tr-dr. 47r where p = density. N=number of blades ---. c=chord '-f2=angula velocity CL=]ift coefficientor if P is the-power input in'ft.--lb. sec. 47rP=pAOU2NcrOL.
'Using the tip section a value of NCCL=22S is obtained.
The value chosen are restricted because CL must not exceed
-O6 to 07 and the section. must be sufficiently thin to avoid
cavitation. There are. 4 blades. ancF the chord of the tip section is 12 in. For constant pressure increment the value of NCCLJr ii
'constant and this determines the incidenôe of the sectionsat
each radius and therefore the pitcir. . -
-Taking these values of pitch and cboid the actual performance of the impeller is calculated by successive approximation. of the rotational inflow factors and 'assuming that the tip clearance is small enough to avoid tip losses. The calculated -thrust torque/ power and the thrust- distribution are then considered and the
516 TKE iniaaow PROPELLER WATER TUN1EL
root. Here the pressure increment is small and increase of pitch angle increases the incidence slowly s that the increase of the
thrust component is limited. The impeller as made is of constant
pitch. the chord. increasing from 12 in. at the tip to 163 -in. at.
the root.
-Putting. itt the actual speeds and revolutions per minute the design-gives the result shown in Table I.
TAELE
I-Chotd Radius Pitch Blade Resultant. Thrust Pressure Torque
ft. It. ft. angle velocity increment increment increment
10 1:7tip 209 1107 842 - 5890 552 1810 102 16 ,, 1172 902 5340 531 1600 7 F2 ,, 1548 1238 3890 516 1130 1.30 d8 -,, 2254 1992 1820 362 550
i
0625 : 2801 2630 1030 . .262 320 r'oot-giving for u -=833 ft. per sec. -- . L)==347rádians per see.
-- -- . thrust=3840-lb.
h.p.a=745 =78 per cent.
The figures in the table are based- on aerofoil data for smooth
aerofoils at Reynolds' number ==3 x 106. The present impefler sections are "iinfinished" sections. If this is allowed for by
decreasingthe lift coefficients 4per cent. .and increasing the -drag
coefficient 25 per cent. the. thrust falls to 3,580 lb., aM the-.
efficiency to '7,27 percent. (Tbere is a difference of this -order
between - the smooth wood impeller and the unfinithed.
man-ganese bronze.) . - -
-It may-be of interest to- add that a model impeller was
made-and tested in a transparent tUbe in. the tunnel;
Threadsattached to a radius wire behind the impeller showed the direc-tion of the race. Near the root the race velocity was very -small
-. insufficient to hold the thiead straight. Change in pitch had
little effect on the thread behaviour; increase in chord bad more-effect but there was no difference in measured thrust and torque.
THE LITHGOW PHOPELLER WATER TUNNEL 517
Fig. 6 (Reference poini 1 in Fig. 3.)
Fig. 7. (Ref. 4.)
/
Fij. 9. '(Reference p'oinz 34 in Fig. 3.)
Fig. 10. (Ref. 10.)
TEE LITaow PROPEL1 WATER '1ThtNEL 519
Fig. 12. (Rf. 23.)
Fig. 13. (Ref. 31.)
20 mii IiThGOW PROPELLER WATER TuNNEL
APPENDIX II
PoPEu
CUTI'ING GnThe outline arrangement of the apparatus is shown in Fig. 15. Badial se,ctions 5 x model size are drawn at the designed pitch
angles and the drawing fixed to
the flexible drum B. The
drum is keyed to the shaft carrying the model propeller.
The shaft is rotated by a, motor through a 32,000 to 1 worm reduction drive 'A, the .diawing and' propeller rotating with the shaft.
The shaft is moved axially
through the traversingmechanism G controlled by the operator and this motion i
transmitted through the lever F to
the "feeler" carrier D;
which is supported on the rails E. The feeler C is movedvertic-ally to a position agreeingwith the particular radius of the drum.
The screw blades are-cut by knives K driven from 'motors 3 'and carried on a circular knife box H.- The knife for each blade is set up with its spindle at the correct rake and there is a screw adjustment to uet the knife at therequired radius. The " knife"
consists of a single blade cutter and the motion of the knife spindle is so arranged that the point of- the utter moves in a
plane parallel to the screw sh&ft centre line. -.
APPENDIX m
-STROBOSCOPIC LXG-BTLNG-
-In ordd to give a reasonably stationary' image of a model
propeller under test in the - tunnel, the screw should only be observed or illuminated while rotating through -a maximum 'of
10, otherwise a blurred image will result. At a model propeller
- speed of 2,000 r.p.m., propeller will move through 10 in
1/18,000 sec. A mechanical stroboscQpe designed to give a
stationary imag ,under these conditions can obviously only
transmit a maximum of about - 1/400 of the incident light, and. hence is unsuitable for dstailed observation, and ror this 'and other: mechanical reasons is not adaptable for successful photo-,
graph.
Various methods of obtaining photographs were tried, and 'eventually through the good offices of Mr. N. C. Lambourne of
Fig. 15.Screw
cuttihg
522 LITGOW PROPELLEn WATEE TUN1EL
Aerodynamics Division N.P.L., 'vhO was concerned with a
simi1'r problem, contact was made (in 1943) with the Research
Laboratories of the General, Electric Company. These
Labora-tories have designed and made eperimentl high intensity flash
- discharge lamps of various kinds d.uxing'tbe war. Those supplied
to the National Physical Laboratory consist essentially of sealed, quartz or glass tube filled with a suitable. gas uhder reduced pressure, with an anode aid cathode sealed into the
tube at opposite- ends, and a third
triggering ,electrode,.the-position and fitting of which may vary for,different lamps. The stroboscopic type of lamp is suitable both for visual and
photographic observation. The ciiZcuit used for & typical lamp
isasfollows:
''
.A condenser is charged through a resistance from a source of: high voltage. The negative and positive terminala of the
àon-denser are connected respectively to the cathode and anode of the lamp. The triggering electrode is pulsed to the cathode from th'e secondary winding of an induction coil, the primary ciruit of which is interrupted by a contact on the shaft of the model propeller. 'The position Of' this contact S movable relative to
that of-the screw. At each pulse of the triggering eircuit, the
gas in the lamp is ionized and the condenser is discharged through
the lamp. ' The: discharge is extremely short 'in duration,. and results in a high intensity flash. This intensity depends on the
- total electrical' energy stored in the condenser, namely,.
jCV2joules where C=capacity in farads
.11 =voltage when chared
while the duration of the flash is dependent mainly on the capacity C. The charging resistance is a justed tocharge completely the 'cimdenser bet*een two successive. pulses of the triggering circuit.
The lapips at present in use are designed to work at 2,00&
volts, and or visual observation sufficient light is obtained by. sucessiyè discharges of -nif. condenser., giving total energy
per flash of 1 oule. Photographs are taken by increasing the capacity and flashing the .lamp once while the camera shutter is open. A capacity of 13 nif. giving aflash of energy 26 joules
has, been 'found adeqtiate. for this purpose, using film of speed.' 32° Sch., and a lens stop of f/3.5. Thelimitations of this method.
THE LITHGOW PIt0PELLEI WATER TU1NEL 523
are at present imposed by the total available power and the
triggering -circuit.. A larger power pack is under construction,
-and it is hoped at some future time to iiake CiII6 films of pro.
pellers in the cavilintion tunnel. The investigationof the forma-tion of cavitaforma-tion bubbles might be undertaken by discharging - several lamps in rapid succession -during a single revolution of the propeller. In any event, the simplicity of circuit and ease of
operation of the lamps make them capable of muchgreater
use-than at present.
-.
BlEtlooitApHy
-"The Design of Wind Tunnel Fans," by . B. collar. Aeronautical Research Ctte.
R. & M. No.
1889.. 1940. - H.M.S.O.,London. -
-"The Aerodynamics of Resistance," by G. N. Abramovitch. Trans. --of t,he CentralAerodynarthcal Inst. of TProf. Jdkovski. Issue 211. "Profehneseimgen bei Kavitatjij," by Ottd Waichner. Hydro-
-meehhjsche Pro1enie des Schiffinntriebes,
p. 256.
Hm-burg, 1932. - - -
-"Speed-and Power of Ships," by D. W. TyIor. 1943. U.S. Govt.. Printing Office, Washington. - -. -. - -_ "Cavitation Experiments on a Model Propeller," by G. Kempf and
B. Lerb. Trans. Inst. Naval Arch., 1932, vol.' 74,p. 155. -" Methodical Experiments with. Mercantile Ship Forms," by G. S. -.
- Baker. Trans. Inst. Naval Arch.,. 1913, vol. 55,, p. 162.
- ' Some Notes on the Theory of an Airseréw working in a Wind
-Chthmel," by R. M. Wood and R. G. Harris. Aeo. Res. Ctte.
-- B. & M. No. 662. 1921. H.IYLS.O., London.
- -
-- -(8) "Elements of Aerofoil and Airscrew Theory," b H.
Glnuert.- 1926.
-- Cambridge University Press. . . - -- - --(9) "Experiments in the Lithgow Pi'opeller Tunnel," by A. Emerson
and L. W. Berry. Trans. N.E.C. Eng. &Shipbldrs., 194647,
-- vol. 63, p. 333.
-- - Discus&jon
Mr 3 M FERGUSON (Member) There is- little to criticize in this paper as it is a plain statement of fact.and - of work done,
and of the methods. adopted to overcome certain difficulties.
There are one or two doubtful points, hd'wever, on which further -
-information -i desirable. The- Authors state that due to diffi-(I)
.524 'rn:E LrvB:GO* PROPELLER WATER TUNNEL
culties in the existing tunnel records could only be taken during steady speed periods. Great care and patience was necessary but good results 'were 9btained. No assuranceis given; however,
that the records taken during these steady periods were
un-mune from the effect ofthe fluctiatng periods. How long did-.a steady period last, that is to say, how many seconds were available for the recording? And after the fluctuating period -did the next steady period repeat the same result as obtained in the earlier part f the record'? Presuma1y the record would be a wavy line very nearly level, or at a mean value; and then .a violently fluctuating section followed by another relatively
straight section, and so on. Did the straight sections repeat
eaoh other for mean value, or did the fluctuation cause the-next
portion to drop slightly? -
-The writer is interested in the description given of the
ro--tating gland of I the bearing as his firm carried out some tests'
on -the driving shaft of a neiv piece of mechanism for testing large
model propellers in the opn, not 'in açav-itätiowtunnel. There
-was some doubt as-to' which type of bearing would give the
'be4 results, and after 'a number of experiments it was discovered .'that plain bearings were highly efficient. Once the shaft started
rotating it floated in 'the bearings, the- thrust balance became
very sensitite and the friction of the driving shaft remained
very teady. If anything
else was attempted theihrus balance became very sticky indeed. It is therefore interesting to note-that the' very ingenious but complicated rotating gland, has
been abandoned and that the shaft is th becarried in two cutless
:rubber beaxings '
-Can- the AuthOrs give little inore information ,on the channel
-' epeed córrectionl
Appaentiy, to release the new tank at
Teddington for more important work, or,at least work for which :it was 'more suited, open ,experiments were carried out in the cavitation tunnel, where there was not only different recording
jgear and mechanism, btit actually different conditions -of
watér flow from that obtaining in tje tank itself.
It is stated
that the thtust and torque curves obtained in the tunnel agreed'-with the open water curves after this correction had been applied. Was that correction obtained by direct correlation
- ,and then applied toall' the screws tested, or did it depend on
-TEE LITUGOW PROP1TT'R WATER TUNNEL 52
On p. 510 'where symbol is deflhed as the cavitaticin,
number, VA is stated simply as "relative speed of blade section
at 07 tip radius."
Could the Authors add one other line t define VA, or at least draw attntion to the fact that it is theresultant of the forward and the rotational velocities?
In a recent publication of the British .Shipbuilding Research Association an endeavour was maae to achieve some standard
presentation of symbols. C was allocated to a drag constant
and J was allocated purely and simply tO the v/nd of the slip
Why therefore did the Authors chOose CD as the' ymbol for
slip factor? -
-With regard to Fig. , the numbers placed, for example, on
the upper curve of C=O'563 rading from left to right are
1, 4, 7 and 4, and in the illustrations the numbers are i sequence
as Figs. 6, 7, 8 and 9. Thee are in agreement as far as the
numbering on the curve goes, but they are presented in the
reverse logical order in respect of incidence of cavitation, that is,
the -first illustration is full cavitation more, or less, and then
that dies away in the sequence until inFg. 9- there is incipient
cavitation. A rearrangement of the photographs would thake
the sequence a little clearer.
The writer compliments the Authora' on, the propellei cutting
gear described. A few months ago, - while on a visit to the
National Physical Laboratory, he saw this gealL in the preliminary
stage and found It most atiractive. There are no unnecessary fittings or elaborate mechanical devices aboit it, arid compared
with some of the Continental propeller-cutting- ma'ehinea it
appears ridiculously bare They in tirn seem to .be overloaded, with very beautiful and ingenious mechanism, with the result
that only an expert can handle them, whereas the gear described
-in the paper is so simple that anyone can undrrstand it; The
simpler a piece of appratus can be made the better
Is' it:possible, by using a separate st of tracinga or diagrams on the curve table to cut the back of the blade as well as the face?
Mr: J. F. ALLAN, B.Sc. (AssoCiate Member): Although the
paper gives an interesting account of tle development work,
and the teething troubles that have ben experienced in the
tunnel no new results of. any importance are given Oil526 TifE LITEGOW PROPELLEE WATER T1INNEL
time, this is disappointing.. The d'esin of the tunnel is. not
unique in any respect, being similarto the Hamburg and Haslar
tunnels. It has been stated that it is difficult to design a bad
propeller, but is rather astoundiig to learn from the paper that the original impeller had. an. efficiency of 25 per cent.; soiieone
certainly succeeded in designing a very bad impeller.
The information given in egard to the choking at the corner
when the suction is on the tunnel is interesting, but this is a
feature inevitably associated with this design. of tunnel. If, in
order to r'aise 'the velocity and lower the pressure of the test
section, the contracted section method i& used a condition is
created in the water in the" tunnel whih makes the flow very sensitive to any further disurbance, and probably when the water starts to go round. the bend cavitation is set up on the
inside of the beid with the resultnt èhoking of the flow. The
Authors indicate a means whereby thi difficulty could be
overcome and it would appear to be well *orth whil to
in-terrupt the work o the tunnel, for all the time involved, in order
to remedy the fault. The same trouble has been experienced
in tunnels of simikr design, although not at such a low velocity
as that which appears to create trouble in this particular instance.
On p 510 comment is made on the location of cavitation on
blade sections. It is perhaps worth remarking. that the propeller
tested ui this case was of a circular back section. There is
stifi a strong body of opinioi that holds that sharp-nosed sections
are the best to prevent cavitation': that is an opinionwith which
the writer is not in agreement. With these cireular back,
sharp-edged sections, back cavitaVion starts very early behind the
ladin
edge., With a thin round-nosed section, with the mean camber designed for the required lift, and the greatest thicknessforward of niid-áhord, cavitation is delayed in the region of the designed lift and it first appears "on the back behind the
greatest thickness. In this ptsition it has much less effect on the
performance 'than when cavitation occisis behind the leading
edge.
The writer strongly associntes himself with Mr. Ferguson in
his protest against th use of CD for t1e J value. On the top
part of Fig. 4 are shown the curves of KQ with progressively diminishing avitation numbers. These curves of KQ, or for that matter the curves of ..K on the :lower part of the illustration,
4
THE LITHGOW PBOPLLE. WATER T1TNNEL
527-are turning down steeply towards the right, but they 527-are quite obviously not heading for the same zero value on the base line
as the non-cavitation curves for which the spots are plotted. The writer is not saying this is wrong; he is merely stating it is an intetesting feature. It means that the propeUer unde the cavitating conditions is de'veloping a different effective pitch from that which it has as anon-cavitating propeller.
In conclusion, now that the tunnel 'is 'iihctioning properly it is hoped that more and yaluable papers vill be forthcoming from the National Physical Laboratory on the subject 'of. cavitation.
-Prof. 'A. .M. ROBB', D.Sc. (Member of' Council): The writer
enters into any discussion oii avitatiOn with diffidence since
it seems that tjiere is nearly one' opinion per experimenter;
and that' does not make for clear understanding by those who can study the phenomen only at' second-hand. He is, however, muób intrigued by the differençie between the character of the results shown in Fig. 4 and those in' the latest edition of "The Speed and Power of Ships,"4 and also by the comnents of Mr.
8.- S. Cook in a- discussion5 on cavitation. 'In Fig. 4-a reduction
in cavitation number leads to separate curves o K and KQ and to. a fairly gradual approach to mQre or legs completely
developed cavitation.
The other rsults cited are entirely
different. In them there is only one curve of KT' and one curve
of KQ over a large range of slip whatever the cavitation number;
but beyond a critical slip there is a series of separate purves,
branching offsuddenly in the sequence Of the cavitation numbers.
That - difference, in character seems sufficiently important to
justify doubt being cast on CUrrent, theoriOs of cavitation. The
results in Fig. 4, -with the indication of marked differences in, KT and KQ with diffeiince in cavitation number whateer' the
slip, and- with the suggestion of gradual development, seem
rational if the growth of cavitation :8 associated with increasing
-thickness of the bOundary layer Sand progressive travel of the point of separation of flow along the back toward the section' of maximum thickness. Moreover, the gradual development 'appears to be consistent with the gra"dual rounding common to - cirves of lift coefficient for aerofoils as the stalling angle is approached; the stall of an aeroplane is the same- phenomenon as the cavitation of .a propeller. On the other hand at least
528 - THE LITROOW PBOPvrT.FR WAIER
one experimenter who has recorded sudden, changes in slope cf curves of K and K0, and not the separate curves shown on
rig. 4, has noted instability of result, and such instabffity of
result is found also in measurements of lift of aerofoils.
Is it
possible that the difference ip diaracter which has Teen noted can be explained as the election from a series of inconsistent values in the region o instability, or have the Authors a more satisfactory explanation? - -.On the -basis of general principle it would appear that if
Fig. 4 were beyond reproach and: were more reliable than the
,. other results to which reference is made it would seem to be
necessary for purposes of design to rely only on curves of KT
and KQ determined at the appropriate cavitation numbers.
From that there follows another consideration.. The sepai'ation of flow associated with cavitation of propellers and the stalling
of aeroplanes appears to be just the áme phenomenon as the separation of flow at the sterns of models photographed6 by
Dr. Baker. If it is necessary to investigate the separation .of
flow aronnd propeller blades ii
a cavitation tunnel it is
also-necessary to investigate it at the sterns of models in a reduced.. pressure channel. Have the Authors any views on that specu-lation as to a possible development in the technique of the
measurement of resistane. If there is. any basis for the
specu-lation there will be far-reaching consequences.
Dr. J. F. 0. CONN, D.Sc. (Asociate Methber): The Authors are commendably frank. They show .that the existing tunnel
is unsatisfactory in that it " chokes" above .a certain speed
-and that some st uctrral alterations are necessary if the tunnel is to operate efficiently over the full range of speed. It is to be hoped that the necessary modifications will be made as soon
as possible. .- :
it is interesting to learn tht the open-water cliaracteristics of propellers can be measured in the tunnel with satisfactory
accuracy, but the limitation in diameter of model screws (7 to S
in.) is unfortunate, having regard to the desirable Reynold?
numbers. The tunnel correction for epeed mentioned on p. 508
is not a.simple matter. The procedure described is essentially that
given by R. M Wood and R: G. Harris,7 whieh is based upon the simple momentum theory: of piopeller action and,. as the
TEE LITHOOW PROPELLER WATER TUNNEL -. 52
authors of that report said, "may probably be relied upon 'so
long as the correction required is reasonably small."
Are the'
Authors of this paper satisfied that this is accurate undercon-- ditions of cavitation?
One of the. most interesting features. df the paper is the. brief
-description of a new screw-cutting gear, and a photograph of this would be 'a welcome addition to thepaper.
Appendix I on the design of the new impeller gives an iter-esting application of wind tunnel work., To design a propecler or fan, to operate with small clearance within a nircular tube, one can scarcely do better than follow the procedure in R. & M.
18891,' but more explanation might be given as to why 0L must.
not exceed O6 to O7. Might not these figures be increased, by adopting constant velocity sections and will the Authors state
-,which aerofoil sections were actually used?
Mr. W. R. G. WnmNG, M.B.E., M.A. : '., It would be of much
interest and enhance the ccnfidenee with which water tunnel work should be regarded if some quantitative indication bf..
the "channel speed- correction" could be given in tie Author?
reply. It is assuiied that these corrections are purely empirical and are established from, 'and sertu to bring the tunnel results
into conformity ivith, corresponding tank 'experiments.
If
this is so, must tunnel results be looked on as valid only where', appropriate tank "cOntrols " are available?
It is noted that the obliquity of flow at the measuring section'
is' 10. Is the considerable angular velocity, which the impeller
'must impart to the stream e)iminn,ted by the 'nature of the
intervening channel; and is elan 10 not of some experimental importance?
As a physical phenomenon the steady periods between velocity
pulsations are somewhat unusual. What is the duration of the
steady period,, and in what way is it evidenced? Possibly a
design of 'tunnel which avoided abrupt changes of curvature--and conicality would elirñinate this hcurvature--andicap, curvature--and at the same:.
time reduce the.impeller power input. Could an illustration of the
propeller be added to this rascinating and informative record
Mr. L.- G. STEVENS, R.C.N.C:: Th Authors' account of 'early
-.530 THE LITHGOW PROPELLER WATER T1ThNEL
overcome some. àf. the diculties encountered 'will be read with interest and understanding by any who have been closely asoci-ated with cavitation tunnel testing. The improvement effected
in the tunnel performance by redesign of the impeller anil also
the projected use of the bronze impellers to investigate-the effect
-of blade roughness is noted witli interest. Can the Authors say -to what extent, if ny, the reduction in noise was due to
'the use of a wooden impeller? Was it modified when the bronze
-propeller was substituted? .
-Referring to the pulsations of water speed, it might be of
inteiest to note that no trouble of this nature has been
experi-.enced in the Haslar thnnel except at low cavitation numbers
-immediately before breakdown of flow the bend on the
down-stream side of the working section. The increase hi area of
the Haslar tunnel between the working section and the bend
is about 30 per cent. and the bend itself is rather easier, th,.is
onflnning -the remedy which the Authors indicate. Breakdown in
flow at the bend has been found to depend rather critically not only on cavitation number but also on water speed, temperature and the object under test in the tunnel. The minimum cavitation
number which 'has been ,-found to be obtainable at Haslar is
about 0l 1. With a propeller working it is higher to an pxtent
lependirig on slip. -,
The Authors refer to the effect of wake conditions-. in the ship
on cavitation and conclude that- in practic it -has been found
satisfactory to neglect the effect of wake variations over the
disc. It would be of interest if this could be confirmed. At Haslar it has been assumed piovisionally that the effect can be --taken account -of by comparing resulLs at a reduced cavitation number, but definite investigation is obviously most desirable Testing in de-aerat'ed water is also 'the practice - at Haslar,
the oxygen content, as determined by the Winkler test, being
kept below 05 c.c. per litre for most standard tests. To obtain
-this standard, which ha been found necessary for consistent
-results, the stream is reversed and the propeller worked at
-laige negative slip for a few hours prior to testing. In this waythe air is "flogged" 9ut of solution and drawn off at thefree
-sudae 1n the coaming. Information as . to experience in the
Lithgow tunnel in this mater woulc be of interest., A difficult -which has been experienced in testing the effect of air content
T LITRGOW PROPELLER WATER TUNNEL 531
'is the rapid de-aeratioti which occurs in the initial stages when
a propeller is tested in saturated 'water. In view of this the
apparent consistency of the spots in Fig. 5 is rather surprising and any further information would bô of interest. Was the air content megsiAred or were any precautions' taken to maintain saturation?
It wciuld be of interest if the, Authors could say
hy the
definition of cavitation number 'agreed at the- conference of
Tank Superintendents at Paris in 1935 has bee'n departed from: It is considered regxettable if this has been adopted as standard
practice, 'in. view of the desirability of ready comparison 1et*een
results from various establishments. It would, be appreciated 'if the Authors would define- VA more closely and if posible add to the diagrams a scale of cavitation numbers caiculated on the
usual basis. - -
-The paper not only gives an interesting account of the tupnèl itselfbut indicates some of the factors which require thvstigation -'to correlate the results with ship trials, including effects of 'air content, roughness, wake variation and a study of the
mechan-ism of cavity formation by.
high speed cine-photography. While it is hoped that in clue, course the Lithgow tunnel 'wiU be able to report good progress on these items, it is clear that 'there is a field for fundamental investigation sufficient to keep. the tunnel fully occupied for a long time to come.Authors' Reqiy
Messrs. 'ERsoN and BEnaY: Before replying to the points
raised in the discussion there are three questions not- adequately exptained iii 'the paper. First, it is necessary to enlarge the
brief statement in. the "channel - speed correction." This is
a standard' calculated correction given in the $per7 by Wood
and Harris and in a more readily available form in a. book8 by H. Glauert Jhe thrust of the model sérew can be expressed as the change in momentum at the screw disc.' The velocity'
'distribution in the tunnel is different from that in open 'water,'.
the channel speed correction gives 'the, equivalent open water speed, that is, the speed of advance: in open water which would
give the sathe velocity at the screwdisc. 'Tie '"correction can be 'Calculated for any screw diameter, speed and thrust.
..
532 TKE LIPHGOW PROPELLER WATER TU1EL'
For' the screw described in the paper 'the speed correction
in-creases from zero at no thrust, to a maximum of. 4 per cent
So long as the total correction is not greater than this, quite
large variations in radial distribution of. thrust' .wffl cause only
small differences in the axial velocity at any radius. For full
pressve running the calculated correction gives, agreement of
tunnel and open water thrust for screws of diameter greater
than 8 in., but the small chaliges in radial velocity distribution may affect the appearance qf cavitation. It is, of course, - the difficulty of separating the influence of the; tunnel walls whiek
makes difficult the testing of hulls or parts of hulls with propellers:
Tiere are two questions of nomenclature. The 1935
Con-ference of Tank Superintendents decided that screw results.
should be presented in' a base of v/nD but did not' give a symbol
for this ratio, so that the Teddington use of CD for v/D was.
not altered. The general use of 3' will save explanations arid.
deflnitions The Paris conference also decided. that the
'cavi-tation number should be called a where a=(p_e)/4pv2.
If the
speed used is the spd of advance, ahas no comparative value sincr the (speed of advance)2 is in general very small by corn-'parison with, the (rotational velocity)2. The velocity, VA usedin the paper is the velocity at which the O7 tipradius section.
meets the water (excluding inflow correction) and a defined
in terms of VA has a significant comparative value instead of being mere1y a base for plotting. Mr. Ferguson has completed the definition of VA; if v is the speed of advance and J =vinD'
then VA2 2+ (nO7D)2 and v2A/v2 =1 + (O77r/J)2 gives
therelation-ship between the values of a defined in terms of v arid VA.,.
The fluctuation in speed and the flow round, the top corner'
are discussed at some length in th paper because they are the
features 'of importance in ,any new design of return flowtunnel..
So far as th,e Authors are aware, all return flow 'tunnels are
imperfect. in design at this ..point and the reduction of speed
fluctuation from. the present 2"per cent. to less tjran per cent..
would not b obtained without, further experimental work. The; experimental method of waiting for the steady periods. does not- appear to entail serious disadvantages. It is found'
that the model screw thrust torqu and r.p.m. become constnt.
very quickly when 'the tunnel speed is constant. For recording
'FHE. LITEGOW PROPELLER WATER TD1NEL 533
-to learn from Mr. Stevens that the Haslar tuiinel is free from
-trouble at the corner presumably the higher value of the
minimum cavitation number at Haslar is due to other diffêr: /
nces in the tunnels.
-Coming now to inthvidul replies, in th propeller cutting
I apparatus the "knife" cuts iound each blade section and the
-cutting drawing consists of enlarged radial sections set off at
the designed pitch angles.
The paper is primarily a description of the Lithow tunnel
with an illustration of the results obtained. Mr. Allan and
Prof. IRobb appear to have been decpived by the gi'eat range
-of J and a shown in Fig-. 4- A little consideratkni of the lift
coefficients of circular back seàtions at redluCed pressure will -show that the value of J for zero thrust must alter as the se,tions
.cavitate. The nature oi cavitation on various types of se'ctjon
i again a matter of pressure distribution calculations or
ex-perimenti on the particular section.
In Fig. 3 at 0D='.°29
cavitation is visible below
O5 and back cavitation
below a=O-32'; at JD=O915 there is face cavitation below
a=045 and back caitation belowa=O32; at 0D=0778 the
tip vortex bubble spreads down the back of the blade below a=05 aid again the main back cavitation begins atapproxi-mately a=O32.
- The analogy between cavitation and stalling drawn by Prof.
IBobb seems to be more difficult to use than the more direct
-consideratiOn of the pressure distribution round a section with limited suction head. The separation effect depends prmariiy -on 1lie. velcity or pressure gradiemt
- In general ship practice a pressure drop due to acceleration
in the streamline flow equivalent to the atmospheric head of
30 ft. of water wpuld be most unusual. It is possible that' the
structure of the eddiesTormed after separation would be different
if the model tests were made with the correct scale pressure
-at. the free surface. But this does not seem to be sufficient
reason for the experiments suggeted by Prof. Robb.'
The informatiOn supplied by Mr. Stevens is of great interest. INo tecords were made of the noise levels with the wooden and
bronze impellers. If the information is sufficiently important
-comparative tests could be made. The reference in the paper -to the wake for fine-sterned twin-tcrew ship, was intended
534 TRE. -LITHGOW PBOELLER WATER TUNNEL
as a warning ,that the methOd of testing is not an exact repro-, dnction of ship conditions. It is considered that, as a guide to the appearance of cavitation it is safer to take the value of J corresponding to the. slowest moving water in the wake: The nine spots plotted fgr air-saturated water were .obtaiiied by taking readings as. the pressure was being reduced ; 'the
value plotted at a=O11 and the succeecting readings (nQt
plotted) ere moving across. to the 'de-aerated curve. When. fresh water is put into the tunnel it is necessary to circulate the water at. low pressure tO remove the air bubbles. Mter this pro.
cess no further de-aeratior. is carried out. it is found that' leaving
the tunnel' .standing' over night,: and emptying and opening,
the rneasuring section do not cause any discontinuity in results
The sections used for the impeller were approximately section.
P,gfven in R. & M. No. 1889, the thickness ratio varying from.
008 at 0625 ft. radius to 006 at 17 ft. radius..
There is sufficient material to allow cons1anSt velocitr sections
to be cut out of the impeller but it is not clear whether they
'wOuld be suitable' because of the turbulençe .of the tunnel flow.
it is realized that the 'imfeller affords cope for experiments
at a mre useful 'Reynolds' number than the ordinary mode,l
screw tests and any suggestions from Dr. (Jonn -would be wel-comed. The 0L for the tunnel impeller sectiOns is limited by
the appearance of cavitation, but it has been found thit, in
- wind tunnel' fns, a form of "stalling" takes place when the
tip section °L about O7. . -. .
Most of the points raised by Mr. Whiting have been covered. earlier in. this reply. The swirl produced by thern impeller is largely removedby. four fins immediately behind The correbt.
design of impeller should include twisted gnide vanes, in front
or behind, so that there is no' rotation of the slip stream. The obliqiiity of flow will require investigation if tets in sections
are nade in the' tunnel; for screw testing it is considered to b
of minor importance., . . -
..
The Authors would like to thank those taking part in the. discussion for drawing attention to' some of the omissions in.
the paper. Some of this vagueness is due to the siniultaneou
writing of a complementary paper9.'
it is hoped that any
obscurity not covered in this reply will be cleared by reference to the second paper.