King-KCA

13'h Congress of Intl. Maritime Assoc, of MediterraneanIMAM 2009, Istanbul, Turkey, l2-15 Oct. 2009

Fifty years of the Kings College Propeller Seriesat the Emerson Cavitation Tunnel

R. SAMPSON, G.H.G. MITCHELL & M, ATLAREmerson Cavitation Tunnel, University of lVewcastl,e, U K.

Since the founding of the Emerson Cavitation Tunnel in 1949, collaborative research into advancing

propeller design has been undertaken at the facility. The British Admiralfy initiated a systematic propeller

series known as the Kings College Admiralty (KCA) series that was developed and tested in the King'sCollege Tunnel (now the Emerson Cavitation Tunnel), between 1950 and 1955. The KCA series was unique

as it not only encompassed 30 propeller designs, but considered the effect of cavitation in the resultingdiagrams. The original design diagrams for the KCA series were constructed using 'drawing office' practice.

No spline fits or regression methods were available to fit the data and therefore the data was faired by hand

using French_curves. This requireci signifrcant levei of checking and re-testittg to.rerify the accuracy of,tliemethod. The intention of this paper is to re-visit the original experimental data taken.when the KCApropellers were tested in 1950 and re-assess the data using modern numerical methods.

1 INTRODIJCTION

The use of standard propeller series has wained inrecent years with the need for more detailed wakeoptimised designs. The ubiquitous Bp-6 diagramshave given way to computerised methods ofpolynomial fits and artificial neural networksolutions. Whilst the format of the design data maybe changing to take advantage of computing powerthe underlying data of the series remains robust.One of the more important and still widely usedseries is the Kings College Admiralty (KCA) seriesfor high speed craft. The series is unique as itprovides design data not only for propellerperformance but for thrust breakdown due tocavitation.

In an attempt to build a numerical method toassist designers in using this data, the originalexperimental results were re-analysed prior toimplementing complex numerical solution methods.This was deemed prudent as the curue fits used onthe data were all hand drawn. Therefore theobjective of this paper is to provide a platform fromwhich to build future numerical models. Therobustness of the early data will be underlaken toassess the curve fit methods and to select a data setto be used in fur1her analysis for the development of

multiple regression or ANN models. Following thisintroduction the history of the Emerson CavitationTunnel is described. This is followed by a review ofthe instrumentation used on the KCA propellers andthe development of the series is given. An analysisof the original model test data provides the mainfindings of the paper and the study concludes withrecommendations for future work.

2 TUNNEL HISTORY

The roots of the Emerson Cavitation Tunnel are

related to an original tunnel at Pelzerhaken inGermany, which was dismantled and brought toNewcastle after the Second World War. The originaltunnel was not a cavitation tunnel for testingpropellers, but was a horizontally disposed tubularcircuit. It is believed tht the tunnel was used forunderwater acoustic research such as the inceptionof cavitation on sound domes and the pioneeringdevelopment of anechroic tiles still used today onsubmarine hulls. A detailed history of the EmersonCavitation Tunnel is given in Atlar (2000).

The original tunnel was part of the Deparlment ofNaval Architecture at King's College, Newcastle.The tunnel had a measuring section of 32 inches by40 inches with radiused corners and was dedicatedto propeller testing. The shell of the tunnel was

74

erected as a vertical closed channel cavitation tirnnelby Vickers-Armstrong with the support of theprincipal propeller manufacturers of the UK. Thefunnel had a temporary housing in the steamlaboratory, a position it still maintains to this day.Figure 1 shows the tunnel schematic in 1949 and,Figure 2 the test section ln 1949.

1.22m width x 0.81m height in comparison ro the0.81m width x l.A2 m height of the old King,sCollege Tunnel cross section. The long propellershaft and dynamometer of the old tunnel wasreplaced py a new electroninc dynamometer unitwith a 90" drive from the top.

In 2007. again drrven by the dernarrds of indusrrythe ECT was modified to make the tunnel more laser.friendly. The measuring section was replaced forone with large windows and a new honeycomb andguide vanes, automated controi system and quickdegas unit were fitted. Finally the impeller, ,huftand impeller bearing were overhauied. Figure 3shows the tunnel schematic itt 2007 and Figure 4 thetest section in 2007 .

'-i:

::i,.€

:,i

j,rii.

Figure l. King's College Tunnel 1949-19g0

illhtf(. ftmrll trmmrn

Figure 3: Emerson Cavitation Tunnel (2007)

With the rapid increase in merchant ship size andin the power transmitted on a single shaft, the needsof the cavitation tunnel changed. A larger measuringsection was required to allow the placement of hullshaped bodies referred to as "Dummy Hulls" infront of the propeliers to help simulate tire extremelynon-uniform wake caused by fuller single screwhullforms. After 25 years of continuous use thetunnel was upgraded to modernise the ageingmachinery and equiprnent and replace tt " .rpp"iporlion of the funnei (contraction, diffuser andmeasuring section). At the official openingceremony in 1980 the tunnel was named. ,,ThiEmerson Cavitation Turutel" after Dr. ArnoldEmerson, the superintendent of the tunnel. Thesection had square corners, bigger windows andbetter access to provide a new cross section of

3 INSTRUMENTATION & EQUIPMENT

In the original cavitation tunnel, to drive thepropellers and measure the propeller loads an ACdrive controlled by means of a phase change gearand a rather novel Variable Speed Gear (VSG) dilvewas adopted. This measuring gear would allow amaximum thrust of 4903 N and a maximum torqueof 510 Nm at a maximurn rate of rotation of ZOb0rpm and was able to test moclei propellers up to 20',(508 rnm) in diameter. Unlike todays electronic

ii:!

(trrulir (o

Itctngultr

Figure 2. Measuring section - King,s Coil"g" T,*n"l

1,

72

: measurement was used to measure and control gas _ *_- \levels. As mercury is now considered ahazzardoussubstance; the Van Slyke has been superseeded bydigital DO2 system.

4 DEVELOPMENT OF THE KCA SERIES

One of the first tasks for the newly erected King'sCollege Tunnel was collaboration with the BritishAdmiralty on developing the KCA propeller series.The Director of Naval Construction, Department ofthe Admiralty placed contracts with the King'sCollege Tumel to investigate the propulsivecharacteristics of a methodical series of propellers.this series consisting of some 30 propellers to betested at 6 cavitation numbers and would generatethe well-known "KCA "or "Gawn-Burrill"systematic series and associated data. The designand range of models including test reqnirementswere to Admiralty requirements and the experimentsand results rested with the University. The extensivetest period lasted from 1950 - 1955 and consisted ofover 3000 propeller tests and some 1600 photo-graphs of cavitation pattems. Collectively the in-formation gathered was equal to that of 6 extensivepropeller series in a towing tank, one for each cavi-tation number of thc iest. The ts'CA propeller series

systems this unit was operated using a completelymechanical system. The propeller shaft was ftee tomove in the longitudinal direction and was held inposition using weights and a spring balanceconnected through a lever system to a thrust bearingsleve on the shaft. The thrust was also conected forDressure differences inside and outide the tunnel.The propeller torque was measured by weighing thereactions on the pinnions of the epicyclic drive to the

propeltrer shaft this measurernent rvas also correctedfor idle torque from the friction of the shaft bearing.In both the thrust and torque measurements no data

logging system was used, the operator recorded th-e

displayed analogue value which effectivelyrepresented the mean load. As the diai gauges wereoften moving this took a level of skill and

experience to obtain consistent results. This systemwas upgraded in 1980 to a Kempf and RemmersH33 dynamometer which has a lower load rating forthrust and torque and is therefore unable to test thelarge KCA propellers. The data logging has alsoadvanced with fulI digitai systems now comrnonlysampling all of the test data between 2-llkJtz. Theflow velocity in the test 'section was originallymeasured using a water manometer. This system isstill in use in the ECT however most tests are nowsupported by the Laser Doppler Anemometry (LDA)measurements. Finally a Van Slyke gas

was important and it was published for 3 main rea-SONS:

1. To help design propellers that do not erode inservice.

2. To allow the best possible propeller designunder thrust breakdown conditions.

3. To allow high speed propeller data to becomparyd against full scale and theoreticalconditions.

Using this above approactr propellers can be easilydesigned to absorb a design power at a design rpm,which do not erode in sbrvice and have high effi-ciencies.

The parent propeller of the series was KCA 110shown in Figure 5. The propeller was a 3 blade pro-peller with constant face pitch ratio of 1.0 blade arearatio of 0.8 and blade thickness ratio of 0.045. Theblade sections were segmental over the outer half ofthe blade at the inner radii the flat face was washedback at the leading and trailing edges, with the bladeoulline being ellipticai.

*. 30'

Figure 5: KCA 110 Expanded blade sections

From this basic propeller, 4 other propellers of thesame pitch ratio were derived but with BAR of 0.5,0.65, 0.95 and 1.1 respectively. For each of these

designs the new propeller was obtained by multiply-ing the expanded blade width with at each radius ofthe basic screw by a constant factor to give the re-quired area. For each blade arearatio, the expandedoutline was used with different face pitch ratios.The fuIl range of pitch ratio variations tested was0.6, 0.8, 1.0, 7.2, I.4, 1.6, and 2.0. The nomencla-fure adopted for the propeller names was as follows.The first digit in the numbers designates the BAR,digits 3, 4, 1, 5,2 refer to BAR of 0.5, 0.65, 0.80,0.95 and 1.1 with the last 2 digits relating to thepitch ratio. Thus KCA 420 has a BAR of 0.65 andpitch ratio of 2.0. Table I shows the section geome-try for KCA 110 and Tabie 2 sho.vs 3 |i5t nf nrnncl-lers in the series.

IJ

10

BAR r/R0.25

r/R0.37 5

r/R0.5

r/R0.625

r/R0.75

r/R0.875

r/R0.93'75

EXPA.R.

0.5 3.35 5.555 6.33 6.74 6.32 5.25 0.510.65 4.38 5.93 1.25 8.26 8.79 6.86 0.6650.8 5.3 8 7.30 8.92 0.18 0.80 0. l6 8.42 0.820.95 6.56 8.90 10.87 2.38 3.16 2.3'l 10.28 l.001.10 7.73 10.49 12.80 4.60 5.53 4.58 t2.t 1 1.18

0.604 0.8 l9 1.000 l. l,l 2t2 I .139 0.946

Table 1. Secl for KCA 110

able 2. Co te pro ler list of KCA Series

Propeller BAR P/D Sigma

306 0.5 0.6 8.3,2.0 5, 1.0,0.75, 0.5

308 0.5 0.8 6.3,2.0 5, 1.0, 0.75, 0.5

310 0.5 t.0 6.3. 2.0. .5, 1.0,0.75,0.5

312 0.5 1.2 6.3,2.0, .5, 1.0,0.75,0.5

320 0.5 2.0 6.3,2.0, .5, i.0,0.75,0.5

406 0.65 0.6 8.3,2.0, 5, 1.0, 0.75, 0.5

408 0.65 0.8 6.3,2.0, 5, 1.0, 0.75, 0.5

410 0.65 1.0 6.3,2.0, 5, 1.0,0.75,0.5

412 0.65 t.2 6.3,2.0 5. 1.0. 0.75. 0.5

4t4 0.65 1.4 6.3 2.0, 5, 1.0, 0.75, 0.5

4r6 0.65 1.6 6.3 2.0 5. 1.0.0.75.0.5

420 0.65 2.0 6.3,2.0, .5, 1.0,0.75,0.5

06 0.8 0.6 8.3,2.0, .5, 1.0,0.75,0.5

08 0.8 0.8 6.3.2.0. .5, 1.0,0.75,0.5

l0 0.8 1.0 6.3,2.0, .5, i.0, 0.75, 0.s

t2 0.8 1.2 6.3,2.0, 5, 1.0, 0.75.0.s

t4 0.8 t.4 6.3,2.0, 5, 1.0,0.75,0.5

lo 0.8 1.6 6.3,2.0, 5, 1.0,0.75,0.s

20 0.8 2.0 6.3 z.o 5, 1.0, 0.75, 0.5

508 0.95 0.8 6.3, 2.0, 1.5, 1.0, 0.75, 0.5

510 0.95 1.0 6.3,2.0, l.s,1.0, 0.7s, 0.5

51,2 0.95 L2 6.3, 2.0, 1.5, 1.0, 0.75, 0.s

5r4 0.95 t.4 6.3,2.0, 1.5, 1.0, 0.7s, 0.5

516 0.95 L6 6.3,2.0, 1.5, 1.0, 0.75, 0.5

520 0.95 2.0 6.3,2.0, 1.5,1.0, 0.75, 0.5

208 0.8 6.3, 2.0, 1.5, 1.0, 0.75, 0.5

210 1.0 6.3,2.0, 1.5,1.0, 0.75, 0.5

212 1,2 6.3,2.0, 1.5,1.0, 0.75, 0.5

214 t.4 6.3, 2.0, t.5, 1.0, 0.75, 0.5

216 L6 6.3, 2.0, 1.5, 1.0, 0.75, 0.5

5 THE KCA SERIES MODEL TEST

The uniqueness of the KCA propeller series liesin the fact that each propeller was tested at 6cavitation numbers. For the experiments the freestream cavitation number was used (ou : 6.3, 2.0,1.5, 1.0, 0.75 and 0.5). The tunnel velocity andcavitation number were held constant whilst thrust,torque and shaft revolution measurements weremade over a range of advance coefficients. In Gawn& Burrill (1955) the results of the KCA propellersare presented in non-dimensional format, for thrust(K1), torque (1OKq), and efficiency (n") to a base ofadr.ance coefficient for each cavitation number asshown in Equations I to 4. The value of V1 was

corrected for tunnel wall interference usinsknown Wood & Harris /11920) method.

the well

Cavitation number

Thrust coefficient

Torque coefficient

Efficiency

As mentioned previously data aquisition was almostnon existant at the time of these experiments,Where the present day open water test can be fullyautomated and executed with aneat report as soon asthe test is finished, in 1950 the reality was quitedifferent. From the original test data it is possible toobserve that each group oftest spots typically took 2hours, as rpm was increased, manometer stabilisedand , the weight pans adjusted to calculate themeasurement. The output therefore relied on thecare and patience of the experimenter and was anaturally attenuated analogue result giving a singlemean value. To this end, no time domain recordswere at all possible and simple repeat testing was theonly cross check. However, also lacking at this timewas the ability to manipulate the tabulated data thatwas gathered. The raw experimental data at non-uniform advance coefficients was plotted in largeformat by hand and the results graphicallyinterpolated to obtain standard J values needed forgeneration of diagrams. Without the assistance ofcomputers the curve fitting which is taken forgranted today had to be performed by hand usingfrench curves. No numerical curve fitting methodwas employed and the fairness of the curves reliedon the skill of the drawing office. The faired datawas subsequently generated into Bp 6 diagrams andthe results cross plotted and checked until suitableconvergence was achieved, often by generatingcurve drawings of enormous size.

The core motivation of this paper is a preliminarysfudy of the original unprocessed test data inconjunction with the hand faired results; the latterbeing published widely. As the use of Bp 6 diagramsnow primarily rests as a useful teaching tool for theNaval Architecture student or for guidange of thekeen boat owner, a more preferential format isnumerical representations suitable for incorporationinto performance prediction software. The data is

p-eO.. = -------------" y, p(v,)'

T'l\r : ------:-;' pn'D"

K^:-9.o pr'D'

K,Jn^=rX'-Ko 2n

(1)

(2)

(3)

(4)

-74

commonly given as polynomial coefficients frommultiple linear regressicn such as Van Lamrneran et

al. 0969) and Yosifof et al. (1986) or more recentlyas tables of weights for the use in Artificial NeuralNetworks (A1'IN) such as Koushan (2007). Either ofthese formats can then be incorporated into NavalArchitecture soffware allowing rapid development of

_ basis propeller designs.Prior to selecting a suitable method for the

analysrs the basis data needeC tc be assessed.

Mitchell (2006) noted that the method and

compilation of the original KCA data may providedifferent- results when modern analysis procedures

were applied to the data. In the original KCApropeller reports by Emerson and Burrill (1955), 2tables of experimental data are given for each test

that is raw and fitted. This paper presents the firststage in the project and compares the fitted curvedata from 1955 with the raw experimental data fittedand faired with a 6th order polyniomial regressioncurve. In this manor the goodness of fit can beassessed. The polynomiai method does havelimitations and it is not ideally suited to the curyefitting for the entire KCA series'due to the thrustbreakdown effect. Fieure 6 shows that whencavitation develops ott ttt" blade sections theperformance is modified and rapidly deviates fromthe standard Kr curvo.

J'Figure 6. Sample of KCA 514

The discontinuify increases the order of the curvessigninfcantly until quite often the Runge (1901)phenomenon occurs at the extremities of the data,that is erratic curue fitting when using polynomialinterpolation with polynomials of high order. Aprcferabie solution is to use a cir'oic s;rlirie iriteipola-tion however the fitting of each curve of every pro-

peller test was beyond the scope of this paper.Therefore this introductcry study was limited to theatmospheric conditions (ou : 6.3), which repre-sented performance curyes, which could easily havebeen obtained from towing tank or cavitation tunnel.On these curves no thrust breakdown occurred.Lines joining the test spots represented constantstatic head in the tunnel but not constant cavitationnumber due to tunnel wall interference and increasein static pressilre',-".ith thrust. The constant ca.,'itationnumber lines needed for the Bp-6 diagrams were ob-tained by plotting the values at the same advancecoefficient (J) on a basis of corected cavitationnumber and lifting off the values at each particularconstant cavitation number.

7.2KT EXp

r KT Fit

10KQ Exp

^ 10KQ Fit

Eta_o Exp

Eta o Fii

G-

0.75 0.95 1.15 1.35 1.55 7.75

Advance Coefficient (J')

Figure 7. KCA 1 16 Open water diagram

Figure 7 shows the open water plot for KCA 116 at

atmospheric conditions. in the plot the experimentaltest spots at non-uniform J values are shown with theopen shapes. The fitted test spots i.e. those fairedand taken at uniform J spacing are shown with filledshapes. From Figure 7, it is possible to see that thereare slight differences in the perforrnance curves, par-ticularly the torque cllrve. These differences arerelatively small, however they have a far greater ef-fect on the efficiency curve, which in itself was stillrelatively small. The lines shown in the figure werecreated as.6tr' orcier poir,norrrials through tire originalexperimental data. A1l of the curves show good

I

uv""o

0.4

75

agreement and the trend suggests a slight biasing ofthe data from the hand fit method however giventhat this was made on the performance curve and notthe efficiency curve this gives a good level of confi-dence in the results. This also suggests that theoriginal KCA propeller data is robust and the analy-sis methods adopted at the time were appropriate.However Figure 7 represents a single case thereforethis study was extended to the entire series. Figures8 to 12 show the open water efficiency curves forthe entire KCA propeller series taken at the atmos-pheric condition (o, : 6.3). In the figures the origi-nal experimental data with a polynomial fit is givenwith the solid line and the hand faired standard J ex-perimental results, which have been used in publica-tion are given with the broken lines.

o.8

o.7

o o5

- O..t

0.3

0.2

0.5

Figure 8. Effrciency curves for BAR: 0.50

o.9

o.a

O./

0.6

- 0.i

0.)

0.I

0

0.5 1 1.5

Advance Coefficrent (-J)

Figure 10. Efficiency iu;;;fb;BAR: O.s0

0.9

0.3

0.2

0.1

0

i Advance Coefficieni (l)

Figure 1 1. Efficiency curves for BAR : 0.95

0.4

o.7

0-6

I

- 0.4

0.9

0.4

4.7

0.6

o oS

s- oc

BAR = 1.1

0.9

O,B

0.1

8.6

5

fl. J

BAR = O.65

4.2

0.1

0

0.l

4.2

(1. l

0

0 0.5 1 15Advance Coeffi.ient ( l)

Figure 9. Efficiency curves for BAR: 0.65

0 0.5 1 t.5Advance Coeffrcrent (J)

2 5 Figure 12. Efficiency curves for BAR : 1 . 10

As the performance curves were difficult to as-sess only the efficiency curves are given in the fig-ures. In general the two methods of curve fit showgood agreement with the most variation occurring inthe high J areas of the test commonly associatedwith lower Reynolds numbers. As the hand fairingmethod relied to a great extent on the skill of the op-

8AR = O.BO

rt2116ja<---\ 120

lite- ffi: \{\,\\\

toe/

l

\

76

erator and some bias may be present. Frorn the plotsthere is no real trend or bias difference between the2 methods. However the published curves (handfaired) are more often underneath the actual experi-mental values making the published results for thiscondition conservative and giving a beneficial mar-gin to the work.

Finally one of the advantages of propeller testingin icavitation tunnel is highlighted in Figures l3 -16 u,hich sho-,r,s a sample of the phctographs takenduring the propeller testing, something not possiblein the towing tank. The photography method wasdeveloped by Townsin (1955) and was able to cap-fure a blade at a repeatable angular location. Whenthe cavitation extent recorded in the photographs iscorrelated with the numerical results the power ofthis standard series becomes clear. In the figuresshown the atmospheric condition shown for KCA420, despite a developed tip vortex and foamingsheet cavitation no thrust breakdown was recordedfor the conditions shown.

00, o,: 6.3

6. CONCLUSIONS

The original and unpublished experimental datafrom the KCA propellel series measured at the at-mospheric condition has been re-analysed and com-pared with the faired data. From the study the fol-lowing r,vas concluded:

. The hand fairing methods used on the originaldata gave a good fit based on comparison withthe orisinal un-faired data.The original un-faired data responded well to 6'n

order polynomial regression techniques.For lorver cavitation numbers the polynomial re-gression technique was not suitable.The faired data gave a rrore conservative esti-mate of the efficiency when compared to the ex-perimental data.

Figure 14. KCA420. J: I

Figure 15. KCA420, J - i.30, o" : 6.3

Figure 12. KCA420, J: I

Figule

77

7. FUTI.IRE WORK

The findings presented in this paper indicate that theKCA propeller series would benefit from beingtransposed into numerical methods such as ANN. Tocomplete such a task the thrust and torque values forthe entire propeller series are required at all cavita-tion numbers, based on either the original experi-mental and faired data. Once this data is obtained aglobal assessment on the goodness of the data for allcavitating conditions can be made and a data setadopted into the future model. It is hoped themethod can be developed which will be of use to thenovice Naval Architect student as well as the sea-soned Professional. It is also desirable to extendthe scope of the series by introducing measurementsin the 4'" quadtant i.e. advance speed astern and pro-peller rotation ahead. This would provide the mostcomplete data reference for the design possible.

8. REFERENCES

Atlar, M. 2000. A History of the Emers;on CavitationTunnel and its role in cavitation researclt.NCT'50 Conference, Newcastle upon Tyne

Emerson, A. and Burrill, L.C. 1955. A series ofModel propeller experiments. Part 1-7, King'sCollege Report, UniversityNewcastle upon Tyne

of Durham,

Gawn, R.W.L. and Burrill, L.C. 1955. The effect ofcavitation on a series of I6in. Model propellers.Transactrons Royal Society of Naval Architects.

Koushan, K. 2007. Mathematical expressions ofthrust and torque of Gawn-Burrill Propeller se-ries for high speed crafts using artificial neuralnetworks. Ninth International symposium on fastsea transportation, FAST2007, Shanghai, China.

MitcheTl, G.H.G, 2006. Personal communication,University of Newcastle.

Runge, C. 1901, Uber empirische Funktionen und.

die Interpolation zwischen ciquidistanten Ordi-naten, Zeitschffifi.ir Mathematik und Physik 46.

Townsin, R.L. 1956. Photographing Cavitation onModel Propellers, The Journal of Photographic

. Science, Vol. 4Van Lammeren, W.P.A. and Van Manen, J.D. and

Oosterveld, M.W.C 1969. The Wageningen B-Screw Series. Society of Naval Architects andMarine Rngineers - Transactions, Vol. 77

Wood, R. & Harris, R. 1920. Some notes on thetheory of an airscrew working in a wind channel.British Aeronautical Research Committee,Research and Memoranda No. 662

Yosifof, K., Zlatev,2., and Staneva, A. 1986. Opti-mum Charcicteristic Equations for the KT-J De-sign Charts based on the Wageningen B-ScrewSeries. International Shipbuilding Progress, No.382, Vol. 33 (1986)

7B