You are on page 1of 22
Subject 1—Reynolds Number for Model Propeller Experiments CHAIRMAN, PROFESSOR L. TROOST, SUPERI REPORT OF INTERNATIONAL Co! INTENDENT, NETHERLANDS MODEL BASIN MMITTEE AND INTRODUCTORY REMARKS BY DR. JAMES F. ALLAN Decisions at the 1948 London Conference 1. The Conference is not in a position to rec- ommend any immediate revision of the Berlin figures of minimum Reynolds number, although these are known to be low for many types of pro- pellers when run in open water without turbulence stimulation, 2. It is necessary that the model propellers should be made to a high degree of precision, and in all published work the measured toler ances and the quality of the surface finish should be stated. 3. It is desirable that the experiments on circular-back blade sections as described in Reports and Memoranda 2301 should be ex- tended to cover other shapes. 4. The value of Reynolds number for propel- lers should be calculated using the blade width and the resultant velocity at 0.7 radius, ignoring inflow. PROPELLER COMMITTEE: Dr. J. F. Allan, Chairman Ms. R. B. Couch Dr. H. W. E. Lerbs Prof. E. V. Telfer Terms of Reference: To collect, scrutinize and review existing propeller scale effect data, and to recommend further relevant research. The Committee has held no formal meetings but has carried on its work by correspondence, with occasional personal contacts as the oppor tunity offered, Following up the terms of refer- ence a survey of published data in various lan- guages was made and the following list was prepared: 1, "Neue Propellerversuche,”” by Fr. Gebers, STG, 1910. 20 2. “Propeller Design Based on Model Experi- ments,” by D. W. Taylor, SNAME, Vol. 31, 1923, Pp. 57-106. 3. “The Separation of Dissolved Air Caused by Propeller Action,” by Ing. Col. G. Rebbeno, INA, Vol. 71, 1929, p. 331. 4. “The Effect of Immersion on Propellers,” by Mrs. Smith-Keary, NECIES, 1931. on’ Geometrically-Similar Ship ” by H. E. Saunders, SNAME, Vol. 40, 1932, p. 75. 6.’ “Immersion of Propellers,” by G. Kempf, NECIES, 1933/4, p. 225. 7. "The Influence of Viscosity on Thrust and Torque of a Propeller Working Near the Surface,” by G. Kempf, INA, 1934, p. 321. 8. “Further Model Tests on Immersion of Pro- pellers, Effect of Wake and Viscosity,” by G. Kempf, NECIES, 1937/8, p. 349. 9. 1938 “Rauhigskeits-und Kennzableinfluss bei Schiffsschrauben,”” by G. Kempf, WRH, Vol. 19, pp. 145-148, 1939 “Influence of Blade Roughness and Scale Effect on Propeller Efficiency,” by G. Kempf, SBMEB, Vol. 46, pp. 279-280. 1939’ “Ergebnisse Naturgrosser Schrauben- versuche auf Dampfer Tannenberg, by G. Kempf, WRH, Vol. 20, pp. 167-174. 10. ‘'Versuche Uber die Profileigenschafeen der Bluttschritten von Schiffsschrauben und ihr Ein- fluss auf deren Entwurf und Auswercung,” by F. Gutsche, Mitteilungen der Preussischen Versuch- sanstalt fir Wasserbau und Schiffoau, Berlin, Heft 10, 1933. “‘Kennwerteinflisse bei Schiffsschrauben- Modellversuchen’”? by F. Gutsche, SJG, 1936, p. 277, and WRH, 1936, p. 4. “Versuche mit umlaufenden Trigfluegeln,” by F. Gutsche, WRH, Vol. 21, No. 2, 1940, pp. 1415. REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS a 11, “Results of Experiments on Model Screw Propellers with Wide Blades,” by R. W. L. Gawn, INA, Vol. 79, 1937, p. 183. Discussion. See also note by Dr. G. Kempf, p. 178, on calculation of scale effect by differences of curves for Cy and Cor 12. “Scale Effect in Screw Propellers,” by J. F, Allan, INA, 1934, p. 111, 13, ‘The Effect of Inclination, Immersion, and Seale, on Propellers in Open Water,” by R. de Santis, INA, 1934, p. 380. “Sull’ effetco scala nelle eliche geometrica mente simili,” by R. de Santis, Annali Vasca Naz. Esper. Arch. Nav., Roma, 10, 1941, pp. 51-71. “La cortezione d’attrito sulla pala dell’ ellica ed il principie di similitudine meccanica,”* by A. di Bella, 1942. Annali Vasca Naz. Esper. Arch. Nav., Roma, 11, pp. 151-163. 14. “Dimensional Analysis of Model Propel- ler Tests,” by E. Buckingham, ASNE, Vol. 48, 1936, pp. 147-198. 15. "Propulsion Seale Effect, Lammeren, NECIES, 1939/40. 16. ‘The Efficiency of Marine Sctew Propel- ers and the Drag Coefficient,” by G. S. Baker, NECIES, Vol. 61, 1944/5, p. 279. 17. “Experiments in the Lithgow Propeller Tunnel,” by A. Emerson and L. W. Berry, NECIES, Vol. 63, 1946/7, p. 333. 18. “‘Seale Effect on Model Propellers,’ by J. G. Hill, TMB Report 660, 1948. 19. “Cavitation of Screw Propellers,” by R. W. L. Gawn, NECIES, Vol. 65, April, 1949, p.- 339; see also p. 349. 20. “Tests on Four CircularBack Aerofoils in the Compressed Air Tunnel,” by D. H. Williams, A. F. Brown and C. J. W. Miles, ARC R and M 2301, p. 3. 21, "Principles of Naval Architecture,” Ros- sell and Chapman, SNAME, Vol. II, pp. 138-139. 22, “Resistance, Propulsion and Steering of Ships,” by W. P. A. van Lammeren, L. Troost and J. Kéning, 1948, pp. 139-142, 164-167. 23. International Conference Reports; Paris, 1935; Berlin, 1937; London, 1948, 24, “A Note on Propeller Scale Effect,” by E. V. Telfer, 1951. by W. P. A.van These publications were translated into Eng- lish where necessary, and all of them summarized 30 that a survey of che information could be made. The information contained in these references aturally covers a great many points which, al- ready have been discussed at these conferences and it does not appear that any important data have been overlooked in the past. The next part of this statement makes brief references to sali- ent points in the publications. Items 1 and 2, by Gebers and Taylor respec- tively, concern a series of tests on propellers of ogival blade section, and both papers indicate no serious departure from the square law as regards thrust and torque, i.e. no scale effect. Item 3, by Rabbeno, discusses the effect of dissolved air content and also variation of water temperature on propeller action. Four stages are distinguished, depending on the blade pressure conditions. ‘These range from no effect at low pressures to a serious loss in efficiency due to production of bubbles at high pressures. Immersion effect, which influences the non- dissolved air content, is dealt with in item 4 and again in icems 6, 7, 8, and 13. These reports refer to experiments both “tin the open’ and be- hind a model. Ieem 4, by Smith-Keary, states that good agreement was obtained between ship and model as regards rpm and thrust. Item 6, by Kempf, states that in such cases the Froude Law of Similarity can be applied only if the pressures fare similar for model and ship. There is agree- ment as regards the general drop in thrust, torque, and also efficiency due to propeller tips ap- proaching and breaking surface. The importance of the interaction between propeller and hull in such conditions is emphasized in item 6. Ikem 7, by Kempf, makes special reference to high-slip conditions and mentions the partial re- covery that takes place in thrust and perform- ance after air saturation has occurred. A mini- mum Rp of 0.5 x 10 at 0.7 radius is suggested for such tests. Item 8 refers to Kempf's discussion of the ques- tion of frictional drag of propeller blades and proposes the assumption that at zero thrust the torque is caused mainly by frictional drag. Ie is suggested that ideal smoothness is achieved on the model propeller at an Ry of 0.6 x 10", Allow- ance for roughness on the full-scale propeller is proposed and an example is given. Item 9 develops this proposal and gives a specific method for applying the correction for seale effect and roughness. The results of che full-scale trials on the Tannenberg are cited, and one conclusion from these is that'the rough ness allowance on the ship propeller just off- sets the scale effect on the propeller friction coefficient. Trem 10 refers to Gutsche’s work on individual foils in “under” and “over” critical flow condi- tions, to cascade effects, and to the application of this to the complete propeller. Limiting Rp 2 REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS, values for the change of condition are given for two types of section. 5 Item 10 also covers Gutsche’s tests on a spe- cial 3-bladed propeller in which pressure on the blades was measured. These tests show the ¢! fect of centrifugal force on the boundary layer in a rotating propeller. Item 11, dealing with Gawn’s series tests on wide-bladed propellers, gives no specific infor. mation on scale effect, although the effect of blade roughness was referred to in the discussion. Item 12 refers to Allan’s tests on scale effect in blade elements of different types and to the ef fect of this on model propeller performance, The “under” and “over” critical condition of flow is noted. A proposal is made to utilize a spe- cial water tunnel under pressure and at high temperature in order to reach high R,values with models of reasonable size. Some ship-to-model performance comparisons are given. Item 13 refers to de Santis’ tests on inclina~ tion, immersion, and scale in open-water tests. His conclusions are: (2) The smaller the propeller, the smaller the thrust and torque and the greater the efficiency. (2) Increasing immersion gives increasing thrust, torque and efficiency but when the im- mersion/diameter ratio exceeds 0.8 there is no more effect. (3) Inclination up to 6 degrees has no material effect on thrust, torque, or efficiency. This item also refers to some tests made by de Santis on a model propeller with smooth and rough surface. The loss in effective pitch due to rough~ ness is shown, and scale effects in thrust and torque are shown. It is concluded that roughness does not guarantee constancy of type of flow. Di Bella’s theoretical work is covered also by this item. He assesses the change in thrust and torque due to friction changes, using Blasius and von Karman formulas. He considers the theoreti= cal results are in line with experience. Item 14 refers to a paper by Buckingham giving a general review of the position, A propeller diameter of 16 in. is proposed as being large enough to avoid the critical range. Trem 15 refers to van Lammeren’s scale-effect series. Extensive reference is made to Gutsche’s work, It is stated that as laminar, turbulent, or transitional flow may exist on a propeller blade, care should be taken in drawing conclusions from open-water propeller tests with ehe same propel- ler at different rpm, Propeller efficiency may in- crease or decrease depending on the R,, and the type of flow. The efficiency of a smooth propel- ler is likely to increase indefinitely with increase in Rye Item 16 refers to Baker's paper in which he puts forward a simple formula for propeller ef s(-s) Gros) ‘the @ constant depends directly on mean drag co- efficient. Scale effect tests on blade sections in the Compressed Air Tunnel (CAT) are referred to (see also item 20) and a method is developed for correcting the drag coefficient of well-designed propellers. Item 17 gives some scale-effect results by Emerson and Berry for 8-in. and 6,4in, diameter propellers tested in the NPL tunnel. The range of Rq was such as to cover a large amount of “under-critical”” flow and a large scale effect in thrust is noted. Item 18, a paper by Hill, refers to propeller scale-effece cests at the David Taylor Model Basin, Reference is made to the effect of defor- mation on model propellers under load. The use of curbulence stimulation on model propellers is suggested. Item 19 refers to work by Gawn on the cavita- tion of screw propellers; the “scale” effects re~ ferred to are concerned with cavitation. Item 20 refers to experiments inthe CAT at the NPL by Williams and others on four ogival- sectioned acrofoils. A maximum R,, of 6 x 10* is covered and it is suggested that little change in the life curve is to be expected beyond this point, except in the region of the stall. The effect of surface roughness is mentioned. Trem 21 refers to the general review of the subject given in “Principles of Naval Archi- tecture”? by Rossell and Chapman, and item 22 to a similar survey in “Resistance, Propulsion and Steering of Ships” by van Lammeren, Troost, and Kéning. Item 23 covers the reports of the previous con- ferences in 1935, 1937,and 1948. The reader will be familiar with that for 1948 which covers the previous discussion: Item 24 refers to a proposal put forward by Telfer that the extrapolation diagram given in his 1949 INA paper is very conveniently applicable to the propeller problem. The details of this method are being circulated separately. ‘Summing up the position, it may be said that the causes of “scale effect” in the performance of screw propellers are: (The ‘change fon laminar to turbulent boundary layer flow on the blades. (2) The continued decrease in turbulent fric~ tion with increase of Ry. In the above statement ficiency, namely: 7= and shows that REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS B scale effect” is defined as ‘variation in the value of Ky and Ko with change in R, of geo- metrically-similar propellers at constant J or slip value.”” The major effect of the change from laminar to turbulent flow on the blades is a movement of the separation point on the suction side from forward towards the trailing edge. This is as- sociated with a large increase in pressure lift and a large reduction in eddy drag. These changes ate most marked in thick, ogival sec~ tions, and scarcely noticeable in thin hydrofoil sections, They take place in the R,, range 0.1 to 0.310%. The continued decrease in turbulent friction with increase in R, causes a continued fall in Ko and a very increase in Ky for ideally smooth propellers, and there is no limit to this effect in terms of increasing Raw In addition to these influences there are im- portant effects arising from immersion, air con- tent of water and blade roughness, but these must be allowed for independently. The methods which have been proposed to cor rect for propeller scale effect apply generally to the blade friction only and assume no change in the lift circulation or pressure distribution round the blades. These methods by Kempf, van Lam- meren, and Baker, are described in items 9, 15, and 16 respectively, and a new proposal by Tel- fer in item 24. In general, they make provision for roughness allowance on the full-size propeller. None of these methods has been widely adopted, probably because of doubts regarding the ac- curacy and extent of our knowledge of the various factors involved, and also because of associated scale effects in the hull factors. It has generally been preferred to adopt an over-all factor between model and ship to cover various unknowns. Some experiments have been made with turbu- lence-stimulating grids ahead of a model propel- ler, item 15, but the results are not sufficiently comprehensive or conclusive. There is also a body of opinion which considers that the stimula~ tion must take place on the blade itself, othe wise laminar flow may be established on the leading part of the blade even in a stream con taining an appreciable amount of turbulence. ‘A further complication arises in the difference in turbulence between the open-water condition and the behind condition, so that whereas the flow may be largely laminar in the former it is probably turbulent in the latter. ‘At this point mention may be made of efforts propeller blades. if this can be achiéved successfully on the model, it is still a major problem to reproduce similar flow conditions on scale. It is, of course, appreciated that jon from laminar to turbulent flow on pro- peller blades may cause either a gain or a loss in performance depending on the shape and thick- ness ratio of the sections, It appears that our present knowledge of flow conditions on propel- lers is insufficient to determine the extent and effect of laminar flow on the model propeller performance. More extensive work on turbulence stimulation on propellers should be done. This should in- clude stimulation on the blade as well as in the stream ahead of the propeller. A method should be developed also for detecting the extent of lam- inar flow on propeller blades ot determining the position of transition from laminar to turbulent flow. When the flow conditions on the model pro- peller are known and controlled it will be possi~ ble co apply a blade-friction correction method with confidence. Ie will be essential to have a more extensive knowledge of propeller-blade roughness so that correct allowance can be made for the effect of this on blade friction. Measurements of blade roughness therefore should be made on repre- sentative groups of full-scale propellers. It would be useful ¢0 carry out similar tests on blade ele- ments of various types and thicknesses, to en- able this approach to be developed. To provide a check on “‘scale-effect”? correc: tions based on the foregoing, it is recommended that several “scale-effect” series of propellers should be tested. High-power dynamometers will be required for this work and it should include the investigation of roughness effects on, say, 2-ft diameter propellers. The several series should cover variation of pitch ratio, area ratio, thickness ratio and number of blades, but should be confined to well-designed propellers which avoid any eddy-making. Work on any one series should be confined to one establishment. There are difficulties in carrying out such a program internationally, but to achieve the results in a reasonable time, it will be necessary to spread the work over several establishments. The members of the Conference should com- ment on, ctiticize and amend the above sugges- tions, so that a useful and practicable program of further research may be determined at the Confer Attention is drawn to the recent publication by the ASNE of a paper by Dr. Lerbs “On the Effects of Scale and Roughness on Free-Running Propellers.” 24 REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS Formal Discussion ComMENTS BY DR. J. F. ALLAN This question forms an important part of the more general question of scale effect on propul- sion factors, discussed as a separate item in this Conference. It was posed originally as ‘The Determination of Minimum Reynolds Number for Propeller Model Experiments”? but is resolving it~ self into a determination of an acceptable method of correcting for propeller scale effect and blade roughness. ‘There is no generally accepted method of mak- ing these corrections, and it is highly desirable that in due course a uniform decision should be reached in this matter. To that end, it appears to me that a critical study should be made, by the Committee, of existing and proposed methods of making such corrections, including: (1) Dr. Telfer’s proposal, which is before the Conference (2) Dr. Lerbs’ proposal as published in a re- cent paper (3) An amplification of the late Dr. Baker's proposal which will be available shortly through the BSRA, There is a lack of large-scale propeller geo- sim data against which to check the method; fa- cilities for obtaining such large geosim data are limited. There shortly will be available a large~ scale propeller dynamometer at the Taylor Mode! Basin. We have made considerable progress to- ward designing and manufacturing a large-scale propeller dynamometer for open testing. I think John Brown and Company have a fairly powerful inserument and it may be that the Rome tank also has such a dynamometer available. I recommend that some attempt be made to coordinate work with such large-scale propeller dynamometers as are available, so that we may avoid overlapping of effort and cover as wide a range of type of propeller and variation of characteristics of pro- peller as may be achieved. ‘We cannot rigidly coordinate such work inter nationally, but a good deal has been achieved in that direction on the cavitation propeller series, and a similar success might be achieved with open propeller geosim testing. Any tank propos ing to test a series of similar propellers is re- quested to advise the Chairman of the Interna tional Committee of the general particulars of the series. ‘As a start, I make the suggestion that the Series I and Series I propellers, varying from 8 in, to 16 in, diameter, which already have been accurately manufactured and tested in the cavita- tion tunnels, could be tested very usefully in the open over as wide a range of conditions as possi- ble to start this accumulation of accurate data on propeller geosims for the purpose of studying scale effects. Ie also would be of great value to carry out roughness tests, especially on the larger-scale propellers of these and other geosims which may be tested by any of the basins concerned with this Conference. The other major problem is the question of the extent of laminar flow on model propellers and the control of laminar and turbulent flow on pro- peller models. We have here a rather difficult problem and one which has not been investigated 80 far to any appreciable extent. Methods of de- tecting changeover from laminar to turbulent flow in the propeller are rather difficult to devise, and suggestions in this direction would be welcome. I feel that the title of this subject could be changed usefully from the determination of mini- mum Reynolds number to the determination of an acceptable method of correcting for propeller scale effects and roughness effects, with appro- priate terms of reference. Consideration should be given also to the possibility of merging this Committee into, say, the Cavitation Committee. One last point concerns the personnel of the Committee, which should be modified to make it more representative. I am not suggesting it should be enlarged, but due to the shifting of per- sonnel since the existing Committee was formed there is a certain unbalance of distribution of membership. COMMENTS BY DR. J. M. ROBERTSON Decision 2 of the London Conference stresses the need for model propellers made to a high de~ gree of precision. At the Ordnance Research Laboratory of the Pennsylvania State College, we have a propeller research program, the experi- mental phases of which require high-precision propellers. Five- to six-in. diameter propellers are needed with desired section tolerances of 0.001 in., but acceptable tolerances of 0.003 in. This contribution reports briefly on the results of the two-year program concerning propeller manufacture. A survey of the known possible methods of manufacture indicated that only a process which terminated with laborious hand-finishing by skilled craftsmen, using precision templates, could be counted on to achieve the minimum tol- erance of 0.001 in. We have been able to obtain REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 25 superior cast-aluminum propellers which require aly a minor amount of hand work and which agree with specifications to within £0,005 in. Invest ment casting methods do not promise sufficient accuracy to warrant the high die cost. Profile nilling of a complete propeller is difficult be- cause of the poor cutting angles and interference between blades. Precision methods of manufacturing blades have been developed for axial-flow turbines and compressors. By going to a blade-by-blade manu- facture, we have been able to take advantage of these methods. There are additional reasons for making separate blades: (1) The blades have a high degree of uniformity (2) Extra blades can be obtained cheaply to al low replacement of damaged ones (3) Overlapping blades may be readily manu- factured (4) Propellers may be made with large num- bers of blades (5) There is more freedom in the selection of This system requires a good method for attach- ing the blades to the hub. ‘The accompanying Figs. 1 and 2 show an B-bladed stainless steel propeller which we have recently obtained. The blades were manufactured by a process developed by the NACA for airfoils. ‘As may be seen in Fig. 2, the hub attachment is relatively simple. Since centrifugal forces are small, a single screw in conjunction with the T-slot serves to hold the blade in place. The blades have been checked by the ‘*Viz-O- Graf? as well as by our own measuring machine and found to agree with specifications co within £0.003 in. The methods we use to check our propellers require a litele comment, Just what you are measuring is very important when you say a propeller is good to 0.001 or 0.003 or 0.005 in. In our process we try to check the location of a point on the blade surface as a point in space and we want to locate it within 0.001 to 0.003 in, So, at five or eight radial stations on the blade, we measure the location of points on the surface and compare the locations with what we have specified. In the propeller shown, by check- ing abour five stations and ten to twenty points on the surface at each station, we find an aver- age accuracy of 40.003 in, Actually, over 90% of the blade area is within 0.001 to 0.002 in. As this is only the first propeller we have buile this way, we feel that it gives promise of increased accuracy as we progress in making more and more propellers. FIG, 1.—Eight-Bladed ORL Model Pr Separate Stainless-Steel Blades er witht FIG. 2.—Hub and Blade Construction of ORL Propeller, 26 REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS ANOTE ON PROPELLER SCALE EFFECT by Professor E. V. Telfer 1951 (1) The purpose of this note is to show that the Froude extrapolation diagram, first given in my 1949 paper and further developed in my 1951 paper, is applicable very conveniently to the pro- peller extrapolation problem. As is implicit in the ship model problem, it is presumed, however, that che pressure distribution remains unchanged wich change of scale at the same Froude number. This possibly restricts the use of the method to showing the influence of blade-friction scale- effect; and experimentally the method is best applied to those types of screws which suffer least from life scale-effect, ie. those having cor rect leading-edge tilt and suitable blade sections. (2) Consider the usual thrust and torque constants T Cre and poe eno pao Let one geosim have the values Cry and Cory while a larger geosim at the same J and nVD values, as required by the Froude law, has the values’ Cr and Co. Due to the lower intensity of frictional re~ sistance, the larger geosim has the greater thrust and che smaller torque. Now as the spe~ cific resistance of each blade section can be re~ presented by a+5R,~® it follows that the difference in blade resistance will be given by WRat ~~ Rpg). This, applied to the whole propeller, gives a particularly simple solution of the extrapolation problem. Thus let Cr: = Cro +Cpy where Cp; is that part of the total resolved propeller friction affected by the Reyn- olds number. Forthe geosim we have Cr2 = Cro + Cra and from this Cro ~ Cry = Cri Cra Now Cp1/Ror Cra/Rna and if Raz = vl/v, where v and I are the mean velocity and length of the propeller blade, then aa = YVR a, where A is the scale ratio of the geosim. Cra = Cpy/A%, therefore vx From this we have from which vx Similarly we can deal with the torque constant variation, Thus Cor = Coo +C"rx Coa = Coo +C’ra therefore fl Cor ~ Con =C’p1 ~C" pp = Ces (1- oso (-4) Con~ Loz 1 Crt vx Thus we can write Coo =Con~ C's (- Similarly, ©. Cr f+ St 1 os =en bo S 4-2 and, extending the same analysis to efficiency, we have Cm REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS Pa Cy and Ca + ou ony £80272. or = 36 ease or Va 40 ANVASNOD, ISRO a bop w= 64 oo ja = 100 FIG. 3.—Diagramm: ic Variation of Cz and Co with Seale Function A (3) Each of the above expressions is a linear function of 1/VX. Thus they can be ploeted co this scale-function base as shown in Fig. 3. It ie seen chat the J constant, nVB lines intersect the zero ordinate at a value 2 above the model Cry value. This p is equal to Cz. Similarly for the torque line the value of equals C’r1. These ewo values Cp; and C’p; represent the nodel resolved viscous friction which reduces to zero for infinite Similatly for the efficiency relation we may ex- press this in the form ™m (ce Crs ( 1 Tee s(CFe — mote ce) kaa) (ze Cn) & Cor Cor) \ Ve. which gives the frictional gain in efficiency as A is increased. In terms of the individual model re- sults, this reduces to Ta ~Ts _ Cra ~ Crs att ™h Cr Cor ~ Coa Cor from which we see that wh) (See, Cee vx} \Cr1 Cor dai er C’r1\ . and since (=~ + =") is constant for the Con Cra~Crs Cor ~Co2 rr Cor model family for constant slip and nVD, we have [pat ela where and finally ‘The value of E will be greatest at low slips and low pitch ratios. (4) One of the puzzling features of available data is that when a single model is tested over a Reynolds number and a slip range, in addition to the thrust constant increasing with Reynolds number the torque constant also increases. This is most probably due to the continued existence of lift scale-effect having a greater in- fluence than the opposing frictional resistance, bue in any case there appears always to be a net gain in efficiency. A clearer understanding of this feature may be forthcoming when propeller geosims have been analyzed as herein sug- gested, using the condition nVD constant for the respective contours. This may show that al- though for each propeller the torque constant may increase with nVD yet for the larger propellers at the same nVD the torque constant will be smaller. It is hoped to use van Lammeren’s tests on the Simon Bolivar screws to investigate this point fully. An example is given in the following section. (5) To illustrate the application of the method, but only to efficiency extrapolation, use has been made of van Lammeren’s Simon Bolivar work; and in particular of the data given in Table X and Fig. 12 of his thesis. Van Lammeren gives for each model and the constant J value of 0.70, an efficiency value corresponding to the nearly constant Froude number nVD = 5.4. These results, together with the tank water temperature, are summarized below. 28 Seale 3625S Temp.°C 165 17.2 13.0 12d 17.7 Efficiency 66.3 662 69.0 68.6 69.8, These efficiencies are plotted in Fig. 4 toa base of Va = 1/VX, the abscissa values being corrected for temperature. (See INA paper, 1951.) I is seen that with che exception of the a= 25 result, which appears to be much too low, (the wake also was low for this geosim as seen from Fig. 9 of my 1949 paper) the other results allow of reasonable linear extrapolation. The ship efficiency is seen to be 74.3% against 69.8 for the a=15 screw. This corresponds to an effi- ciency ratio of 1.078 and, of course, refers to the smooth surface. aS \ \L \ a” “i & \ PM py a 8 E + toi a \a * Ase OF Ya FIG. 4,—Variation of Efficiency with Scale Function For the rough surface, considered in the light of the viscous roughness convention, we can proceed as follows. The efficiency extrapolated to infinite scale is lo, the ship efficiency smooth is 75 and rough 7p. With r the viscous roughness then Tr = 71s~ Mo ~Ts) a treatment which is completely analogous to the corresponding ship resistance and wake prob- lems. The problem, of course, is to determine r. If the ship propeller were made of painted steel, Kempf's pontoon tests would give an r value of 0.35, if of “brighe” steel an r value of 0.25, REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS. both referring to rolled surfaces. The normal polished bronze has a much smoother surface and may be taken as smooth as the usual model white- metal propelle:s. For rough-cast bur still normal practice surfaces, an idea of r can be got from some tests of McEntee reported in SNAME, 1916, p. 85. McEntee tested a 1Gin. diameter 3-bladed screw of 1.08 face-pitch ratio made in cast iron, cast steel, and cast bronze. The bronze propel- ler was tested both as cast and as polished; the others, except for che usual edge dressing, were tested as cast. Broadly the change in maximum efficiency was from 72% smooth to 64% as cast, bronze included. If now we apply this efficiency ratio at the correct scale abscissa (a=14) for the Simon Bolivar, this would re- duce the smooth efficiency of 69.8 to 62 and hence r would be 69-8 = 62_ 75.9 = 69.8 as-cast surface structural equivalent to that of a tiveted ship or slightly greater. Accepting this value, the loss in efficiency for the as-cast ship screw becomes 4.3 — 1.28 (75.9 ~ 74.3) ~ 74,3 ~ 2.04 = 72.26 1.28, implying an TR and thus, the ship efficiency drop is not nearly so much as that on the model. Actually the rough- ness of 1.28 is probably too high as such, since it appears fairly certain that some of the e! ciency loss observed was due to the as-cast thicknesses being greater than that of the final polished surface. This is evident from the fact that all the as-cast propellers show a greater effective pitch, than the smooth blade. As rough ness itself would reduce the effective pitch, the actual gain can be due only to greater thick- ness. In the absence of more precise knowledge it thus seems more rational to reduce the value of 7 somewhat and the unity value previously proposed for the plated hull would appear to be acceptably near the truth for the as-cast un- polished propeller. For the polished bronze propeller, in the present state of knowledge its roughness can be taken as zero. The foregoing shows that so long as normal material surface condition merely induces vis~ cous-roughness resistance and not sand-rough- ness resistance, the effect of normal propeller surface condition on efficiency is always likely to be small; and this it is felt always has been professionally appreciated. I am fairly sure that unless a ship screw is well covered with a shell growth, the Prandtl-Schlichting roughness convention cannot possibly apply. REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 29 The inherent simplicity of the method now developed for propeller performance extrapola- tion, involving as it does no precise knowledge of the blade frictional resistance and being com- pletely analogous to that already developed for ship resistance extrapolation, should appeal to the profession. It accomplishes much more than any other system of so-called formulation and therefore should, it is hoped, eventually merit general adoption. CONMENTS BY DR. E. V. TELFER I have developed a means of combining the Froude law for propellers and the Reynolds By so doing I avoid the necessity of describing any particular section of the propeller as that at which the Reynolds number should be calculated. At this stage I wish to mention the emphasis placed by Mr. Ross and Dr. Robertson on the necessity for really accurate models before any attempt at geosim comparison is attempted. AS a result of my own investigations here, and my own professional experience, I would say to those responsible for the actual making of the models that the most important error encountered in model making and in model checking is in the actual slope of the blade at the trailing edge. Some very important propellers have been liter ally miles out in the trailing-edge slope, particu: larly at the two-thirds radius, which completely governs the thrust which can be developed by the propeller. That slope is the most important contribution to the effect of pitch in that section. In our recommendations I strongly emphasize the necessity for checking trailing-edge slope as the most important variable in comparative propeller tests, My method has another advantage in that it en- ables comparative roughness tests to be carried out. By making use of some of McEntee’s early work I again show that the type of roughness you get in propeller surfaces appears to be the same as that in ship surfaces. It is what I call a viscous-toughness resiscance and not a sand- roughness resistance. I show how quite a big roughness difference on the model gradually reduces itself to a relatively small roughness dif- ference on the ship. Only further research can determine whether that is the accurate deduction to make from this experience. I strongly em dorse Dr. Allan’s plea for geosim tests by as many people as have the facilities to conduct them. . In connection with the carrying out of syste- matig series of tests, I should like the meeting to consider whether the very basis of such com parative series should not be changed, In the past we have had the Froude series carried out at a constant translation velocity. Then, through Schaffran and Taylor up to Troost and van Lam we have had the constant revolution seties. In the discussion of Dr. Troost’s North- east Coast paper I emphasized the necessity of carrying out such comparative tests at a constant thrust, because the constant-thrust criterion is that which is eventually used in the ship-model comparison itself, By making constant thrust the basis of the comparison you at once upgrade the advantages of the low-pitch-ratio screw and keep in better perspective the possible disadvantages of the high-piech-ratio, although apparently high efficiency, screw. If a position could be arrived at in the course of our deliberations on that point it would be very useful to the profession from many poiats of view. COMMENTS BY PROFESSOR B. V. KORVIN-KROUKOVSKY ‘At the Experimental Towing Tank we are inter- ested in propeller testing primarily from the point of view of measuring hull efficiency. Since this involves, in our case, the use of small models and necessitates a comparison of the propeller in the free-running and in the behind conditions, it becomes necessary to appraise the effect of the scale or Reynolds aumber and the effect of turbulence. As has been the case in all phases of small-model testing, while the immediate activity is centered on the small-model teck- nique, by far the broader and more important re~ sult is a better understanding of scale effects for all model sizes. The first series of tests for scale effect was conducted with the model hull and propeller of the motorship San Francisco for which the data were first published by Kempf [1]. A model of the same propeller was made from Kempf’s draw- ings by the David Taylor Model Basin, the free~ running tests were made, and the data were kindly furnished to the Experimental Towing Tank. At the Experimental Towing Tank a small replica of the Taylor Model Basin model was made, using a pantograph attachment to a lathe, and the model was used for both behind and free~ running conditions, with a few tests behind the turbulence-inducing grid in the latter case. The data on the model sizes, the Reynolds numbers, and the performance at the value of V/nD corresponding to maximum efficiency are PNanbecs im rackets seer to references listed at the end TABLET MODEL PROPELLER PERFORMANCE AT V/nD OF MAXIMUM EFFICIENCY REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS OP meema con ay given in Table 1, taken from Reference [2]. els SSRag RSF By These reat data were reduced to the performance e sddd dds sé of a blade section at 74% of the radius, using Goldstein's theory and generally following the & principles described by Driggs (3] and Lerbs {4]. if The profile drag resulting from this analysis is a\s 2 Rosng SPS 8S plotted against Reynolds number in Fig. 5, also pega8 see 33 taken from Reference [2]. For comparison, the " wind-tunnel data for the RAF- airfoil section, é obtained in an early NPL. wind tunnel, are also plotted. The salient features broughe out by this plot 528 ge (a) While there is a large scatter of the data, gee $5 the general tendency to follow the line of the slope corresponding to the turbulent boundary & layer is apparent (b) The profile drag is shown to be about twice the friction drag of a flat plate of the same pro- jected area (c) The data obeained from the propeller tests generally agree with the data obtained in a wind tunnel of the type in which a high degree of tur bulence is expected. The profile drag listed under (b) is very high, equal to double the friction drag. Possibly this may be explained by the old-fashioned type of sections, in the case of the RAF-6 airfoil and the Joukowsky-type profile in the propeller mod- els. Both items (a) and (c) tend to give the im pression that the boundary layer is turbulent, and further comfirmation may be considered to ‘exist in the fact that brief and incomplete tests ea behind a turbulence grid did not show any change in propeller characteristics. If this conclusion is correct, then it is surprising to find the existence of the turbulent boundary layer at the propeller sections operating in calm water at such a low Reynolds number. There apparently must exist an effective turbulence stimulation, and a hy- pothesis is made that it is created by the una- voidable vibration of a rapidly-rotating propeller even though this vibration is too small to be ob served. A contributing cause, of course, is again the type of blade section with a thick leading edge and harsh curvatures in its vicinity. Not so good a picture is observed in the case of the life coefficient, which, in all cases, was found to be well below the coefficient which could have been expected on the basis of the airfoil and propeller theories after applying Hill's curvature correction [5]. The comparison with RAF-6 wind-tunnel data indicated that this is due to the deficiency in the angle of zero lift at low Reynolds number. The variation of this angle with Reynolds number is shown in Fig. 6. The hypothesis that the vibration of the rotating 0.0225 0.0217 0.0246 0.0208 0.0234 0.102 0.109 o124 0.106 0.126 (O.7AR X 10° 0.045 0.056 0.067 0.073 0.078 0.086 0.153 0.335 8 4 Z t g =60n 800 1000 1200 1308 1400 204 363 792 175 300 po 385 fe spf Model 3542 0.75 0.323 fe 0.80 Eat Son Deacription® ETT Model 52 98: 80 DTMB Model 2214 Fab at Maximea Efficiency D I i Diameter b= J lier Hodel Designation ‘Number x Laboratory ope Designation Peo REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 31 PROFILE DRAG COEF., Cy =0/1/2¢ AU" 8 & 3 30 40 5060 80 10% REYNOLDS NUMBER, Rn = Ub/V FIG. 5,—Profile Drag Coe! propeller model provides turbulence stimulation is supported by this diagram, which shows that at any given value of Reynolds aumber a rapidly- rotating small propeller approaches a large-scale condition more closely than a slowly-rotating large propeller. In all cases, however, the de- ficiency of the zero-life angle is appalling and indicates that a difference of some 2 degrees in the effective pitch angle of a model and a full-scale propeller can be expected. In this connection, it should be noced that Dr. Lerbs jents of Blade Sections at 0.7 Radius [4] has found a satisfactory agreement between the lift coefficient as obtained in propeller tests and in the wind cunnel at a Reynolds number of about 0.77 x 10°, which can reasonably be ex: pected on the basis of Fig. 6. It is interesting to observe, furthermore, that there is no significant difference between the behavior of the “large” and “small” propellers, and that tests of propellers normally accepted as being of reasonable size and at Reynolds num bers as large as 0.3 x 10* can be expectéd to 2 ° RAF EFFECTIVE Uo, DEGREES 2 se Rar 6 RovMPTOTE 4.06 3 6 7,8 910 6 7 8 9108 REYNOLDS NUMBER, R, = Ub/v FIG, 6.—Angle of Zero Lift a» for Various Sections 32 REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS give results considerably different from the full scale condition, particularly in regards to the- effective pitch. The tests of the M.S. San Francisco model were undertaken primarily because of the availa- bility of Kempf’s daca for comparison. As far as the propeller is concemed, it did not prove to bbe a good choice because of the special type of blade section, which is not commonly used. It was questioned, therefore, whether the same conclusions would have been reached in the case of a propeller of a more modern type, such as the propellers used in the van Lammeren (or Troost) series. Two propeller models were available from the currently-conducted self-propulsion tests, a four bladed and a three-bladed one, both designed on the basis of the van Lammeren charts, series B [6]. There were no comparative” tests, but the cest data were compared to the data expected from the charts. This comparison for the three- bladed propeller is given in Table 2. Ie is inter esting to note that although the Experimental Towing Tank model was much smaller, the speed of rotation permitted i to attain higher Reynolds numbers than the one used in the van Lammeren TABLE 2 COMPARISON OF PERFORMANCE OF VAN LAMMEREN TYPE B-3-50 MODEL PROPELLER AS ESTINATED FROM CHARTS AND AS OB TAINED IN THE ETT TEST AT J =0.60 A Be © Kr 0.099 0.087 0.097 Ko 0.0156 0.0125 0.0141 Efficiency 0.605 0.663. 0.657 (A) ETT test—model diameter 0.390 fr; 0.288 ft; 800 rpm; Ra = 0.117 x 10" (B) As estimated from van Lammeren's charts which are based on Ry = 0.136 x 10° (C) ETT test—same model as in (A); 1400 rpm; Ry = 0,205 x 10° Note: The ETT model propeller is assumed to be machined “‘as specified.” No check measure- ments have been made. A more complete analysis, such as that de- scribed in the case of the M.S. San Francisco propeller, has not yet been made, but it is inter esting to observe that the ratio of the efficiencies of tests made at 800 and 1400 rpm is the same as the ratio of Reynolds numbers to the power 0.15, thus again indicating the behavior chara teristic of the full-developed turbulent boundary layer. In this series of tests, the free-running tests were repeated behind a grid consisting of a set of Yein. diameter vertical rods spaced 1 in. apart and placed about 1% diameters ahead of the model propeller. The rods produced, of course, a wake velocity which was allowed for by the usual method of thrust identity. At 1400 rpm the wake fraction was about 0.10 in value, At both 800 and 1400 rpm, the test behind the eurbulence rods, indicated 3% reduction in the torque for the same thrust, but indicated no change in the variation of the efficiency with Reynolds number, thus again supporting the view that the boundary layer was already eurbulent and, therefore, was not sensitive to the additional curbulence stimula- tion. ‘The rods also produced a sizable wave so that the propeller blades operated effectively at a fluctuating angle of attack. Possibly this is « sponsible for the reduction of the torque. In conclusion it may be said that in the free~ sunning condition it may be easier to obtain high Reynolds numbers with smaller, more rapidly- rotating propeller models, and that such model vibration as occurs in che process appears to be a beneficial factor. The degree of this vibration, however, is small enough to be imperceptible to the observer. The question of the propeller model size is therefore largely the question of obtaining a sufficiently high standard of precision in manu- facturing, polishing, and measuring the model. The probability of the variation of the lift co- efficient of the blade sections, and the resultant change of the effective pitch with Reynolds num ber within a commonly accepted range for model testing, looms now as the major aspect of scale effect, and at the moment there are not enough data available to evaluate this effect. Apart from this uncertainty of the effective pitch, charts such as van Lammeren’s, based on model tests, apparently can be extrapolated to full size by applying a correction to the blade profile drag based on the turbulent friction curve. In this connection, charts of the type shown on Fig. 7 will be found useful. The upper most curves on the charts give the ideal effi- ciency of a propeller operating at a given V/nD ratio as obtained from an advanced theory, say REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 33 95 90 é a 3 B EFFICIENCY = 7;- 7p & -60 55 50 o1 23 4 5 Tee Wiemrev? FIG, 7.—Curves of Maximum Efficiency on Several Parameters for J = from Kramer's charts [7]. A series of curves of lower efficiency results from the introduction of the drag-lift ratio € by the formula of Bienen and Von Karman as described in [8]. The effect of the changes of the blade profile drag due to any cause—poor section, Reynolds number rough ness, and the like—readily can be seen. REFERENCES, [1] Kempf, G.:"*A Study of Ship Performance in Smooth and Rough Water,” SNAME, Vol. 44, 1936, pp. 195-22: [2] Korvin-Kroukovsky, B. Speed and Size on the Performance of # Model Ship Propeller at Optimum Efficiency,” Trans. New York Acad. of Sci., Ser. Il, Vol. 13, No: 6, April 1951. 2.3 4 5 6 7 8 9 10 1. Pe> P/ Yh Pw REV? [3] Driggs, 1. H.: “Simplified Propeller Caleu- lations,” Journ. Aero. Sci., Vol. 5, No. 9, July 1938, pp. 337-344. [4] Lerbs, H. W.: “On the Effects of Scale and Roughness on Free Ruraing Propellers,” ASNE, Vol. 63, No. 1, February 1951, pp. 58-94. (5] Hill, J. 'G.: “The Design of Propellers,” SNAME, Vol. 56, 1948, pp. 143-192. [6] Van Lammeren, W. P. A., Troost, L., and Kéning, J. G.: “Resistance, Propulsion and Steering of Ships,” The Technical Publishing Company, H. Stam, Haarlem, Holland, 1948. (7 Kramer, K. N.: “The Induced Efficiency of Optimum Propellers Having a Finite Number of Blades," NACA, TN 884, January 1939. [8] Bienen, Th. and Von Karman, Th.: “Zur Theory der Lufesehrauben,”” Z.d.V.d.l., Vol. 68, Nos. 48 and 51; Vol. 69, No. 25. 34 COMMENTS BY MR. At the National Research Council we have been testing propellers of around 6.5 in. in eter for some little while now and I would summarize a few of the results obtained, For open water tests, each of which is carried out at constant revolutions, we find chat a Reyn- olds number of 0.14 x 10° for 100% slip is the critical region below which we cannot get con- sistent results, With increasing Reynolds num bers, say cortesponding to 800 rpm on the above diameter, up to 1200 rpm, we get increasing efficiencies, an increase in thrust coefficients, and a falling off of torque coefficients. ‘Around 600 rpm, again for the above diameter, the resules do not correspond with a turbulent drag coefficient line. We get higher efficiencies and lower torque coefficients than for 1200 rpm but, surprisingly, a decrease in thrust coefficients, below the 800-rpm values. T. MATHEWS COMMENTS BY SR. L. MAZARREDO As Dr. Allan says, the major scale effect in the blades of a propeller is the movement of the separation point on the suction side, which af. fects not only friction on the blade but also the life characteristics of the profile. Therefore, it would not be expected that the methods which take into account the water viscosity, introducing only the section’s profile drag, will lead to re- Tiable resules. A direct approach to the problem is preferable by testing (as the report states) several propeller serie Some of these experiments already have been performed and may provide a first step. But since the results seem not to be conclusive, or at any rate comprehensive enough, to show any general conclusions, these tests must be supple- mented by new ones. A large amount of work is involved, since the tests must not only be per formed, but be analyzed. It would be interesting to study each propeller at various load conditions. Hence we agree that such a program must be distributed, assigning to each of the basins a complete series. Some of them might be tested jin more than one establishment for control pur poses. oo Referring to turbulence stimulation on propel- lets, much work ought to be undertaken. I sug- gest that priority be given to stimulation in the water, ahead of the propeller, because this method, besides providing the necessary turbu- lence, (if successful), would provide more similar conditions between the open-water test and those behind the ship, as well as those in the cavita- tion tunnels. REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS Ie would be interesting to know the effects of water turbulence, because the results will be useful when studying the comparative tests car ried out in the cavitation tunnels. To attain this aim it would be, of course, convenient to begin with tests that show the effect of different turbu- lence degrees on different propellers, including possible overstimulation. Water turbulence has been proposed several times and I think can be best accomplished by putting a grid or a wire-screen disc ahead of the propeller. COMMENTS BY PROFESSOR M. YAMAGATA In dealing with the gir-draw (air Leakage) prob- lem of a propeller working near che water sur- face, Dr. Kempf adopted Reynolds number as a criterion for model-propeller experiments. It seems rather preferable to use the Froude num ber, expressed by nVD7g, instead of the Reyn- olds number, which may be confirmed also by Dr. Kempf's experiments, By analyzing the results ‘of Mr. Shiba’s extensive air-draw experiments, the critical value of nVD7g has been found to be about 2.5, roughly depending upon blade-area ratio, immersion-diameter ratio, etc. within his expetimental ranges. ‘As, in the case of D=0.20 meter, the re- suleant velocity Vx at 0.7R, corresponding to nVD7@ = 2.5, has a mean value of about 8.5 me- ters per second over a range of advance con stant J from 0 to 1.0; the corresponding Reyn- olds number is Va(Bare) _ 8.5(Baan) where the units are in meters and seconds. From this we may know that, unless too small model propellers are employed, the minimum value of mVD7g will give a Reynolds number larger than the critical value. COMMENTS BY MR. R. W. L. GAWN Briefly, the requirement is to expand the re- sules of model propeller tests to afford realistic predictions of ship propeller performance. One requirement to satisfy the law of similarity is that the Reynolds number should be the same as that of the ship propeller. So far that is physi- cally impossible, but it will be instructive to see what the scope is. The Reynolds number of ship propellers varies considerably with the class of ship and the speed. For example, the Reynolds number of a fast warship propelier at full speed REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 35 is of the order of 100 million and at cruising speed, about 50 million, The value for fast lin~ ers is about the same as for warships. The pro- peller of a cargo ship at service speed operates at about 20 million and less for slow ships. A minimum Reynolds number of 0.5 million was suggested at the 1937 Conference as suitable for tests on model propellers. Many tests have, how- ever, been made at a larger Reynolds number. For example, the tests at Haslar referred to in Refer- ence [19] of the Committee’s report were made at a value of 2.7 million, and the open-water tests in Reference [11] at 0.8 million. I made reference at the 1948 Conference to some tests of propellers of 20 in. diameter, part of a methodical series commenced fourteen years ago, covering a wide range of pitch, blade area, slip, blade section, blade outline, number of blades, and size of propeller. The work was in- terrupted by the war, but a series for one blade~ thickness ratio has been completed and progress made on another series of modified ratio. The Reynolds number extends to 4.3 million. Gen- erally the experiments were made at constant revolutions, but a number of propellers were tested at various speeds of advance and at various constant revolutions. No. variation in thrust or torque coefficients was recorded over the range tested. While he was at Haslar, Dr. Lerbs made an interesting analysis of each of these results. ‘The fictional resistance coefficients deduced from the measurements of thrust and torque in- dicated that the propellers were working in tur- bulene flow. This confirms the deductions from the tests of blade sections in the CAT given in Reference [20] of the Committee's report, which showed that scale effect is very small for seg- mental sections comparable with those of the propellers at the working range of incidence. The wind-tunnel tests extended up to 7 million Reynolds number. We want to cover a value of 100 millions, although for many ships a smaller value is acceptable. There is a wealth of ex- perimental data for frictional resistance up to 50 nillion Reynolds number, and a certain amount of information at higher Reynolds number. We are not up against quite the same sort of problem as with the skin friction of the hull, for which predictions are required up to 1000 mil- lion Reynolds number. We have got frictional data within the range of Reynolds number of ship propellers. The Committee sums up the causes of scale effect on the performance of screw propellers as: (a) A change from laminar to turbulent boundary- layer flow on the blades (b) The continued decrease in turbulent fric- tion with increase of Reynolds number. The analysis to which I have referred shows that the transition occurs at remarkably low Reynolds numbers, well below those for the test series considered, The decrease in turbulent friction with increase of Reynolds number can be approximated very closely for expansion to the ship propeller, since the effect is not large over the range of Reynolds number of interest for ship propellers. ‘The Committee recommends that several scale~ effect series of propellers should be tested with high-power dynamometers to include investiga- tion of roughness effects on propellers of about 2 fe diameter. The Committee notes that the scope of tests required is considerable and an- ticipates that cooperative effort will be required for reasonable progress. The work on the series of propellers of 20 in. diameter at Haslar has been mentioned in case the Conference cares to take advantage of the substantial progress already made in formu- lating new programs of propeller tests. No oppor- tunity has yet been found to publish results of the completed portion of the work, but it is an- ticipated that this will be in the near future, per- haps some time next year. Dr. Allan referred co the Series I and Series II propellers of the cavitation test program, and suggested they should be tested in open water. That is included in the program which the Cavi- tation Committee have proposed. Limited prog- ress has already been made. The remarks of the Committee as to the neces sity for insuring that propeller surfaces are smooth are fully endorsed. Both calculation and experiment confirms that serious loss may arise from even a moderate degree of roughness. COMMENTS BY PROFESSOR F. HORN The suggestions brought forward by Dr. Allan on behalf of the Committee seem to me to be quite on the righe way. In fact, special emphasis should be laid first upon assuring turbulent flow along the whole surface of the propeller blades by some sort of turbulence stimulation. I con~ sider this as the only way to get a true basis for a reliable prediction of the workings of the full- scale propeller. The ascertainment of the means by which turbulence is assured will be an im- portant task of future research. Personally, I ‘Suppose that the use of sand strips on both sides 36 REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS behind the leading edge will be favorable, In this research the means which already have been’ proved in ship model research for ascertaining the extent of laminar flow will, of course, be utilized. ‘Once in such a way a true experimental basis has been secured, at which full eurbulence over the whole blade can be assumed, then further im- portant progress in a sound evaluation of the di- rect test results, as represented by the coeffi- cients Cr and Cg, seems to me to have been attained by the very thorough analysis by Dr. Lerbs in his valuable paper on the effects of scale and roughness on free-running propellers. In this paper, Dr. Lerbs has proved that the val= ues of Cz ad Co, measured at a certain value of A, can be expressed with great accuracy by the life and drag coefficients C;, and Cp of the blade profile at 0.75 radius. Therefore, since at fall turbulence Cz will not vary with Rn, the change of Cr and Co with Ra, and the magni= tude of these values at the full-scale propeller, can be deduced from the change of the drag co- efficient with Rj. Within this method, both smooth and rough full-scale propellers can be considered. Furthermore, Dr. Lerbs’ statement that the roughness conditions outside of 0.5R are prac tically responsible for the whole roughness ef- fect, will be of special importance and draw at- tention to the need of very smooth surface finish of the full-scale screw in this outer range. In my opinion we have succeeded, thanks to the very able work of the Committee, in coming a good deal nearer to a satisfactory solution of a very complicated problem, at least in so far as we now can see a rather clear way before us, although the remaining work still may be rather laborious. A satisfactory result would be the more welcome as the objections which still exist against the use of small models for propul- sive work would then be greatly removed. COMMENTS BY DR. S. L. SMITH I think it is generally agreed that scale effect may occur on many models, and none of the work- ing rules for minimum Reynolds number in order to avoid scale effect is entirely reliable. We, of course, as users of model data are interested in the efficiency and the performance of the full- sized propeller. Therefore it is most desirable that work on propeller scale effect should con tinue. By analogy with the turbulence stimulation de- vices on model hulls it may be possible to stimu late full curbulence on propeller blade sections. This possibility has been considered by the ap- propriate committee of the BSRA and we are now at the stage when we are actually about co com mence preliminary tests on model propeller: The object of these tests is to determine in open- water conditions how best to obviate laminar flow over the blade sections. Lam glad to say that the late Dr. Baker's work on propeller scale effect, which was done for the BSRA, will be published in the near future by the INA in London, COMMENTS BY M, L. P. MAILLARD These are the comments of Vice Admiral E. G. Barrillon, First we must thank the Committee for synthesis work made and congratulate ourselves that in the 1948 conclusions “‘no important data have been overlooked.”” The document listed as item 12, of the Intro- ductory Remarks, has been analyzed bur has not supplied any conclusions. It would be in- teresting to know if the propositions to under- take trials “in a tunnel under pressure and at high temperature” have been followed through. In ‘Summing up the position” the Remarks cor~ rectly report the present state of the question, but it seems premature to state that “'the major effect is a movement of the separation point on the suction side.”* Experience on this sub- ject is insufficient. Definite conclusions can not be formulated as long means of observing laminar flow on the propel ler blades. Greater emphasis should be placed on the following propositions: “A method should be developed for detecting the extent of laminar flow on propeller blades or determining the posi- tion of transition from laminar to turbulent flow.”” It is only by a comparison between such ob- servations on the scale model and on the actual vessel that a sufficient knowledge of the scale effect will be obtained. s we lack effective ComMENTS BY DR. H. W. E. LERBS. Dr. Allan has pointed out in his report that the scale effect of free-running propellers depends on several independent variables. Simplifying the problem by excluding free-surface effects, the variables for a given section are Reynolds number, relative roughness, the turbulence of the flow outside of the boundary layer, and cen~ trifugal forces. Since the influence of each of these variables on the propeller performance is quite different in different ranges, the combined effect is complicated and our knowledge is not REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 7 yet sufficient to understand the problem in its details. ‘A model basin is, of course, much interested in a means of estimating this effect. For this purpose, several tests with series of geometri- cally similar models have been carried out in the past which give an idea of the order of magnitude of scale effect within a small range of Reynolds number. Due to the need for more general infor mation with respect to the influence of the pro- peller geometry and the range of the Reynolds number, numerical methods have been developed in recent years with the aim of reducing the scale effect of a propeller to the underlying physical processes which cause it. ‘0 develop a method which is not restricted by basic assumptions made previously, arbitrary variations of the lift and drag coefficients of the sections have been considered, these variations being aay functions of the radius. The variations of the propeller characteristics are connected with these variations of the section character- istics by propeller theory. Afterwards, special cases can be considered on the basis of these general relations; chat is, that the frictional pare of the drag coefficient varies in the same way as the turbulent friction of a flat smooth plate, the lift coefficient being unaltered, which case corresponds to the scale effect on an over= critical, smooth propeller model. Or the model may be smooth, whereas the full-scale propeller is entirely rough or its roughness varies arbi- tratily with the radius. It may be stated that the general equations, in the case of an over- critical model, can be replaced by a numerically simpler method by which the calculations are confined to one section only. In principle, the general equations can be ap- plied also to an undercritical model test, but in this case our present lack of knowledge with respect to undercritical section characteristics becomes apparent. To establish the basic equa- tions which govern scale effect represents only a small amount of the work necessary in this problem. By these equations the compli- cated scale effect of the whole propeller is shifted to the comparatively simpler scale ef- fect of the sections, but much experimental work still is necessary to obtain the characteristics of series of sections as functions of the afore- mentioned variables, including the undercritical range of Reynolds number. In this range, only a few experimental data are yet available, and moreover these measurements do not apply to propeller sections if laminar separation occurs at small angles of incidence because of the ap- preciable influence of centrifugal forces on a separated boundary layer. As a further complication due to the centrifu- gal forces, the polar curve of a section can no Tonger be’ considered independent of adjacent sections. For these reasons the prediction of scale effect, in particular of undercritical mod- els, would be greatly improved if effective tu- bulence stimulation were possible. In such a case, the great difficulties with respect to the influence of centrifugal forces on the polar curves of the sections could be avoided and the afore- mentioned simple numerical method which holds for a turbulent boundary layer could be applied. Also, a means to determine the extent of laminar flow would be of great help for the same purpose. This brief comment will indicate why the David Taylor Model Basin deems the following work necessary and-urgent for the fundamental study of scale effect on free-running propelle: (1) Work on turbulence stimulation should be carried out and means to determine the extent of Iaminar flow on propellers should be developed. This is, in our opinion, the main problem of ex- perimental research on scale effect on propellers. (2) A limited number of propeller series should be tested for scale effect over a range of Reyn- olds number as large as possible with the avail able facilities. The goal of these tests should be to check the different theoretical methods for predicting scale effect of propellers. These in- vestigations should include studies on rough pro- peller surfaces. In the event that turbulence stimulation on pro pellers does not prove successful, it will be nec~ essary to test: (3) A limited series of propeller profiles both in the over and undercritical range of Reynolds number and to investigate the influence of cen- trifugal forces, relative roughness, and free stream turbulence on the polar curves. This work would enhance our fundamental knowledge on scale effect considerably, and would enable us to cary out predictions of scale effect on free-running propellers with sufficient accuracy. COMMENTS BY PROFESSOR FRANK M. LEWIS There has been quite a bit of discussion on the question of scale effect on propellers. The interest seems to lie in how to take the results of a standard propeller test and extrapolate them in some way to a full-scale propeller so you can apply that result co a ship. I question whether that is possible, whether it is neces sary, of whether it is even desirable. 38 REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS Suppose, for example, you take a propeller test for a 12-in. propeller. Let us assume that by some method we can succeed in getting a reli- able test, a test in which all the basins will agree within a small fraction. We will assume that standard conditions are such that the propel- ler is in full turbulence. You take the results of the tests and plot curves of Kz, Ko, and effi- ciency. What do you do with these data? You have a ship on trial, and from the trial results you have a torque and occasionally you have a thrust. From those two factors you can deduce the ship's wake fractions and that is all you can get. The next propeller that comes along, you hope those factors apply. If instead you take your I2in. propeller re- sults and extrapolate them to ship size you will deduce a new set of slightly different wake fractions. It may be desirable to make some correction to your test results for the roughness on the propeller that you think you will have in the ship, but I see no necessity in going to an extrapolation to a full-scale propeller. The scale corrections are contained in the wake fractions obtained from the 12-in. pro- peller. Since we cannot test a full-scale pro- peller, the scale corrections cannot be separated from these wake fractions, except on a hypo- thetical-theoretical basis. COMMENTS BY MR. J. M. FERGUSON This morning Dr. Allan referred to large model propeller tests carried out in John Brown and Company’s Experimental Tank at Clydebank. Last week, Mr. Comstock, in one of the discus- sions, also made a similar reference. A few de- tails of the propellers and the dynamometer used for the tests may not be out of place. In the early 1930's, the problem of the ac- curacy of the propeller tests for the new Cunard liner then under construction at Clydebank was a vital one. Till then, the usual maximum size of model propeller for open-water tests was 9.6 in, in diameter. Scale effect was doubtful and the relative accuracy of the models was most im- portant. Domestic and other considerations placed a practical limit on the size of propeller which could be made and tested in our tank. Finally, a 2din. diameter was adopted as the largest practicable model and a propeller dynamometer was designed and made to suit this size. It was considered that scale effect from model to ship would be small and thar the open-water efficiencies of such models could be accepted for the ship propeller. An aluminium-silicon alloy—about 5% silicon —was chosen as the material for these 2é-in. propellers. The saving in weight as compared with brass or bronze, particularly in the immersed condition, enabled us to keep the scantlings of many parts of the dynamometer to a reasonable minimum without sacrifice of strength ot rigidity. The face of each blade is carefully finished by hand with a continuous check of progress and ac- curacy on a pitch-measuring machine, the measur- ing heads of which ate scaled in thousandths of inches. The back of each blade then is worked down to the required shape, using the contour method. Since our normal range of model tests seldom exceeds 50 to 60% real slip, the power of the dy= namometer is modest; it has a5 hp motor. So fat, the capacity of the motor has not been taxed to the limi. Torque and thrust are weighed directly on simple balances, which are sensitive to about 0.01 Ib. The motor is arranged on the armature-reaction principle, which simplifies the measurement of power. Our usual procedure is to run each model pro- peller in open water at an immersion of 24 in. to the shaft center, over a range of speed of ad~ vance and rate of revolutions. Each condition is made at constant speed of advance and variable revolutions, The range de- pends upon the characteristics of each case. The values of thrust and torque are converted into non-dimensional form. Each speed condition chus gives a portion of the complete thrust and torque- constant curves. Each portion usually over- laps its neighbour but a smooth line can be drawn through all the spots and there has been little or no indication of discontinuity in the Separate portions of the curve or evidence of scale effect. ‘These curves thus provide the data for estima tion of open-water propeller efficiency and ship revolutions for our estimation of the power re- quired for the ship. Informal Comments ComMENTS By PROFESSOR L. TROOST With reference to scale effect of profiles as used with ship propellers, I call attention to the thesis of my collaborator, Dr. J. Balhan. He car- tied out measurements on four Karman-Trefftz profiles in the NSP cavitation cunnel at Reyn- REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 39. olds numbers of 3 x 10° and 4 x 10° and at lower speeds, Ry = 1.6 x 10° to 2x 10%. From the re- sults it is considered that’ scale effect is pres- ‘ent in the lower range of R, but absent in the range of 3 to 4 x 10%, There was found no dif ference in the profile characteristics going from 3 to 4 x 10°so that we may consider R,, = 3x 10° to be sufficient for avoiding scale effect. COMMENTS BY PROFESSOR C. W. PROHASKA The subject here is of the greatest importance, also for Subject 5, ‘Scale Effect on Propulsion Factors.” In my opinion it is absolutely neces- sary to correct the characteristic curves of the propeller before attempting from che tank data to derive wake values for the ship. We should expect for a clean ship a lower wake than for the model, but very often we find the opposite, which must be due to scale effect on the propeller. The proposition made by the Committee that geosim series should be tested might help us to ‘latify this problem. If the characteristic curves of the propellers of these series were calculated by suieable methods for model and ship condi- tion, and che Latter both for smooth and rough propeller, we might be able to get a broad out- line of the scale effect on propellers. ComMENTS BY Dr. W. P. A. VAN LAMMEREN, I should like to put a practical suggestion before this meeting. We are going to carry out propulsion tests on a series of Victory ship models, and we have also to carry out a series of open-water propeller tests. The smallest pro- peller will have a diameter of 5 in. and the big- gest one for the model boat a diameter of 3 ft. The biggest propeller of the series of rowing models will be about I ft in diameter. I think we could make a propeller between those, say of a diameter of 2 ft, and I should like to offer to the Committee to bring out and report, before the meeting of the next Conference of Tank Superintendents (which might be held maybe in two or three years), the results of these tests. The method by which we are to carry out these experiments could possibly be discussed in the Propeller Committee. I am anxious to know whether another model basin could share with us in carrying out the same tests. We could deliver the propellers, and I should like to know whether another model basin is prepared to carry out the same tests so that we can have a check on our own results and so thatthere will be available in the next Confer ence two series of the same model tests. COMMENTS BY DR. K. E, SCHOENHERR The subject under discussion is an important one. In the years that I was connected with the Model Basin some 2000 propeller models were tested under my direction, so I know from experi- ence that model propeller tests often yield ine consistent results. However, it would be an error to conclude from this that propeller model tests are unreliable. Besides testing propellers 1 have also de- signed propellers for well over 100 commercial vessels. The data that I have used most often in working out these designs are those presented in propeller charts familiar to all of you. If the basic test data from which these charts were con- structed had suffered materially from scale ef- fect, it would seem reasonable to expect occa sionally a’ bad design; however, I can say truth- fully chac all my designs proved entirely accepr- able. From this we may conclude that tests on B-in. of 9-in, propellers run at fairly high speed and rpm are not influenced materially by scale effect. Nevertheless, scale effect—even if slight—may mask other important effects and therefore a thorough investigation of this some what elusive phenomenon seems worth while. From the standpoint of the propeller designer other investigations are needed. Thus, we badly need systematic tests on hydrofoils suitable for propeller blades. Most data of this nature were obtained in wind tunnels and with the require- ments of the airplane rather than the marine pro- peller designer in mind. I believe that, if prog- ress is to be achieved, this gap in our knowl- edge should be closed. Our information on high-pitch-ratio propellers, propellers with abnormally large hubs, and con- tra-rotating propellers is also incomplete. Such information is needed for torpedo work. The model basins would perform a great service to the profession if they would supply this in- formation. COMMENTS BY DR. R. T. KNAPP, I would like to suggest to the Conference that they are not alone in their interest in the class of devices which includes propellers. The manu- facturers and users of hydraulic machinery, espe- cially cenerifugal pumps and turbines, are en- countering similar problems. There are also many similarities. For example, the striking similarity between propellers and centrifugal pumps is seen in the common method by which energy is transmitted from the machine to the fluid. In view of all of che similarities, it would be very desirable to correlate the data from all 40 REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS of the different types of machines included in the class to see how much new light can be shed on the performance characteristics of each type by the knowledge that has been accumulated concerning all of the others. In recent years many large hydraulic machines have been built that are directly comparable so far as size, Reynolds number, and horsepower are concerned, with the largest propellers. Further~ more, problems of mutual interest have been stud- ied, such as the effect of entrance conditions on the performance. This is directly comparable to the effect of the distortion of the flow caused by the ship on performance of the propellers. The propeller designer may find it profitable to study the test results from large pumps and turbines. Usually, both model and full-scale tests are available and generally it has been found possi- ble to get relatively better accuracies on the full-scale tests with these machines than with propellers. Such tests should shed more light on the old problem of scale effect. In view of this community of interests, it is suggested that this Committee establish direct liaison beeween their work and similar work in the hydraulic ma- chinery field, both in this country and in Europe. ConMENTS BY MR. L. PEHRSSON I should like to say a few words about Pro- fessor Lewis’ remarks on the meaning of scale effect on model propellers. I think that if pro= pellers are designed from model tests where scale effect has been present the design in many cases is noticeably affected, Fig. 8 shows two sets of propeller curves, one essentially without and the other with scale ——Without scale effect —ivim ee —s FIG. 8 —Characteristic Open-Water Propeller Curves effect. From the curves you can see that a pro- peller with optimum diameter calculated from the curves without scale effect will have a larger pitch ratio and a smaller diameter than a pro- peller calculated from the other curves. Therefore | consider it important to have as little scale effect as possible in model propeller tests, both in order to have the best propeller dimensions and in order to avoid mixing the in fluence of propeller scale effect and the scale effect on wake when analyzing ship trials. CoMMENTS BY MR. W. P. WALKER Practically no new data have been made avail- able to the Reporter since the last Conference. ‘As the problem is a vast one I suggest the Com- mittee subdivide it as follows: Long-Term Project (1) Tests on propeller geosims (2) Tests on blade sections, including rough ness effects. Short-Term Project (3) Detection of laminar flow on propeller blades (4) Control of such laminar flow Continuing Project (5) Theoretical extrapolation to full scale Items 1 and 2 are suitable for the larger tanks, items 3 and 4 for the smaller tanks. The Confer- ence’s resources are thereby more efficiently ap- plied to the solution of the problem. COMMENTS BY PROFESSOR L. C. BURRILL I wish to stress the importance of accurate measurement and suitable tolerances for model propellers. May I say that I have been responsi- ble for the design and finishing of thousands of propellers. Speaking from that background, I endorse Dr. Schoenhert’s statement that full- scale propellers do behave in accordance with expectations, We do obtain the results in the ship size which we deduce from standard-series data, derived from model propellers of reasonable size. For the testing of model propellers, I re= alize the difficulty of making even a 1G-in. pro- peller correctly to scale; the difficulty of making 2 9-in, or a Gin. propeller must be immeasurably greater. Knowing the features included in the design, I have seen model propellers in the past which departed visibly from the drawing, and even differed from blade to blade. I therefore agree with an earlier speaker that part of this so-called scale effect probably is not scale ef- REYNOLDS NUMBER FOR MODEL PROPELLER EXPERIMENTS 4 fect at all, but is in the models. For this reason, I strongly recommend that in carrying out tests with models of different sizes, the propellers be so designed that they can readily be machined all over the face. Elaborate sections should be avoided, and simple pitch variations be adopted, to obtain that degree of accuracy necessary for proper comparisons. T agree substantially with Professor F. M. Lewis in his remarks about wake analysis, but I disagree with him on one important issue. I be- lieve it is important to pursue these studies of propeller scale effect, and to find out how and why different results are obtained wich a G-in. model and a 9-in. model. If we can reconcile the 6+, 9, 12+ and 16-in. propeller results on a logical basis, we then can use such models to extrapolate to larger sizes. If, on the other hand, we cannot reconcile the small models with the big models, then we should discontinue the present habit of knowing chat there is a Reynolds number limit which should be applied, and then of disregarding quietly that limit when the ship model scale requires a smaller screw than is prescribed by chat limit, still expecting the results to be correct. COMMENTS BY DR. J. F. ALLAN For the guidance of the panel which has to meet after this meeting and formulate the de- cisions which will be put forward later to the Conference for approval, I interpret the situation approximately on these lines. The subject has developed from one of terms of minimum Reyn- ‘olds number, into one of correction of model pro~ peller resules for scale effect. The study of methods and processes of correction for scale effect divides itself, as Mr. Walker suggested, into long-term, short-term, and continuing items. The desirability of testing a number of pro peller geosim series has been indicated and we have noted the possibility of a cavitation series; also Dr. van Lammeren’s statement regarding the geosim series for the Victory ships. That is a stare in this matter of geosim series. There also has been general agreement on the necessity for investigating the extent of laminar flow on model propellers, and developing meth ods of controlling that flow. Some mention also has been made of the importance of blade-sec~ tion studies. Dr. Knapp’s comparison between hydraulic pump work and the ship propeller is extremely important. The Committee should make note of the suggested liaison between these two model studies with benefit to the study of the propeller scale-effect question. The other point which has been emphasized and which of course has been recognized all along is the paramount importance of accuracy of manufacture in model propellers. COMMENTS BY PROFESSOR L. TROOST I think it important to state that we can obtain very dependable results for all practical pur~ poses in self-propulsion work with propeller mod- els of 8 to 10 in, in diameter. For the academic part of our work, comparative tests with larger models are indicated, however. This should not deter the practical user of model basin work to have his ship models tested with & to 10-in. propeller models and to make use of the results of good systematic screw series, based on mod els of this size. Ie has been made clear that we are emerging from propeller scale effect into propulsion scale effect, which is quite a different thing. It may be necessary for the Committee on Propulsion Scale Effect and the Committee on Propeller Scale Effect to come together, because the one cannot be separated easily’ from the other. Since we propeller designers are not designing large propellers to work in the open, but behind a ship, it should be made clear in our minds that propeller scale effect has a very definite bearing on propulsion scale effect, wake scale effect, and so on.

You might also like