A project sf Volunteers in Asia mformarlce Test of a Savonius Rotor Technical Report TlO by: M.H. Simonds and A. Bodek Published by: Brace Research Institute MacDonald College McGill University ste. Anne de Bellevue 800 Quebec, Canada H9X 1CO Paper copies are $ 2.25. Available from: Brace Research Institute MacDonald College McGill University Ste. Anne de Bellevue 800 Quebec, Canada H9X 1CO Reproduced by permission of the Brace Research Institute, McGill University. Reproduction of this microfiche document in any form is subject to the same restrictions as those of the original document. McGill lJn!.versity Faculty of Engineering BRACE RESEARGH INSTITUTE Performance Test of a Savonius Rotor. Technical Report No. TlO January 1964 by M.H. Simonds, B.E. and . A. Boded BRACE EXPERIMENT STATION St. James Barbados, West Indies For further information .please contact: Brace Research Institute Macdonald College of McGill University Ste. Anne de Bellevue 800 Quebec, Canada H9X 3M1 CONTENTS Page Summary Contents Nomenclature Introduction Description Conclusions Acknowledgment References Appendix: Application of Savonius Rotors to Pumping Table of Results FIGLIRES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. The 18 sq. ft. Savonius Rotor 9 Dimensions of the Rotor 10 Silk Band Brake 9 Torque Characteristic of Savonius Rotor 11 Power Characteristic of Savonius Rotor 12 Power Characteristics for Range of Wind Power Machines 13 Savonius Rotor Harnessed to a Dkaphragm Pump 14 Variation of Pump Flow with Wind Speed 15 Variation of Total Head and Pump Efficiency with Flow Rate 16 Estimated Pumping Performance of Savonius Rotor and Fan Mill 17 Performance 'Test of a Savonius Rotor bY M.H. Simonds, B.E. and A. Bodek SLlMMARY A performance test was carried out on an 18 sq. ft. Savonius rotor on an open site. The shaft output was determined by measl-~-ing the brake torque exerted by a silk band brake. The shaft speed and power output were non-dimensionalised with wind speed; the results are presented as characteristics of torque and power coefficient plotted against tip speed ratio. The peak power coefficient of 0.1~ was attained at a tip speed ratio of 0,7, In an appendix, consideration is given to the application of Savonius rotors to pumping, and test results are given for the 18 sq.ft. Savonius rotor 'Karnessed to a diaphragm pump. It is concluded that a Savonius rotor pumping system operates quite satisfactorily, and is indeed a practical design of windmill. It is, however, only about half as efficient as the conventional fan mill. When used for pumping water, the diameter of the diaphragm pump must be -chosen to suit the total pumping head. The system seems best suited for pumping in cases where the well depth does not exceed about 20. feet. Further work on the Savonius type rotor should be of two kinds. One approach should be to investigate the effects of changes in the shape of the rotor wings, etc., through wind tunnel studies. The other approach should be related to specific applications in which use of different materials, etc., should be tried, depending on specific local conditions. -1- Symbol d S t v T N P md N p= -- 60 v cT= g+d c= p = P 2y' CT u2pv3s Quantity Diameter of rotor Projected area of rotor Air density Wind speed Shaft torque Shaft speed Shaft power Tip speed ratio Torque coefffcient Power coefficient Units Ft. sq. ft. Slug/cu.ft. Ft./set. lb. ft. X.P.M. ft.lb./sec. -2- INTHCDUCTICN A Savonius rotor is a vertical axis wind rotor (See Figs. 1 and 2). Such rotors were developed in Finland by J. Savonius in the period 1925-1928l. Savonius himself naturally carried out many experiments, the majority of which were simply comparrZtive model tests. However he did harness a 20 sq, ft. (projected area) rotor to a pump, and from the figures he presents, it is possible to get a rough idea of the rotor's performance. During World War II, New 'York College of Engineering did some wind tunnel tests on a rotor2. The scatter of their results is much greater than can be justified for wind tunnel testing, and in other ways the report is not satisfactory. However a more detailed picture of the rotor performance can be pieced together from this report. The present test is part of an investigation as to whether Savonfus rotors might have application as wind power units in underdeveloped countries. The idea is that such rotors could be constructed and erected by local iabour out of readily available materials at very low cost. The rotor tested was made out of two 44 gallon oil drums, plywood and plates, a 2" diameter,pipe, two self-aligning ball bearings and timber frame. The appendix deals with the application of Savonius rotors to pumping. Description The leading d<mensions of the rotor are shown in Fig. 2. The rotor was constructed out of two 44 gallon 011 drums, which were bisected, and mounted on plywood end plates and a 2" diameter shaft. The shaft runs in two self-aligning ball bearings, which were bolted to the timber frame. The brake consisted of a cylindrical brass drum, of diameter 6.813", with a silk band of thickness 0.015" rubbing on it, The silk band was tensioned wiih weights and spring balances, as shown in Fig. 3. The brake torque is the product of the radius of the drum and the difference in tension between the two ends of the band. The drum was cleaned with fine abrasive before use, and once the bxnd tension was adjusted, the brake torque remained constant regardless of changes of R.P.M. A tacho-generator was used to measure rotor R.P.M., and for the recorded test results, a vane-mounted pitot-static tube and micro-manometer were used to measure wind speed. Earlier tests had been done, using a Taylor "Wind Scope"; however, the values of torque coefficient obtained from these measurements were unexpectedly low. The Wind Scope was later calibrated against the pitot-static tube and found to be reading 1 mph high at 10 mph, this erl-or decreasing with wind speed. The effect of this 10% error in wind speed was to make the Lorque coefficient 20% low and the tip speed ratio 10% low. These errors had a cumulative effect on the torque characteristic. The experimental points used to plot the torque characteristic of Fig. 4 have a scatter of 3- 30X This is due to the fact that conditions were nothing like steady as a result of atmospheric turbulence. The method of obtaining results was as follows: The brake was set to a particular value of brake torque whLch thereafter remained steady; the wind speed and the R.P.M. were continuously observed, and during periods of relative steadiness, readings were taken. The power characteristic of Fig. 5 was derived from Fig. 4 using the relation:- C p = 7-b %: +clusios 1. The essential results of this report are summarised in the graphs of Figs. 4 and 5. From these two graphs, it is possible to predict the shaft output of any similar rotor of given size at a given wind speed and R.P.M. 2. The performance of the Savonius rotor may be put in the perspective of Gther wind power machines by plotting it on Fig. 74 of Ref. 3 (See Fig. 6 of this report). It can be seen that the Savonius rotor develops power coefficients which are much lower than those of the various types of propeller wind mill. Acknowledgment Mr. A. Bodek was responsible for the desi'gn, manufacture and erection of the rotor and pump. REFERENCES 1. "The wing rotor in theory and practice",.S.J. Savonius, Finland, 1928. 2. "The Savonius rotor", Appendix C of Report on Wind Turbine Project, New York College of Engineering, 1944. 3. "The gsneration of electricity by wind power", E.W. Gelding, E. and F.N. Spon Limited, 1955. 4. "Water pumping by windmills", L. Vadot (Ingenieur-Conseil aux Etablissements Neyrpic), La Houille Blanche, No. 4, Septembre 1957. -4- APPENDIX -- 1. General Having establtshed the shaft output of the Savonius rotor, it remains to see whether there are any applications for it as a wind power unit. Rotors, widely different in size and construction from the example tested, have to be considered. On one hand, a large diameter slow running machine constructed out of natural fibres might prove a practical proposition for pumping ir. certain parts of the world. On the other hand it is conceivable that tall, s&mall diameter, high speed rotors might prove suitable for the generation of small quantitites of electricity. Area for area the Savonius rotor is about half as efficient as a multi-bladed fan mill. However, in many places there is no shortage of wind, and hence no binding need to use Lt 'efficiently'. Also, the simplicity of vertical axis wind rotors militates in their favour, if this simplicity can be matched by simple equipment to which to harness them. In this appendix, application to pumping only is considered. Any proposed pumping system which uses a Savonius rotor must be justified by very careful comparison with the fan mill equipment which is commercially available, and which over the years has reached a high state of development. The Savonius rotor would only be applicable in areas where, through lack of manufacturing facilities or shortage of raw mterials, it could be fabricated more cheaply than alternative equipment. 2. Design Cr;,teria for Windmill Pumping Systems (a) For completely fulfilling a specific application in a given wind regime, the system of windmill, pump and tank should have the lowest cost. The specific: job to be done must be clearly stated, and the wind regime defined as accurate:-- as possible. However considerable engineering safety factors need to be app?Fed in the planning stage to allow for the variability of nature. It must be remembered that the lives of men and beasts depend on sufficient reliability being built into the system. Design tends towards lightly loaded mills which pump sufficient water on the days of light airs. It is not of much consequence if such mills do not perform too eff,ciently on the days of strong winds. (b) The pumping system should be easy to erect and maintain. (c) The windEil1 should be able to look after itself safely in storms. 3. Practical Details In attempti;lg to engineer Savonius rotors into pumping systems it is worth noting some of the significant features of fan mill design. The wind wheels of these machines are multi-bladed, to achieve good starting torque and good power -5- coefficients at relatively low shaft speeds. Owing to the limitations on suction head, in most applications the pump has to operate well below the mill head, and a slowly reciprocating pump rod is the most practical drive. At the mill head, the rotary shaft motion is converted to reciprocating motion: with small mills a reduction gear is used. The design of the mill head is refined by the use of machined gears and oil bath lubrication, so that the mechanical efficiency is high and the equipment requires minimal maintenance. The mill is mounted on a tower of sufficient height that it operates in a clear wind, (The minimum tower height recommended for a 6' diameter ~411 is 25'). In conditions of very strong wind, the mill protects itself by turning out of wind. A li,mited amount of experience has been gained in pumping water with the Savonius rotor tested. Diaphragm pumps have been used, mainly because they could readily be made (no close tolerance engineering) of a size to match the rotor, wi,hout the use of a reduction gear. Some preliminary work was done using a rather small pump and a swash plate drive. Serious vibration problems were en- countered, and the system pumped very inefficiently. A second and larger pump was built with a crank drive, and this is illustrated in Fig, 7. This arrangement runs very smoothly. After initial fabrication, the plunger and stroke length were adjusted so that the pump was well matched to the rotor under the conditions of head and wind speed of the tests. In Fig. 8, pump flow is plotted against wind speed at a static head of 9.7 ft. The dynamic head losses in the piping system were calculated. The variation of total head and pump efficiency with flow rate have Leen estimated and are plotted in Fig. 9. The pump efficiency falls off rather sharply with increasing flow. This efficiency could probably be improved by reftned design of the valves and passages. However it is not proposed trr camy out further pump development without specific applications in mind. Note in Fig, 7 that the centre of the rotor is a mere 7 ft. above the ground, and that the pump is very close to the rotor. A pump can only draw water from a level less than 15 ft. below it. For a rotor of this output, it seems doubtful whether it would be worth fitting a bevel reduction gear and cross head to operate a remote pump. Performance Calculations Consider a Savonius rotor harnessed to a pump which is pumping against a total head of H ft. Assume a fixed displacamant pump which pumps W lb. for each revolution of the rot..~. -6- Thus, mass flow = N.W. Ib/min. and water power - N.W.11. ft.lb./min. If the efficiency of the pump and its drive is 7 then water power = TX shaft power or N.W.H. = 92TT T.N. . I W =++A- Putting in some numbers, if the rotor is the same as the one tested: S = 18.1 sq, ft. d = 3.1 ft. . . . N = 6.15 V P Assume that both the total head and the pump efficiency are constant: H = 30 ft, 7 = 60% The pump size is chosen so that the rotor operates at maximum power coefficient in a 10 m.p.h. wind. . . . At V= 10 mph P = 0.7 for maximum power coefficient N= 63 R.P.M. c, = 0.11 ;c= 1.5 lb. ft. Thus the pump size is calculated w =7+= Oe6 x 2rx 1.5 = 0,19 lb. 30 Sfnce W ,7 and H have all been assumed to be constant, the rotor operates at a constant shaft torque of 1.5 lb. ft. at all wind speeds above the starting wind speed. To determine the pump flow at a given wind speed, the torque coeffi- cient corresponding to the wind speed and 1.5 lb. ft. of shaft torque is calculated; thence the corresponding value of tipspeed ratio is read off the torque character- istic. Pump flow = 36.9p.V.W. Imp. gall. per hour. The variation of pump f-low with wind speed has been calculated and the results have been plotted in Fig. 10. Note that similar curves for different pump sizes could easily be plotted. The overal,l output of the pumping system over a definite period of gLven wind distribution can be determined by integration, for various pump sizes: a pump size can then be chosen to optimise this output. However, this is probably a piece of academic lily guilding which is not worth carrying too far, owing to the crudeness of the assumptions and the difficulties of specifjring -7- wind distributions. Also plotted in Fig. 10 is the pump flow which can be expected from a fan mill of the same area, pumping against the same head.. Again, pump size has been chosen so thar the mill operates at best power coefficient. at a wind speed of 10 mph. Under the assumed conditions the two machines would start at substantially the same wind speed. The output of the fan mill is about 1.8 times that of the Savonius as all wind speeds significantly above the starting wind speed. To take the comparison between Savonius rotor and fan mill a little further: the Southern Cross Company of Australia is representative of manufacturers of good fan mill equipment. The smallest mill in their range is the 6' diameter LZ mill. It is approximately estimated that this mill could do the work of three Savonius rotors of the size tested. The price of this mill (including wind wheel, engine and 25' tower) was US $95, f.o.b., Brisbane, May 1960. The cost of the 18 sq. ft. Savonius rotor built entirely from locally available material (Barbados) by unskilled labor was US $23, It must be noted that the two quoted prices do not include the cost of the pump. However, it can be assumed that the pumps would be of similar cost for both the fan mill and the Savonius rotor. The foregoing performance work is about as far as one can go in general terms. The next step is to see whether a technical and econumic case can be made for using Savonius rotors in specific locations. If such a case can be made, then engineering development of the most suitable configuration of rotor and pump can be carried out. - 8- TABLE OF RESULTS I. _ Brdce P.M. Friction Manometer Differ- Wind TiP echo) R.H.S. L.H.S. Force ence STeed R. P.M. Speed ,en. > wt. Spring Spring lb. Zero Reading cm Hz0 m.p.h. Ratio )L - 180 7 13.7 9.0 11.7 4.664' 4.25 0.42 18.7 190 1.13 1 .30 4.308 0.36 17.3 137 0.88 l?i 4.35 0.32 16.3 131 0.89 130 4.38 0.29 15.5 x05 0.75 1 .!i 4.27 0.40 18.3 142 0.86 1 c.0 4.21 0.46 19.6 160 0.91 70 4.53 0.14 10.8 73 0.75 lbi 4.20 0.47 19.8 154 0.815 . ')!' 9 -\I 4.34 0.33 17.4 126 0.80 1 !!(! 4.39 0.28 15.3 105 0.76 Torque Coefficient Cb .069 .080 .090 .lOO .O73 .063 .2G7 .062 .088 .103 50 4.666 4.41 G.26 14.7 84 0.63 ,111 i? 7 18.5 IL.0 14.5 4.26 0.41 18.5 142 0.85 .087 f ‘, 4.38 0.29 15.5 89 0.64 .124 L * 4.42 0.25 14.4 46 0.35 .143 (. ; 4.44 0.23 13.8 67 0.54 .156 i-i?; 11.5 4.0 7.5 4.42 0.25 14.4 110 0.85 .074 110 4.45 0.22 15.5 115 0.94 .084 i 2.i 4.41 3.26 14.7 131 0.99 .071 95 4.47 0.20 12.9 100 0.86 .093 '1 i i 4.40 0.27 15.0 120 0.89 .069 110 4.43 0.24 14.2 115 0.90 .077 ILO 4.34 0.33 16.6 149 1.00 ,056 '180 0 0 0 0 4.39 0.28 15.3 190 1.38 0 1.20 4.43 0.24 14.1 180 1.42 'L 50 4.50 0.17 11.9 159 1.48 Lb0 4.50 0.17 11.9 148 l.38 190 4.67 4637 0.30 15.8 200 1.41 x5 0 20 9.5 10.5 4.704 4.47 0.23 13.9 88 0.70 -113 105 4.4i 0.29 15.5 110 0.79 .089 75 4.49 0.21 13.2 78 0.66 .123 i70 4.53 0.17 12.0 62 0.52 .152 ho 4.54 0.16 11.6 62 0.59 .162 14.50 3.5 7 19.0 12.2 0.20 13.0 36 0.31 -183 60 4.47 0.28 15.3 62 0.45 ,131 45 4.46 0.24 14.1 46 0.36 -152 80 4.35 0.35 17.1 84 0.54 ilO5 120 4.36 0.34 16.8 127 0.84 .108 110 0 9 4.0 5.0 4.50 0.20 13.0 116 0.99 -062 120 4.46 0.24 14.1 127 1.00 .051 100 4.55 0.15 11.2 105 1.04 .082 1L5 4.40 0.30 15.8 153 1.07 -041 120 4.48 0.22 13.5 127 1.05 ,036 1 30 4.70. 4.49 0.21 13.1 138 1.18 -059 -9- Fig. 1. The ?S eq. ft. Savoniue Rotor Fig. 3. The Band Brake - 10 - I Projected Area = 37.3 x 70 = 2611 sq. in. = 18.1 sq. ft. Distance Between End Plates - 70 in. Shaft Dia. - 2.0 in. Circumferential Speed per 1 Rev. = 9.76 ft. per min. Fig. 2. DIMENSIONS OF THE 18.1 SQ. FT. SAVONIUS ROTOR. - 11 - I 1 I 1 I ! I I I I I I I 0.2 0.4 0.6 0.8 1.0 1.2 Tip Speed Ratio - p Fig. 4. TORQUE CHARACTERISTIC OF SAVONIUS ROTOR.