You are on page 1of 7

AUV Propellers

:
Optimal Design and Improving Existing Propellers
for Greater Efficiency
K. P. D'Epagnier
Massachusetts Institute of Technology / Woods Hole Oceanographic Institution Joint Program
Department of Applied Ocean Physics and Engineering
Woods Hole, Massachusetts 02543 USA
Email: kathrynd@mit.edu, Telephone: (508) 289-3269

Abstract-Propeller design is a significant process that is suffice [4]. Therefore, many AUV designs incorporate a
overlooked in many design situations. Using propeller vortex commercially available model airplane propeller.
lifting line code and two-dimensional analysis with Brockett In a situation in which fabricating and testing propellers is
diagrams, a propeller design can be optimized for an
Autonomous Underwater Vehicle with a given design speed by prohibitively expensive, or it is desirable to use existing
varying blade number, propeller diameter, hub position, and propellers, modifications can be made to improve performance
chordlength distribution over a given range of RPS appropriate in propeller efficiency. This analysis seeks to provide both an
to the AUV's thrusters 111. Initial design decisions from PVL ideal design for specific AUV operations, as well as methods
code chose a three-bladed 0.625m diameter propeller at 2RPS,
with a hub located at r/R(0.2) and an efficiency of 0.7091. Final for improving performance using off-the-shelf model airplane
propeller design modified the chord distribution to improve propellers.
distributed lift coefricients and to ensure that shallow-water II. DESIGN METHODS
operations would not experience cavitation effects. The final
design geometry used the same values of the propeller blade A. Propeller Modeling
number, diameter, RPS and hub location as the initial design, but Today, computers allow a detailed optimization of
at an efficiency of 0.7642.
propellers. Propeller design is used to model the hub
While off-the-shelf propellers, such as those designed for geometry with an image vortex. Hub and tip loading effects
model airplanes, can have high calculated efficiencies relative to can be considered in rigorous analysis. After optimizing an
actuator disk theory, the lift coefficient at each radial position
and the resulting stalling effects must be considered. initial design, the propeller performance undergoes two-
Comparison of an existing design with an optimized design dimensional analysis, including determining minimum
improves existing propellers and contributes to the process of pressure coefficients and local lift coefficients. Rigorous
choosing future off-the-shelf propellers. design methods use multiple iterations of many parameters to
calculate a highly efficient propeller design.
I. INTRODUCTION There are many methods for propeller design [1], but in this
Intuition tells us that model airplane propellers are not the study, propeller vortex lifting line code will be used along
ideal solution for an AUV.' In the case of an airplane, the with two-dimensional analysis using the Brockett diagram [5]
fluid density of air is much less, -1/1000 of that of water. to meet the AUV specifications. These two design methods
Also, the typical RPM value intended for a model airplane's will lead to design of an optimal propeller, as well as calculate
propeller is between 6,000 and 10,000 RPM, which is the efficiency of existing propellers, which will determine
significantly greater than that of the AUV used in this study performance flaws. The resulting analysis will provide
[2]. recommendations for altering the existing parameters to
The reason for lack of optimization lies mainly in the cost. improve overall propeller efficiency, considering the effects of
In designing a vehicle, off-the-shelf components are very minimum pressure coefficients and ranges of lift coefficient.
desirable. It would be much more costly to design and B. AUVSpecifications
fabricate a propeller design than it would be to order one that There are several aspects which must be considered in
has already been tested and can easily be replaced. The AUVs propeller design. The propeller design for the AUV must
considered in this paper often do not require high speeds meet the following criteria as in [4,6]:
compared to REMUS AUVs [3], and off-the-shelf propellers Design speed, Vs, is Im/s.
Inflow wake velocity variation, war =/-0.03m/s.
Design depth is 200m to lkm.
'The AUVs modeled in this study are the SeaBED prototype Cavitation effects will be tested for a depth of d = 2m.
and its successors, JAGUAR and PUMA, all of which are Propeller type is assumed free tip double screw.
intended for low-speed operations.

1-4244-01 15-1/06/$20.00 ©2006 IEEE

It is interesting to note that the Va to Vs is one for this particular AUV.. PVL Code. 1.. Rotational .. .. 2 The axial flow of the design decisions.8..594m.g5 where r is the location on the radius and R is the radius length were measured using calipers. The chordlength distributions.. considered in analysis: ommmom ommmom 0. MATLAB code. V*. 2Data from the experimental Huang Body 1 relating axial flow to radial position was splined such that it returns values for axial flow for any range and any spacing of values for radial position. Chordlength Optimization include a case of a ratio of up to 1. c/D with respect to r/R. The blade chosen that best satisfies considers several important factors.... -~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . (2) propeller provides the most efficiency at the design speed of 1..5 C Vortex Lifting Line Codes Propeller vortex lifting line codes provide a powerful o045 2 25 3 method for assessing the characteristics of a propeller. The data was extended to D. was used in combination with Figure 1.. and not an influential part fluid flow entering the propeller [9]. In this Efficiency analysis. diameter is not equal to one. were within an acceptable range of 0. is 0..0 104 in terms of r/R. and Efficiency.... a goal model a value of zero or less at the hub. H.. 0. .plt in PVL. 58 . L LC 0. Otherwise. where r/R = 0.. Va/V S range of high efficiency (0..... CL =V (4) L V*SC range from 1 to 3 RPS....... The total velocity of the lifting line. CT = 1 PV2 (iR2) ... (1) number of blades with a smaller diameter provide greater T efficiency for the lowest range of N. .5 at the hub.... [8]..9.7 The following are characteristics of the existing propeller . Three to six blades were assessed.. Furthermore...exe was used with an input of variables. The diameter varied from 0... as in [1]. F= 27zGVsR.. .. but only the existing propeller.Speed... and revolutions per at each radial location.. D.. the Further optimization of the chosen propeller shape is practical assumption was made that axial flow is constant for achieved through consideration of the result of chordlength on all r/R values less than 0. the three-bladed propeller has a wide Axial flow ratio.. lift coefficient is around 0..6E c The propeller has three blades. J = Vs /ND.. as well as the efficiency of the contour plot. the spline would pressure distribution and the local lift coefficient. The required forward thrust at operational cruising speed for the AUV is 150N.. with an case that the ratio of the propeller blade diameter to the hull initial calculated efficiency of 0. IN(RP. and is 0. z. . the ratio of desirable values for N. since the values of the First... it will be taken as an outlier. . 2..7091.. a maximum efficiency on the 3-blade propeller is for smaller function was created to determine the effects of axial flow for values of N. (3) advance coefficient. and calculating tr less ut* given from APLOT. rotational speed than the others. c The diameter. . Since this is the lowest of the considered values alternative propeller placement or diameter of a body affecting for N.... . This is because a larger The advance coefficient. . is 0. results for a three-bladed propeller are shown. using a the given design criteria. or 75N per propeller [7]. The Initial propeller geometry was chosen by varying lift coefficient can be determined by dimensionalizing gamma parameters of the blade number. Nonetheless..3 to 1. C Tests taken from thrusters of a previous AUV show that it is . Here. or z = 3.5 to 1...625m in diameter.. The hub is 0.. whereas the three-bladed The desired thrust coefficient.2. The first factor is in the all the desirable values of N. diameter. using the following two equations: second.... The geometry is as shown in Fig. Diameter...0m/s. PVL.. Additionally. The initial design exhibits three blades.0m. The blade numbers analyzed range from 3 to 6. D(m)Diame0t r. . since a three-bladed including: propeller is more efficient in all ranges of diameter and The number of blades. Choosing a Blade Based on Rotational Speed. which is not possible. In finding an initial propeller design...75 - . can be found by using B i..03 Im. The analysis considers both the optimal propeller design for The results were assessed as shown in Fig.. Rotational speeds appropriate for the analyzed thruster Then.225. 111111 most acceptable to model the propeller at 2RPS.70 or more) at all ranges of Due to the placement and size of the thrusters.

cavitation is not likely to propeller efficiency by super-cavitating [10].49 0.0543 14.0212 0.0426 0.0330 0.7984 17.0000 4. TABLE I FINAL OPTIMIZED PROPELLER PERFORMANCE VS.540.0331 0.023 0 0 line velocity.48 0.0162 T 0. camber ratio.039 0 0 27rNR(r/R) 0.0748 0.4936 0.0323 0.0239 0.0254 0.4000 1.6047 0. In addition to being less efficient.3000 1.3000 1.8479 0.2269 2.4599 0.0335 0.2391 0.078 0 0 0. V*.2148 2.4448 0.5931 0.0590 0.0380 0.9439 0.9042 0.5685 0.0228 43.1693 1.5000 2.0. USING ML TYPE NACA.2000 1. aspects defining the propeller geometry. fo/c and the inflow variation bucket width.0317 35. used to determine the thickness ration.Pvap JoCHART = 0. However. partial pressure coefficient.5278 51. in which lift and non- Brockett diagram with NACA-66 sections [4] in finding the cavitation requirements are easily fulfilled.6000 1.4867 0.8460 0. Due to the relatively low RPSs needed for through pitting. is shown in Table I. A large aspect of the 2-D design in this analysis utilizes the The data for the final propeller design.079 0 0 0.8538 57.7912 0.0239 118. is calculated thus: P/D tan(Bi + ))i = Aa = tan 1K var 2 tVe ) (7) where i =1.7000 2.3519 0.0660 0.0679.0847 0.3438 0.22 TABLE II BLADE GEOMETRY FOR OPTIMIZED PROPELLER.2023 26.4000 1.6570 20.1275 1.9968 28.0599 0.4235 0.3120 0.4871 0. Au.1545 1.6749 15.8371 0.9991 0.55 0. (9) 1 [)VJ*2 2 .8.1756 0.6878 0.5478 40.7362 0. the thrusters are mounted such that the thruster pressure coefficients at a shallow operating depth of 2m.33 0.2973 17.5000 1.2578 0. Super-cavitation occur [6]. to/c and local pressure including the rake/D and the skew is assumed to be zero.9500 3. as in [1].9500 0.9809 0.0314 48. is In this design.5581 22. present problems will be investigated by analyzing propeller Additionally.5609 0. (5) c . The value of sigma can be found from the following equations: Jo .2207 0.0175 0.070 0 0 0.7000 1.8000 1.2500 1.3256 46.1885 1.8. LCHART = 1.r (6) Cavitation is an important design consideration in shallow- Sigma is used later on in comparison with the minimum water operations.0801 0.041 0 0 NRr = 0.2000 1. Geometry for determining total lifting 1.0000 0.5383 0.0293 27.030 0 0 0.24 1.46 0.9242 19.2500 1. Some high-speed AUVs avoid reduction on this vehicle and the operating depth.0609 0.2788 51.0050 16. where ho = operating depth.4075 0.2683 0.0638 0.9000 3.9000 0.0032 0.0293 67.2029 1.0091 15.8645 15.8000 3. the case in which cavitation could is not an option for the low-speed AUV analyzed in this paper.9462 0.4866 0.1253 0.0005 13.2771 0.0635 0. Cp [1]: pitch to diameter ratio._ C (8) Patm+ pgh .(r / R)()).4060 45.4287 0.9355 0.0146 17.27 0.0235 21. The distribution coefficient.050 0 0 0.055 0 0 0.9896 32. (10) .046 0 0 0.8621 32.9455 0.2193 2. P/D.025 0 0 Figure 2.4754 0.8232 0.5165 0.2864 14.6000 2. AND NACA-66 SECTIONS r/R P/D c/D Au fo/c to/c rake/ Skew 0. from NACA Mean Line a =0. where a > Cp in order for cavitation cavitation creates bubbles which can damage propeller blades not to occur [1].56 0.0277 0.38 0.0572 0.0312 0.0330 0.JoCHART ( CLR h = h .8896 0.0258 97.0340 0. A=0. housing diameter is too small to facilitate super-cavitation.9909 24.4814 38.0693 0. RADIUS POSITION r/R V* B Bi CL G -CPmin 0.7157 0.53 0.

35 --hub=.1435 0.0017 0.9500 3.9756 27.0455 0.2058 0.6000 1.02 0.2319 1.3195 0.2448 46.4634 16.1 0.63 0.74 0.0742 2. 1 1 hub=.1767 59.1720 0.73 Actual Hjb=.4000 1. This will help to make improvements values use optimal calculations for the pitch over diameter.1250 58.2500 1.35 0.0007 15.0921 0.0169 169.594m.9000 3.8 0. 0.1572 0. The design requirements can be summarized into Table II.1476 0.65.0115 1.81 0.5000 1.2697 0.019 0 0 1.0739 1.5 .8000 1.35 hub=.2873 0.1494 0.771: .2000 1.067 0 0 0.55 *t 0.33 hub=.4 0.9210 0.1103 0.0933 0.0301 0.1221 0.0591 0.4 hub=.8000 3.0349 32. to the existing propeller.0143 198. 1 1 0.2170 0.023 0 0 0.6 0.3000 1.3561 0.3000 1.49 1. as well as to provide guidance on .0637 19.6000 2.1137 0. hub=.72 0.6646 15.71 0 0.0888 68.8892 0.9381 0.6877 0.0726 1.9000 1.6255 0.1820 0.3387 38.7000 2. TABLE 111 EXISTING PROPELLER PERFORMANCE VS.8917 32.33 0. CL and Efficiency for varying hub diameters and z = 3blades.7890 0.4 0.6597 0.75.1505 0.037 0 0 0.5026 0.93 0.7862 73. quick comparison. 104 0.7 hub=.46 TABLE IV BLADE GEOMETRY FOR EXISTING PROPELLER. D = 0.0000 3.9926 0. r/R Local Position.4 0.0289 24.12 0.45 Local Position.5828 1.3573 0.3675 0.1563 65.0751 2.0929 0..I Actua Hub=.9215 31.2997 1.75 hub=.35 0.0770 3.0130 0.2000 1. It is important to note that the Table 4 propeller can be made.3741 0. a propeller geometry are shown for the existing propeller design comparison between both off-the-shelf and the designed in Tables 3 and 4.6 0.021 0 0 0.1565 0.8661 14.0425 0.1280 1.1707 0. r/R Figure 3.0533 1.0814 1.032 0 0 0.8699 0.5000 1.0846 2.7000 1.ariation of existing propeller on CL Effect of hub diamneter \ariation of existing propdler on Efficiency 0.028 0 0 0.2771 1. USING ML TYPE NACA.0826 0.0386 0.8.2233 1.2 0. In order to make a from which the optimized propeller could be fabricated.9663 0.0862 0.1551 0.053 0 0 0.5871 0.05 0.15 0.4 0.018 0 0 Effect of hub diameter \.0117 17.09 0.5565 0.4000 1.2877 25.0490 23.4153 0. A=0. RADIUS POSITION r/R V* B Bi CL G a -CnP 0.047 0 0 0.2 0.0342 63. AND NACA-66 SECTIONS r/R P/D c/D Au fo/c to/c rake/D Skew 0.76. T 1 5 0.6989 1.0211 140.0532 0.5825 39.2903 1.0289 93.7506 0.8456 17. the equivalent of Tables 1 and 2 of Now that an optimal propeller has been designed.- 0.0691 0. selecting future off-the-shelf propellers.9533 1.3 0.95 0.2923 51.6704 21. N = 2RPS.9993 15.0000 0.2633 0.061 0 0 0.5251 1.0647 1.25 0.2500 1.0754 0. 104 0.0726 0.6695 0.0185 19.9500 1.0922 0.0365 44.02 0.5663 1.45 0.7904 19.

r/R Loced Position. with the original chord distribution.8 1 0.758 hub=.33 0.65m D=.45 Figure 6.4 0. while '*' indicates CL<0.5 LocEd Position. r/R Figure 4.4 0.81 0.2 > 0.15 0.7 is NOT satisfied.5 criteria determined the chosen propeller design: The lift coefficient would be reduced to 0.756 0 0. 3.5 605 -X D=.5 which is measured as 0.7. CL and Efficiency for varying hub diameters and z -3blades.762 0.7661 Actuad Hub=. wMile satisfying CL<0.6 0. variation of blade hubs for different 3-D Propeller Blade Design diameters.2 0.3 0. appropriate analysis determined the effect of hub diameter variation on the N4 o lift coefficients. Analysis determined that the lift coefficients on the existing propeller were unacceptably high.2 0.7 or below at the 11.2 0.4 0.25 0.1 0. Efficiency vs. as shown in Fig. 1 1 0. 104 hub=. MATLAB representation of optimal modified off- hub(r/R) the shelf propeller.77m 0.35 hub=.3 0.05 0.8 0. MATLAB representation of final propeller design.78 D 0. These values are significantly different from the actual P/D.5 0. the effect of different blade diameters on lift 0.5 605 hub.35 0.25 0.15 .427 for r/R equal to 0. N= 2RPS.79 hl 0- 0 i. Due to high values of lift coefficients for the model-airplane U05~ propeller. Effect of hub diam-eter \aridion af existing propeller onClO Effect of hub diameter \redation of existing propeller on Efficiency 0.73 = 0.75 O D=. 'o' indicates CL<0.3 0.4 0. Figure 5. while still considering the effect of hub variation on propeller efficiency.35 hub=.76m 11.4 0.76 Actuad Hub=. .1 hub=. 104 hub=.70 3-D Propeller Blade Design 0. Therefore.76m Effect of tlade and hub ciarmeter wiriation on Efficiency.33 hub=.5 coefficient and calculated efficiency were modeled.74 D=.82 0. 0.594m 0. The efficiency would be maximized for z = 3 and N = 2. and below 0.31 0.25 at the blade tips.83.0. Figure 7.764 hub=.77 -05< 0. 0.7.1 0.11 0.5 0.76 115 O actuad D=. A list of .75m 0. D = 0.

Camilli for providing 0. and did not with a hub extending to r/R = 0.2). calculated efficiency. analysis showed teaching the propeller design methods used in this analysis. analysis. computing the efficiency for the original propeller design. for an investigations showed that the lift coefficient. Stejskal. and to overall AUV performance. efficient for low-speed operations (although they have a limit). Kerwin for use of his PVL code. the propeller and the hub diameters. and should not be overlooked. B. greater than could be measured using the methods in this [2] G.7. rewarding performance improvements as a result of increasing Thanks also to Professor J." OCEANS 2000 MTS/IEEE Conference and Exhibition. 2000.5. Hydrofoils and Propellers. If a value higher than the adjustment of many parameters. the hub size of the propeller could be blades. The lift coefficients at a Thanks are due to C. from an efficiency of 0. A design depth of 2m resulted in no cavitation effects. and therefore 0. 5. and increased RPS resulted in the need for smaller propellers. it was found that larger diameter propellers were more significant contributions. distribution. is the interaction among the likely to cause stalling. In order to complete a rigorous AUV's design speed. Cambridge. 147-151. the propeller diameter should The design had a modified chord distribution which resulted in be increased to improve efficiency and reduce the effects of a lift coefficient ranging from less than 0.31. Kerwin. but is essential were considerably higher than that of the design propeller. and an efficiency of 0. the MATLAB code and other 0. Under the assumption that the existing propeller is sold off. decreased. efficiency losses due to high CL values should be However. modification and the final propeller design is shown in Fig. due to minimal information on AUV thruster the case of this particular propeller. it can be eliminated as operation decreases. Sclavounos and R. First. vol. at 2RPS. some does not take into consideration stalling effects of the propeller approximations were made on the range of practical RPS due to CL values greater than one from the hub to r/R = 0. . In this design. but this would be counterproductive due to the fact September. Forrester. The existing propeller required attention specifically to the Further investigation into optimal rotational speed will lead lift coefficient and the value of P/D. The PVL lifting line code to better designs of AUV propellers. off-the-shelf geometry were held to the range of CL values less A visual comparison of the off-the-shelf propeller than 0. propeller. C.7810 to 0.5 to less than 0. E. Under the assumption that the same off-the-shelf The initial propeller design had an efficiency of 0.9287. the final design had an efficiency of 0. the requirements for its range Using the results shown in Fig.76m. Purcell. Therefore. and a diameter of increased. so will the RPS of the propeller.31 were information on previous tests ofAUV propellers.9. MIT CopyTech. 200 1. Additionally. The actuator disk efficiency Optimal propeller design emerges from the interaction and serves as a check against efficiency. July 20.625m. within a satisfactory range. 1. Furthermore. As an AUV's speed of actuator disk efficiency is found. could be improved from both of these distribution. Austin. when detailed AUV information is available. the efficiency value calculated for the existing methods used for this propeller design provide a useful tool for propeller was erroneously dependent on an optimal circulation further applications. while satisfying the design requirements. In erroneous.. Therefore. and propeller geometry exists for greater propeller diameters. which decreases the lift coefficient. If the and an actuator disk efficiency [1] of 0. it is important to recognize that a design III. the made. "Aircraft International Carbon Propellers. The P/D ratio can be increased in future blades by ACKNOWLEDGMENT increasing the chordlength or decreasing the propeller diameter. but the same principles for the geometric design performance was far lower than the actual value taken at r/R = remain the same. the lift coefficient values performance. Roman and R. 6 Calculations show that many factors contribute to propeller and 7. the design chosen was 0. pp. as well as the added effort and cost of modifying the The effect of variable hub diameter for the chosen blade propeller. The optimal diameter would be as close as possible to the that the modeled efficiency decreased as propeller diameter original diameter. T. "New Consideration was made in cutting down the tips of the capabilities of the REMUS autonomous underwater vehicle. The optimized design blade did have a lower diameter of 0." personal analysis.8154 account for loss of efficiency through stalling effects. A factor that was not considered. The efficiency improvement was actually [1] J. Kimball for the-shelf for a scaled range of diameters. The coefficient of lift and pitch over drag make general. the propeller diameter would get prohibitively large. et al. which would modify the existing propellers in order to provide The final optimal propeller was designed with three greater efficiency.76m on CL and efficiency are shown in Figure 4.44 percent improvement. [3] M.8154.7642. efficiency. There are several methods that merit further investigation. prevents stalling. however. 2006. of acceptable CL values were more stringent. Therefore.76m blade diameter and a hub at r/R equal to 0. In values. requirements for the propeller design will become more The pitch over diameter value calculated for propeller complex.51 percent from optimizing the chord performance. modification methods. a hub starting at r/R(0. Allen. CONCLUSIONS AND RECOMMENDATIONS with PVL results of high efficiency does not consider stalling effects of high lift coefficients. thruster speed. von Alt. and thus overall improvement of 5. since the effect of stalling was not accounted for in communication.24. and power consumption. Thanks to Professors P. N. stalling.7091. the efficiency REFERENCES showed a 3.

et al. April 19. Marine Hydrodynamics. [6] C. vol. "Imaging Coral I: Imaging Coral Habitats with the SeaBED AUV. 206-208. "Minimum Pressure Envelopes for Modified NACA-66 Sections with NACA a-0.[4] H. Roman. Armstrong." DTMB Report 1780. [10] J. N. Roman. 5. Camilli. 2006." he May 6.. . Gilbes." Subsurface Sensing Technologies and Applications.8 Camber and BuShips Type I and Type II Sections. [9] R. "Cosmos FloWorks AUV Modeling. "Huang Body-1 MTWAKE screen matching experimental wake MTSOL grid finer near body." personal communication. Brockett. pp. Singh. "PVL Code. 2006. F. 2006. [8] J. April 3. R. Newman. "SeaBED MATLAB Code. Kimball. January 2004. C. MA: The MIT Press. R." personal communication. 1977. 1. [5] T. Kerwin. April 21. Cambridge. no. 2006." personal communication. [7] R. February 1966. Eustice.