17 Replies Latest reply on Nov 18, 2014 2:20 PM by Txon Aguirre

    Problem With Propeller Flow Simulation

    Robert McGonigal



      I am trying to perform a flow simuilation of of a propeller with an incoming velocity of 51.33 ft/s and angular velocity of 2575 rpm. I created a cyclinder around the propeller to use as the rotating region and made sure that the component control includes only the propeller and not the rotating region.


      However, when I run the analysis, my velocity profiles are wrong. I am getting little to no velocity on the back side of the propeller (the side where thrust should be generated)


      I have run a seperate tutorial with the same model used in the tutorial with success, I just cannot get it to work with my propeller.


      I have tried changing the angular velocity, axis of rotation, turbulence intensity, number of blades (2), removing the incoming velocity, all with no luck.


      Here is a link to the tutorial which I followed for the propeller used in the tutorial with success.




      Also, attached is the solidworks file of my propeller design.



      The propeller model





      Maps of Velocities





        • Re: Problem With Propeller Flow Simulation
          Jared Conway

          setup looks fine

          when the simulation completes, what is the reason for completion? are the goals converged?

          have you tried improving the mesh at all? recommendations are in the solving eng problems doc

          is it possible that your prop isn't as good as you hope it is?

          • Re: Problem With Propeller Flow Simulation
            Txon Aguirre

            Hi Robert, and hi Jared:

            This is my first post here, ...


            Let's see, May we know the dimensions of your propeller ? Diameter, Pitch, Area ratio ..... Best thing is performing the preliminar calculations by hand, and then setting the problem in Solidworks.

            May be, as Jared said, that your propeller does not fit the working conditions.



              • Re: Problem With Propeller Flow Simulation
                Txon Aguirre

                Hi Robert:

                I've downloaded your file and had a look at it:

                1-] I see you've set three configurations. The correct rotation for the blade profile is +269.65 rads/s.

                2-] For better results try using a domain which height and width are 2.5 times the prop diameter. Leght upwind around 2 diameters, and lenght downwind around 4 diameters.

                3-] The blade profile of your propeller is quite strange and out of what is found in the common practice. That circular leading edge leads to the early flow separation (around at 11% of the chord lenght) developing highly turbulent areas around the blades that do not allow for anykind of tidy flow.

                I include a pic were the Pressure Coefficient graph is drawn (light red area) superimposed to the flow lines and the profile.

                Check the very stepped shape of the graph. For a profile to properly work, pressure changes must be as smooth as possible allowing for the fluid, to smoothly flow, from the leading to the trailing edge.


                From your data(SI units):

                Air speed 15.64 m/s, RPM= 2575= 269.65 rad/s, Chord lenght: 0.2286 m; 70%radius= 0.625m

                Resulting speed on the blade at 70%radius results to be 169.48 m/s, quite high for this shape, with an angle of attack of 12º.

                4-] Blade tips are too flat. Best projected planform of the blades is an ellipse. At least try giving that shape to the tips. The "donut like" turbulence that can be seen in the simulation is mostly due to those abrupt ends of the blades.


                Wrong profile selection, together with the flat tips, leads to fluid recirculation. Its not allowed to flow out of the propeller disk.

                I've also observed a strange blade root design.


                All eddies and turbulence, apprt from diverting the flow from the desired path, mean energy leaks that is not used for propulsion and hence, decreases the system performance.

                CCP= Pressure center

                Cp= Pressure coefficient.




                  • Re: Problem With Propeller Flow Simulation
                    Jared Conway

                    looks like a great synopsis from the engineering side of things. did you find anything from the rotating region setup that might be out of whack?


                    might be worth looking at some examples with known results to make sure the OP is comfortable with the software and then look at this example deeper like you did

                      • Re: Problem With Propeller Flow Simulation
                        Txon Aguirre

                        Hi Jared:


                        With regard to your suggestion, I've worked a standard ship propeller model for Solidworks in order to be able for making some comparisons:


                        PROPELLER specs (International Units):

                        MODEL: Wageningen-B


                        Diameter: 5m

                        Pitch:  5m

                        Rake: Standard 15º

                        Blade area ratio: 0.4

                        Blades number: 4



                        Fresh water: dens= 997.97 Kg/cu.m.

                        Propeller shaft working depth: 6m

                        Fluid speed: 8 m/s

                        Wake factor: 0.3

                        RPM: 115


                        From Kt, Kq, J standard curves. No Reynolds number correction.

                        Hand results.

                        J= 0.835

                        Kt= 0.127

                        Kq= 0.024

                        efficiency: 0.7

                        Thrust: 290.094 kN

                        Torque: 275.310 kNm

                        Cavitation: No cavitating conditions.


                        From Solidworks Flow Simulation.


                        Standard Wageningen B4. Computer generated model from Wageningen series specs.

                        As for good practice, a hydrodynamic bulb has been fitted ahead of the propeller for fluid flow fairing (open water tests).

                        Bulb specs:

                        Lenght= 16.8m

                        Breath= Lenght/7

                        Longitudinal Section: Bi-elliptical.

                        Bow ellipse: a=1.2m: b= 6.22m

                        Stern ellipse: a=1.2m: b= -10.58m


                        Propeller specs.

                        Same specs than above.

                        Building material: Aluminum brass.



                        Analisys: External

                        Reference axis Z

                        Mesh level=5

                        Rotating mesh: 115 RPM (Left). Diameter 6.55m

                        Rotation axis: Z



                        X:= [7.5,-7.5]


                        Z:=[-22.7, 37.7]


                        Fluid speed X,Y,Z := (0, 0, -8)

                        Ambient Pressure: 6m depth equivalent.



                        Surface goal: (whole four blades)

                        Force Z, Torque Z


                        RESULTS for J= 0.83

                                                      VALUE          Averaged          Minimum          Maximum              HAND

                        Force Z       [kN]      208.151          211.285            208.151            218.810               290.094

                        Torque Z     [kNm]    220.501          222.406            220.501            228.043               275.310

                        Kt                           0.0908             0.0922              0.0908              0.0955               0.127

                        Kq                          0.0192              0.0194             0.0192              0.0199                0.024

                        Efficiency:                 0.62                 0.63                0.627                0.637                  0.7



                        As it can be seen, Solidworks predictions are a lot below hand calculations for this propeller.

                        May be some changes in the Flow Simulation set up, could return better results. It would be fine making some more tests with different propellers. Perhaps, results could be more accurate. Anyway, we'd get a better idea of the Simulator behaviour.


                        Im open for any sugestion, and willing for anybody to repeat my calculations. May be Im the one who is wrong, and not the CFD. :-)


                        Kind regards.

                          • Re: Problem With Propeller Flow Simulation
                            Jared Conway

                            Only way to know is a physical test. But I tend to think flow has less assumptions than a hand calc and therefore should be more accurate.

                              • Re: Problem With Propeller Flow Simulation
                                Txon Aguirre

                                Dear Jared:

                                The Kt,Kq,J curves are directly obtained from physical propellers into channel tests and regularly used in the Naval building industry. Anyway, I think I've figured out where the results  discrepancy may come from.

                                May be a difference in the measurement of the blade sections relative angles of attack, resulting in slightly different physical pitch angles. Let me take a few days for checking.


                                Appart from what has been said above, there are more error sources than any problem with a set of curves/tables that are certificated and daily used in the REAL industry. Im a lot more for 3D modeling errors than for any question regarding the calculations.


                                As the 3D model I've used was built by a computer software, may be it does not fit the real prop with the required accuracy. I'll check the model, section by section, and correct it by hand if needed, for repeating the comparison again.


                                Kind regards

                                  • Re: Problem With Propeller Flow Simulation
                                    Jared Conway

                                    I agree if your comments rrelative to sources if error. It is very difficult to exactly match physical in a simulation.


                                    regarding where you got the results, sounds good, I was basing my comment directly on this comment you made


                                    "As it can be seen, Solidworks predictions are a lot below hand calculations for this propeller."

                                      • Re: Problem With Propeller Flow Simulation
                                        Txon Aguirre

                                        Hi all:

                                        Im under way with the model check. As far as I've got by now, Im a lot more for anykind of error in the 3D model than in the calculations.

                                        By only increasing the 3D propeller model blades pitch angle by 0.25º at each section, results fit reasonably well.


                                        I've performed a set of parametric studies(Using the same Wageningen B4 propeller), using Flow Simulation and standard KtKq curves, and have realized that the discrepancy is linear with RPM.

                                        There is an aproximative relation, used for "First approach procedures":


                                        Thrust= dens PI (0.7R)² AeAo (N²L² - V²(1-w)²) ---> Thrust = Flow Momentum difference


                                        Where: dens=water density; R=prop radius; AeAo= Blades area ratio; N= revs/second; L= Nominal pitch(lenght units); V= free flow speed; w= wake factor


                                        As N and L are raised to two, the relation can be taken as linear. Hence, the observed discrepancy could be understood as a diference in pitch.


                                        After some more accurate calculations, the 0.25º difference is the result. Seems to be a very, very tiny difference between the 3D model and the real propeller.



                                        Kind regards

                                          • Re: Problem With Propeller Flow Simulation
                                            Jared Conway

                                            SSo your conclusion is that the model was slightly off and by adjusting the model the results from flow match physical.

                                              • Re: Problem With Propeller Flow Simulation
                                                Txon Aguirre

                                                Yes Jared:

                                                I've checked the same propeller model (Wageningen B4-5050L)  Standard Wageningen propeller, 4 blades, 5m diameter, 5m pitch, Blade area Ratio 0.4, turn-Left.


                                                If anybody wants to repeat my tests --->No problem for sharing the model.....

                                                Wageningen B series. D=5; P/D=1; AeAo=0.4; Z=4

                                                A fairing body has been fitted ahead of the propeller in order to allow for "open water" condition tests.

                                                Vfree stream= 8m/s along Z; Ambient pressure: 8m depth.(101325 + 8 x 997.97 x 9.81)

                                                Parametric study: RPM from 110 to 210, 10 steps.

                                                Domain size: X[-12,12]; Y[-12,12]; Z[-37,27]

                                                Initial mesh level 5: Refinement level1 tabulated.


                                                The 3D model has been built from three different sources; by hand, by a commertial software and by my company propietary software.


                                                I've found slight differences between those models, mostly at the blades tips and leading edges.

                                                The way that "commertial software" exports to Solidworks, models goos trailing edges, but does very bad with the blade tips and last 10% of the blade around them.

                                                Our own software, models accurate blades but the leading edges refinement, once imported into Solidworks must be done by hand.


                                                Modeling by hand, from scratch, gives the better shape control, but work is too slow and must be checked many times along the development in order to be sure about how is it going on.

                                                Each model, into the same conditions returns different results from Flow Simulation, all of them around the KtKq curves hand calculations, and all of them with differences running into similar ranges.


                                                As conclusion:

                                                From a general view, all those procedures give accurate 3D models, but slight differences in critical areas, as blade tips and leading edges, result in Flow Simulation calculation disagreements.



                                                More tests, with different propellers would be required in order to get a good idea.



                                                Kind regards

                                                • Re: Problem With Propeller Flow Simulation
                                                  Txon Aguirre

                                                  Dear Robert and Jared:

                                                  How to attach a model for sharing ?