1 Reply Latest reply on Jan 15, 2017 7:45 AM by Dave Bond

    Marin Propeller simulation - Thrust results don't match ..

    Shar Fa

      Hi guys!

       

      I'm working on validating through flow simulation a known marine propeller with the following data:

      600 RPM

      V_water=1.4 m/s

      Thrust is suppose to be around 23 [N] (calculated), with effic. > 0.7 for the very least.

       

      Creating the model on solidworks i had the control of how to place the blades (their pitch) against the hub.

      It was set as 50.5 deg at r/R=0.2 manually, and the simulation results show a symmetric velocity and pressure profile around the blades with this setting (as opposed to when I tried setting at different angle which either shown no symmetry or no high-low pressure profile above and behind the blade section view).

       

      Problem is the results I get for the thrust and torque (too low as well) do not make sense.

      Thrust is too low (around 1-3 [N] after full convergence) with a torque of approx. 1 [Nm] which gives a VERY bad efficiency rate (does not make sense).

      Removing/Changing hub geometry has negligible effects (tried no hub, cylinder hub and the attached 'improved' hub adding 1 [N] compared to a goal of 23 [N])

      Hub2_AntiCloclwiseBackFlow.png

      Any guidance would be much appreciated

      Thanks!

        • Re: Marin Propeller simulation - Thrust results don't match ..
          Dave Bond

          Looking at the date you posted this query you probably have the boat built and are happily sailing down the river.

          One thing you will find out that simulating something as simple as a propeller is actually quite complex. First observation is your picture of the flow is showing a characteristic of insufficient run time. Whilst the answer may look to have converged in terms of your goals the model needs to fully allow the swirl to settle down particularly outside your rotating region downstream.  If you set up your flow trajectories then you will normally see a bunch or rotating arrows close to you model. If you let the model run even though the goals have converged then you will see the swirl trajectories move further away from the prop. I find it useful to set up the goal plots in Solver with a very narrow Absolute scale around the parameter. Whilst it may look flat on a large scale it usually doesn't on a very narrow scale. it depends what answer your looking for, see what the flow is doing or get a numerical number.

           

          Normally if the model is not converging on what you think is the correct result (ie out by a factor of 8-10) then the cause is probably the mesh size. Unfortunately as you add cells to improve the result the computational time goes up which compounds the first issue of not having enough calculations I talked about above! You are using an External Analysis type rather than Internal. I would suggest creating a tube around the prop and using internal analysis. Set all the outer surfaces to environmental pressure. The length before the prop can be relatively short but the distance behind the prop needs to quite long. 5-10 diameters maybe more, play with it until the answer no longer improves. Think of stirring a bucket of water with a small stick, when you start stirring nothing seems to happen but water is still flowing over the stick. As you keep going the water starts to rotate. The same is happening with your model, take awhile to get wound up.

           

          Given that Lift is proportional to velocity squared then it stands to reason to decrease your mesh size (increase cell count) near the outer edge of the prop by using a series of rings as local meshes with different amounts of refinement to give more cells towards the outside.You can also project lines onto the prop to break the surface into several sections then use Equidistant Refinement with varying levels of refinement to better capture whats going on. Bear in mind the top of the prop may need to be treated differently to the bottom so you could you different local meshes on each side.

           

          I also find XY plots very useful in understanding results. You can place a line in a sketch on a plane or project it onto a surface then get a graph ( I prefer sending to Excel). They can be solids too except you need to uncheck merge when you extrude then go into Flow Simulation> Input Data>Component Control and uncheck the solid if you only want to use it for measuring. You can think of these as measuring devices. One word of caution currently if your measuring more than one thing like velocity X and Y set resolution to maximum otherwise you may not get the parameters measured at the same point on the line.