I am trying to simulate a bollard pull test on my thruster model to validate CFD results with the experimental results provided by the manufacturer (T200 Thruster Documentation). Going by the product specs I should ideally have a thrust force value of close to 50N but what I am getting is almost half that (the model is on a 1:1 scale ratio). I downloaded my results on paraview, extracted surface normals and multiplied them with pressure for all surfaces and still get a low thrust of 28N (Assuming this is the right way to extract the force output?). Hoping someone can shed some light on this. Cheers.
I think this issue has to do with your material selection. Currently youāve set your material to an arbitrary type which will not represent the medium of what you are trying to simulate your propeller in, hence the lower force values. You need to instead import the material type via āImport from material libraryā and select your medium. Afterwards you can then set your density.
Lets see if this fix works and weāll move on from there.
I donāt have any specific details of the experimental set up but I have a fair idea how these tests are performed. So basically the thruster is thrown into a tank of water and attached to a fixed bollard to calculate various outputs.I will try running my case initializing with pressure and see if it affects the results.
Iāve been trying to see where it went wrong in order to obtain the expected values and so far all methods are not only far from the expected performance, but also donāt deviate as much from the first run. This usually means there is some big oversight somewhere that Iām still looking for but in the meantime I suggest obtaining exact documentation on how the methodology for testing such a device by getting in contact with the manufacturer themselves.
An easier comparison could be done with a rpm vs thrust curve that we then can, on top of ensuring we are obtaining the right data, validate the robustness of the simulation for different rpms.
Thanks again for looking into my project ! Have not had any luck with the results unfortunately. The thing that baffles me the most is that I have imported this model straight from solidworks and have run some test simulations on Solidworks Flow & CFX with the same boundary conditions, I get a global thrust output of 47N which is closer to the experimental data . Even I feel something important is being overlooked, just cant figure what it is yet.
Have you performed a mesh check analysis in Fluent and thought about improving the mesh inside SimScale a bit more as you still have 152 illegal cells? \rightarrowMesh Convergence Study?
Why donāt you use āfreeslipā boundary conditions on the sides?
Where do you get your inlet BC as well as the rotation speed from?
I think I may not be calculating the global thrust properly from the results. And one thing i overlooked thus far was that this is a ducted propeller, so the nozzle part contributes to the thrust as well. I am running a sim as I type this selecting all faces of the model including the blades and see if it gets me closer to the desired value. Will let you know how it goes. Cheers
Have not performed a mesh convergence study yet, but I feel my mesh is refined enough in the areas of interest to make a huge difference in the results. But thats something I ll think about.
Will try using free slip bc condition on my next run!
T200 Thruster Documentation - I got the rpm from this link and as for the velocity, the usual bollard pull tests in a tank are tested under the stationary water conditions and then gradually increased.
Let us know how things go! Very interested in the results/validation of this project. Would be amazing if we can create some nice post-processing pictures to decorate your hard work
Some mesh refinement and addition of free slip bcās pushed the global thrust output to 31N. That being said I am officially out of ideas on how to proceed further. Would love to hear your ideas. Cheers.
Sorry for the delayed reply. My new school term just started and was rather busy.
So far my major tackling point is probably the mesh and geometry, if what you mentioned about other solvers being able to produce the correct results, then I donāt think the simulation section or rather the setup of the boundary conditions should be wrong. Iāll be looking into each section to see if I can garner anything. No promise that Iāll be able to get anything useful so Iāll let you know if I do.
Interesting that the meshing did push the value up, however weāre still far away my a large margin so I still suspect it is some large oversight somewhere.
In the meantime if you are short on time, the runs you did in the other solvers will suffice as your validation case.
Sorry for the late reply ! Hope you are having a good start to your school term. I have not had much luck in validating the results.Kept refining the mesh as much as possible, went from 1 million to 8 million cells and the thrust force as it stands is 33N. Still way off. I ll keep playing with the bcās and let you know if something works. Cheers.
Hi @Targaryen, are you evaluation the Y+ values to ensure that walls are being modelled properly? Might be worth seeing if changing the layered cells parameters change results.
How would I check my y+ values. I dont think I have a boundary layer around the walls. I tried setting an inflation around the blades but i dont think it generated any layers.
Hi Victor, for incompressible Y+ can be added as a āresult control itemā. Add a field calculation and it is under dimensionless wall distance.
I looked at your mesh setup and here are a few pointers:
Expansion ratio should be > 1 otherwise they will actually get smaller, so for example if a layer is 1mm and the expansion is 1.1 the next layer will be 1.1mm. If you have n layers then it will be 1mm*1.1^n for the final layer but since SimScale works in reverse, you need to define the final layer thickness.
you are working in relative cell sizes (defined under operation, Layer adding controls then Use relative size for layers?: True). You can change this if you want to work with absolute values (meters) but otherwise, a relative value of 0.001 is too small try something around 0.1-0.3 for better results.
Not much change in results though.