Hi, If you got some force readings then perhaps that bug is not the cause, but kindly try that anyway, if you would. But if the cone's velocity is not high and it is not super heavy, then it should be stopped by the granular bed, should it not? I saw in the animation the cone almost "phased through" the material. I can only see it from the side so I am not 100% sure, but did you see the same in Paraview? You can apply glyph (sphere) filters to the particles (with their diameter being the scaling factor) and/or reduce the opacity of the cone to help you observe. The cone should push the particles out of the way while descending, and finally rest in the crater it creates. Pay attention to the particles that are originally on the trajectory of the cone, do they get inside the cone?
In a co-simulation, the GPU module does not manage the Chrono objects such as the cone, it just reports the force that cone should feel to Chrono core, and let the Chrono core handle their dynamics, then receive updates of the locations of objects from Chrono core. Looks to me that in this test the reported force and torque are not honored by the core module somehow. Can you give a better rendering of one of the cases that worked for you so we can go from there? Thank you, Ruochun On Sunday, July 3, 2022 at 5:05:53 PM UTC-5 [email protected] wrote: > Hello, > > Thank you so much for your reply. The forces I obtain from this simulation > are considerably large (on the order of 10^4-10^6). If the particles do not > interact with the mesh, I am not sure how these forces were generated. I > assumed that the bed particles are not showing resistance because the speed > of the cone was small (that is probably incorrect). Also, when running the > simulation with Young's modulus value of 1e10, I get no errors at all (the > animated results are still the same). As for the file that I attached in my > first post, you do not need a JSON file to run the simulation. My latest > Chrono build was from the newest commit (I rebuilt it a week ago). However, > I will try one more time. > > Thank you, > Mohammad > > On Sunday, July 3, 2022 at 1:20:47 PM UTC-7 Ruochun Zhang wrote: > >> Hi, >> >> This animation does not look good at all to me. The cone seems to have >> gone through the granular bed with no resistance at all, and the granular >> bed did not seem to respond to this impacting object. Also, if you >> monitored the cone force, I suspect you just got an extremely small >> reading. Do you agree? I would like to ask you to do one thing: could you >> pull from the *develop* branch of Chrono (newest commit), then build and >> run your script again with it? This is because there was a bug in granular >> module that could potentially disable mesh--particle contact unexpectedly, >> if you used an older version. I do not think it is very likely that your >> simulation suffered from this bug, but let's rule out this possibility >> first. >> >> After that, the task is to make this simulation physical. I believe the >> "no contact pair when Young's modulus is below particular value" is just >> the by-product of some unphysical phenomenon, such as huge penetration, >> super large angular velocity, failure to detect contacts etc. And you have >> to debug based off a "minimal" run that is physical. To be physical, it has >> to look right at least... In your case, you have to find a test scenario >> where it runs without errors, and the mesh and particles seem to interact >> with each other normally (cone punches a crater on the granular bed). >> >> By the way, we still cannot run your script because there is no JSON file >> attached. On top of that, I am sorry that I currently do not have time to >> run or debug it even with the JSON file. But maybe I can read it to get a >> brief idea. I also think this post discussing the general >> question-asking practice <https://stackoverflow.com/help/how-to-ask> can >> be useful to consider. >> >> Thank you! >> Ruochun >> >> On Saturday, July 2, 2022 at 4:27:13 PM UTC-5 [email protected] wrote: >> >>> Hello, >>> >>> I have implemented one body and included the two meshes there. However, >>> the results I got did not match what I expected where the forces were not >>> producing the stress values that I was expecting on the tip of the cone. As >>> a result, I decided to switch back to my old method and see what >>> differences I get. I am only using three of the four bodies that are >>> created (I should probably comment on the one that I am not using out). One >>> of the three bodies is used as a guide for the motor. Also, I assumed that >>> both implementations would generate the same results since I am basically >>> trying to do the same thing in two different ways. However, I get >>> completely different results as well. I have animated the results for a >>> specific young modulus value and everything looks good when animated. >>> However, the physics is still different for both implementations. I am >>> including the animation of the file that I attached earlier as well as the >>> meshes I used and my other implementation for one body. >>> >>> >>> Thank you, >>> Mohammad >>> On Thursday, June 30, 2022 at 12:09:35 AM UTC-7 Ruochun Zhang wrote: >>> >>>> Hi, >>>> >>>> Did you visualize your simulation? I doubt if this script can produce >>>> the correct physics. Last time, I stated that I believed one Chrono body >>>> was enough, and you had to collect force from both pieces of your meshes. >>>> This script uses at least 4 ChBodies (it may work, but I do not understand >>>> how they are used here) and seems the force is collected from the tip mesh >>>> only. Again, that's my speculation, because I don't know the mesh you >>>> used; >>>> but I doubt it would work, with low *or* high Young's modulus. It may >>>> happen to not throw an error at certain settings. >>>> >>>> Perhaps more importantly, I know the tip force is what you are after >>>> but did you manage to get a one-mesh-one-cone test scenario running, >>>> visualize it and validate the physics it produces? I think you have to >>>> start there, and if you have that the rest should follow naturally. >>>> >>>> Thank you, >>>> Ruochun >>>> >>>> On Thursday, June 30, 2022 at 12:07:59 AM UTC-5 [email protected] >>>> wrote: >>>> >>>>> Hi, >>>>> >>>>> I am trying to use the Chrono gpu to achieve a similar effect as >>>>> the run a cone penetration test but have been running into some >>>>> difficulties. I hope that somebody here would be able to help me out. >>>>> >>>>> When setting the young's modulus for the particles and boundaries to >>>>> 1e9, I get the following error: >>>>> >>>>> - No available contact pair slots for body xxxxx and body xxxxx >>>>> >>>>> However, when I run the simulation with values larger than 1e9, the >>>>> simulation runs smoothly. >>>>> >>>>> It seems strange to me that there seems to exist a critical value of >>>>> cohesion that causes models to behave in unexpected ways. I am wondering >>>>> if >>>>> anybody here has had success with using low values for young's modulus in >>>>> Chrono GPU. I am attaching my file for your reference. >>>>> >>>>> -- You received this message because you are subscribed to the Google Groups "ProjectChrono" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/a6f835d8-3f27-437e-b20a-673102a0dc57n%40googlegroups.com.
