When trying to run a simulation , despite the initial prediction of 22 - 33 minutes and about 5.8 - 8.7 core hours , the simulation would not finish after 72 mins and it took more than 100 core hours spent. I canceled the simulation because there are only 500 core hours available in the plan, and there is a danger of getting out of quota.
Could you provide some information why this is happening, supposing the simulation is canceled are the core hours saved or wasted?
Hi @Zaff_Zaff, I think you could remove/simplify the very detailed portions of your CAD first. E.g. the bearing and the spheres, the capsules…
This should reduce the runtime by a lot. Defeature the model so that it’s composed of simple shapes at first and build on complexity as you have a sucessful run.
Oh, and a last comment - 20 modes are way too much. Try going for 5 or 10. You can also have them around a center frequency instead of retrieving the first ones.
Thanks for your initial info. I have some doubts though , I share it with you.
Bearings play a significant role in gun dynamics, supposing I remove them , is there any other alternative to capture the physics (the balls and needles in the bearings experience centrifugal forces)?
For the double row ball bearing there are 36 balls in total which are held by seals and pintles, is it a good idea holding just the balls and removing the rest details (pintles , seals)
Another thought of making a more balanced setup while keeping all the geometry is reducing mesh size , though the credibility of the results would be under consideration
As far as the population of the modes are concerned, the rotor rotates either at 500 RPM (8.5 Hz) and 750 RPM ( 12.5 Hz), how would a center frequency approximation be set ?
I see. I can’t think of a boundary condition that would cover the bearings, but I’m not sure the frequency/harmonic studies would provide you with the physics you intend to immitate either. The reason is that you can only assign linear contacts to these sorts of analyses, whereas in reality you’d need a Physical Contact to replicate their behavior.
Yes, removing the small details would certainly help - perhaps this is an alternative. Another one would be to make them all a single body and simlify the spheres so that there are more points of contact between them and the part below.
Exactly. The mesh overall looks good in my POV, I think simplifying the CAD is the way to go.
Maybe setting a center frequency at 10Hz or 11Hz and retrieving 5 modes at first? Consider this approach.
As a last note, 20 modes are indeed too much, note that you’d be running 20 simulations (that’s why you’re consuming way too many core hours in the process - besides the considerations regarding the CAD).