Hi Yu-Hsiang, Sorry that for some reason, your reply didn't trigger an email message to me. So I didn't know you replied, until today when I was randomly browsing the forum...
So if you are still working on this project, hear me out. We probably didn't miss that much, since the past month I was busy defending my final thesis and got little time to develop it anyway; and what you want probably does require a bit of development. The reason for that is, although like I said the particles can portal through boundaries, the contact cannot, meaning if they are densely packed, boundary particles may bounce back and forth across the periodic boundary because they can only have contacts with other particles from one side of the periodic boundary in simulation, where they need to have contacts with both sides. The solution to this problem won't be particularly elegant, but I want to create a demo that correctly does this. However, this does take time and I cannot be sure if anything can materialize in a month's time. I asked if the packing was dense because if you don't need inter-boundary contact, then it's easier. but it seems that's not the case. So to fully achieve your goal some time is needed. Another thing I'd like to mention is that neither DEM-Engine nor C::GPU can run solely on CPU. If there are only around 1000 particles, then Chrono::Multicore may actually be a good starting place instead. It can simulate interactions between complex-shaped geometries and supports user callbacks, and it runs on CPUs. I don't know about the periodic boundary thing, but you can create another thread on the forum to ask and I am sure people can answer it. Thank you, Ruochun On Thursday, September 14, 2023 at 2:56:48 PM UTC-5 YuHsiang Lan wrote: > Hi Ruochun, > > That's a good news. It will be good enough for me to have a CPU version. > Our initial case won't be that large (says 1000 spheres, stacked in about > 6-10 layers from the ground) > > The domain size for x and y can be the same at this moment. > Once we have a periodic x-y-squared box, I can duplicate it to get desired > aspect ratio on my end. This will be good enough at this stage. > > We want randomly densely packed spheres. All spheres. All radii are the > same constant. > The target average porosity is about 40%, but 40-50% are acceptable. > Ideally, all spheres are just touching each other, but we accept small > overlap as well. > > Thanks, > Yu-Hsiang > -- > > On Tuesday, September 12, 2023 at 7:21:27 PM UTC-5 Ruochun Zhang wrote: > >> Hi Yu-Hsiang, >> >> Chrono::GPU cannot. >> >> DEM-Engine can achieve that by enforcing a per-step check that moves all >> DEM elements that suffice certain criteria to a new location. But it could >> require a bit of additional development. >> >> DEM-Engine can also partially support a no-computational-cost single >> periodic boundary (only x or only y). If the domain sizes on the x and y >> directions are the same, then it should support a double periodic boundary >> as well. It would allow the elements that pass through that boundary to >> re-appear on the other side, but no inter-periodic-boundary contact will be >> detected in the process. >> >> I might be able to help you make this work, but I'd like to understand it >> a bit more. Are those pebbles densely packed in your fluid domain? Or do >> they tend to be relatively spread out? And are there only spherical pebbles >> needed? Do you need or have a plan to use complex-shaped pebbles? >> >> Thank you, >> Ruochun >> On Tuesday, September 12, 2023 at 6:45:26 PM UTC-5 YuHsiang Lan wrote: >> >>> Hi Luning, >>> >>> The problem is to generate randomly packed spheres in a double-periodic >>> 3D box. >>> We only need Project Chrono's DEM simulator to generate the sphere >>> centers, so periodic boundary is only for pebbles. >>> >>> From sphere centers, we have our in-house code to generate the hex mesh >>> for the void space between spheres. >>> Then, we run large eddy simulations with our computational fluid >>> dynamics code, NekRS. The final target is to study the boundary layer from >>> the river bed to the medium. >>> >>> Thanks, >>> Yu-Hsiang >>> -- >>> >>> On Tuesday, September 12, 2023 at 6:01:54 PM UTC-5 [email protected] >>> wrote: >>> >>>> Hello Yu-Hsiang, >>>> >>>> Can you provide more details of your problem? Which solver are you >>>> using for you application? Is the periodic boundary condition for both the >>>> fluid and the pebbles? >>>> >>>> Thank you, >>>> Luning >>>> >>>> >>>> >>>> On Friday, September 8, 2023 at 8:14:10 PM UTC-5 YuHsiang Lan wrote: >>>> >>>>> Hi, >>>>> >>>>> I'd like to generate an infinite river bed with spheres and then solve >>>>> fluid dynamics on it. In order to reduce the computational domain, I'd >>>>> like >>>>> to have a 3D box container filled with random packed spheres with double >>>>> periodic BC in both x and y directions. >>>>> >>>>> I found posts in this forum that there is no periodic support in 2018 >>>>> and 2020. >>>>> Do we have now in 2023? >>>>> >>>>> Thanks, >>>>> Yu-Hsiang >>>>> -- >>>>> >>>> -- 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/7334d5c6-1c09-479c-be0c-35c7077ff64en%40googlegroups.com.
