dicehub

#VAWT ran in #OpenFoam using @dicehub 'case run' feature - took 31 hours to run, then downloaded the results and did all post-processing on laptop using #Paraview.

Ran this #VAWT case on @dicehub recently - it took 31 hours, on just 4procs. More of an experimental run to scope out some of the issues/constraints. Model from thingiverse.com/thing:6734541. The rotation is 60rpm, wind speed 5m/sec. Ran as #incompressibleFluid in OF12 using #NCC

Done with @dicehub, beautifully enhanced wit @GoogleLabs

#incompressibleVoF catamaran ran on @dicehub this time with CFL reduced to 1 (instead of 10) for better resultion of the timesteps, and more accurate solution.

Boat hydrodynamics ran on @dicehub with #incompressibleVoF #OF12 solver (#interFoam). Case ran on 36 procs in 25mins, mesh 650k cells, simulated time 10 seconds.

#incompressibleVoF #OF12 (previously #interFoam) ran on @dicehub - 100 mins runtime, 36 cpu, 650k cells

#interFoam case ran on @dicehub

Did you know you can just setup your OpenFOAM case on @dicehub , generate the case and download it without the need to run it in the cloud?

This week been looking at gas mixing tank demo using the #multicomponentFluid model in #OpenFoam v12. I created this simple 3D geometry myself in Paraview. Its been quite challenging, but thanks to some help from @dicehub got it working. Buoyant gas mixing over 20 seconds.

#UFOCFD (octree) F1 car with rotating wheels, 30M cells ran on @dicehub GPU in 55 mins

#UFOCFD (octree) F1 car ran on @dicehub GPU in 55mins

#UFOCFD (octree) F1 car, 100m/sec, 30million cells, 20k iterations, ran on @dicehub T4 GPU in 55 minutes, all post-processing #Paraview 6.0.1 #ChristmasEve

#UFOCFD (octree) re-ran this XB-1 case with open ducts, to allow airflow through the intakes and exhausts, ran on @dicehub T4 GPU in 40 mins.

#UFOCFD (octree) 38M cell mesh ran on @dicehub T4 GPU in 42 minutes @boomsupersonic XB-1 Mach 1.18 model from thingiverse.com/thing:7128378

#UFOCFD (octree) @boomsupersonic XB-1 ran on @dicehub T4 GPU, 5 million live cells, 10 minutes runtime, 10k iterations, Mach 1.18, 400m/sec (ISA sea-level conditions), #Paraview v5.10, model from thingiverse.com/thing:7128378

#UFOLBM flow over sphere Re=600,000 (U=10m/sec), mesh 30 million cells, 100k iterations, runtime 2 hours, on single T4 GPU via @dicehub. Had to refine the mesh due to checkerboarding on coarser mesh. Bounce-back wall boundary used. All post-processing done with #Paraview.

→ @dicehub is a cloud-based platform for engineering simulations. It offers real-time collaboration, pre-configured templates, and scalable computing power to simplify complex simulations. 🔴 Live on Dev Hunt → devhunt.org/tool/dicehub

Just ran this case on @dicehub in 4hr 13min. The mesh size is 62.5 million cells and ran for 100k iterations. The buildings are 100m wide so the cell size is quite large (4metre). So 5m/sec gives Re=30million. This ran on a single T4 GPU, saving VTK slices every 1000 iterations.

#UFOLBM 32M cells, 100k iterations, AIJ CaseD ran on single T4 GPU via @dicehub - just need to fix the upper lower boundary condition because it looks periodic [top (nut), middle (vel), bottom (rho)]

#rhoSimpleFoam F1 car, 100m/sec, ran in parallel on @dicehub (18 processors), 1000 iterations, 1.8M cells