According to your experience, how long a simulation can take? I want to know if my computer's performance is normal or I need to get a better one. I am simulating a 3D model of 20,000 cels (Hex), it takes about 4 hours to simulate 10 sec. Is that normal?.

Thanks a lot

Har

Thats not enough Information, which models do you use? ( phases , turbulence , .......... )

First of all Thanks for being interested in helping me.

The information is:

Mesh: 20,000 cells (Hex) Cells: 50 mm (edge) Model: Mixture (3 phases) water, air and sand. Viscous model: k-e Solver: Piso. BCs: Inlet (vel inlet), outlet (pressure outlet), top of the tank (outflow). Unsteady flow. Time run: 3 hours (approx) Time of simulation: 10 sec

Specification of the engine: UNIX SERVER (Not personal workstation - this is for the whole university): Quad 3.33 Ghz Intel Xeon 64bit processors with HyperThreading, appearing to be 8 processors, yielding 53162.8 BogoMIPS, 28 Gigs of RAM, and an 29 Gig Swap Partition. THIS IS DIVIDED BY THE NUMBER OF PEOPLE WHO CAN BE SIMULATING AT A TIME.

Thaks for your comments.

Humberto

hello HAR

Nice machine you have at your disposal and multiphase problems are perfect canditates for such high end computers. However I hope you are not running in parallel, because for such small grids the data transfer between the processors would detoriate the performance dramatically. It would be intersting to know how many processes are actually running on this machine, what is the time step in your simulation and how many iterations per timestep does it take to reach convergency? Max

I use different time steps during the simulation,

A. 0.001 sec, 1000 Steps....then B. 0.004 sec, 1000 steps....then C. 0.007 sec , 700 steps.

Each step has 20 iterations.

Total run time: 3.5 hours.

How is it look?