How to fix Error from fLesGetHbrgEig <40> in acusolve (acusolve failed to launch after certain iteration)

Mudassir khan

Hello Expert 

I have developed a CFD model solving dam break condition air-debris multiphase problem. I have discretized CFD domain with 448789 nodes, 64440 2D element, and 1278663 3D element. My PC has 16 GB RAM with 8 cores. I started the simulation for transient condition for 5s with time step size 0.05. First Simulation run for very long time almost for 24 hr and after that it terminate showing following error. Please help me eliminate the following error and to obtain quick result 

Time-Step= 35 ; timeInc= 5.000000e-02 ; time= 1.700000e+00
acuSolve: Flow stagger "flow": FORM-LHS
acuSolve: pressure res ratio = 7.300083e-05
acuSolve: velocity res ratio = 1.323510e-02
acuSolve: CGP No iterations = 49
acuSolve: CGP 0/1/n norms = 1.851304e+00 1.840254e+00 1.217291e-02
acuSolve: CGP Iter. CPU/Elapse = 1.442100e+01 2.073000e+00
acuSolve: GMRES No iterations = 270 (27.00)
acuSolve: GMRES 0/1/n norms = 3.894095e+00 3.133886e+00 3.866773e-02
acuSolve: GMRES Iter. CPU/Elapse = 1.380160e+02 2.000700e+01
acuSolve: pressure sol ratio = 4.382404e-01
acuSolve: velocity sol ratio = 8.762230e-02
acuSolve: Turbulence stagger "turbulence": FORM-LHS
acuSolve: eddy-visc. res ratio = 6.386642e-02
acuSolve: GMRES No iterations = 32 (0.80)
acuSolve: GMRES 0/1/n norms = 7.033538e-04 7.033538e-04 6.870205e-06
acuSolve: GMRES Iter. CPU/Elapse = 2.734000e+00 4.040000e-01
acuSolve: eddy-visc. sol ratio = 1.966387e-01
acuSolve: Levelset stagger "levelset_step_1": FORM-LHS
acuSolve: levelset res ratio = 1.000000e+00
acuSolve: *** ASSERTION in Function <lesGmres> File <lesGmres.c> Line <1788>
acuSolve: *** Error from fLesGetHbrgEig <40>
acuRun: *** ERROR: error occurred executing acuSolve"
acuRun: Sat Sep 24 18:17:19 2022

acupro

Multiphase can be quite computationally intensive.  First suggestion would be try a smaller time step and see if this helps - maybe 0.01 instead of 0.05.  It may even require smaller time step than that, but that's at least a place to start.  It may also require finer mesh, but I would start with smaller time step.

As you look at interim solutions (assuming you're writing output somewhat frequently) do the results look realistic?  Does the maximum velocity magnitude seem realistic?  Etc.