Hello. I was trying LES model using library caser 803 with 5 vertical cells which gives 85x60x5 grid. Initially, I thought there should be no problem running it, but convergenece was very poor. % error is magnitude of 10^5 for all variables. Time step size was 0.01 sec. 300 iterations for each time step. Inlet is uniform velocity of 0.3m/sec. Property is water at 20degree. I know that inlet condition and inital condtions are important. However, I just wanted to check if PHOENICS can do LES properly. Any one knows how to run LES on PHOENICS? Thank you in advance.

Library cases T610 and T611 perform 2D LES for vortex shedding behind a square cylinder. Strictky, LES should be 3d, but this library case exemplifies how to use the PHOENICS LES and get some quick results that agree reasonably well with the experimental data. Perhaps your time step of 10ms is too large, but the value required depends on the eddy turnover time scale.

Dear Mike, Thank you for the advice. I confirmed that those two cases converges well using 1/1000 sec as a time step assuming eddy life time is larger than that. I changed all required setup of library case 802 following t601 and t602 cases. Then ran it with 1/1000 sec time step size. However, the %error is still order of 10^5 and convergency is poor. I don't see how one case converges and the other doesn't with same setup. Could you kindly test my case and see if you can make it converge? Thank you.

Dear Mike, I managed to make the simulation converge. It was mainly on choice of scheme and GENK=T in group 19 of Q1. Otherwise, the simulation did not converge. I find it a bit confusing about the setting. Thank you

I've been using the LES model and it seems to work pretty well. A dimension with only 5 grid points looks pretty small for an eddy to be formed and may cause instability?