I have computed a delta wing using LES, the solver is pressure-base segregated solver, with the the SIMPLE for pressure-velocity coupling, and standard pressure interpolating technique. The under relexation coefficients are kept the default numbers. But the convergence characteristic is poor. What can I do to improve its convergence ?

LES is the hardest thing to get the convergent solution. First it looks nice then after that fall into irregular plot result. My suggestion please decrease the underrelaxation coefficient. If not work, please change the mesh density to more dense.

Thank you for your suggestions. I have tried decreasing the under relexation coefficient, it does not improve convergence. Maybe the way i decrease the coefficients is not good, how do you decrease it ?