[zhengjg]

I use Fluent to resolve unsteady subsonic flow (Mach number = 0.27) around a stalled naca0012 airfoil with chord length based Reynolds number of around 1.5 million. The density-based solver with 1st order implicit time stepping scheme is employed.
The vortex shedding frequency is known. For each shedding cycle, 50 time steps are used. However, with the density-based solver, the solution cannot be be converged after more than 50 sub-iterations for each time step. However, for the pressure-based solver, the solution is converged in around 10-20 iteration for each time step.
so, why does the density-based solver not converge and how can I solve the convergence problem of density-based solver?

Any suggestion is welcome.
Thanks!