[DanielVe94]

Hello everybody!

I´m running a simulation of the flow around an Ahmed body with slant angle of 35 degrees and synthetic jets actuation on the slanted surface. I created a rectangular computational domain with the following boundary conditions:

- Velocity inlet of 40 m/s and turbulent intensity of 1%
- Pressure outlet at the back of the domain
- Wall conditions for the rest of the surfaces (to simulate a wind tunnel scenario)

I´m using SST k-ω turbulence model with COUPLED pressure-velocity algorithm. For the spatial discretization I set second order scheme for pressure, second order upwind for momentum, second order upwind for turbulent kinetic energy and second order upwind for specific dissipation rate. I specified a flow courant number of 50 and relaxation factor of 0.25 for both momentum and pressure. Also, the Under relaxation factors are as follows:
- Density :1
- Body forces:1
- Turbulent kinetic energy: 0.8
- Specific dissipation rate:0.8
- Turbulent viscosity: 0.95

Then, I first run a steady state simulation with 5000 iterations to simulate the base case scenario (no synthetic jet actuation) and residuals behave reasonable good (mass converges to the order of -5, k and omega converge to the order of -8, and velocities to the order of -6). After that, I change my simulation to transient (taking the staeady state data as initial iteration point) and I specify a sinusoidal time variation velocity profile on the actuator boundary to simulate the synthetic jet condition by mean of a UDF. After a couple of time steps I got the following error message:

# Divergence detected in AMG solver: pressure coupled -> Decreasing coarsening group size!
# Divergence detected in AMG solver: pressure coupled -> Increasing relaxation sweeps!

I tried reducing the time step of the simulation from 0.0005s to 0.0001s but divergence problem persists.

Any suggestion is welcome,

Thanks
Daniel

[LuckyTran]

Try running transient with the same constant velocity and get your numerical settings taken care of.

Ignore that the letters read courant number in the coupled settings. What's the courant number for your case? I don't have a feel for whether these time-steps are small or large. If you are running super small time-steps it's best to set the courant number in coupled settings to Inf and not use any under-relaxation. If you use a lot of underrelaxation, then your inter-timestep solution might not be converged which you carry into subsequent time. When do you see this divergence problem? The first timestep or many timesteps later?

After you do that and it works, then answer how do you implement sinusoidal velocity?