Hi there,

i am working on a pseduo 2D-Simulation of an airfoil. When using the SST-Model i get oscillating residuals so the simulation is not converging. I tried to set timescale manually but that didnt help either. (should i use smaller or bigger timescales). I use a rectangular simulation volume with about 35000 Hexas. To simulate differrent angles of incidence i am using two inlets with cart. vel. components and two outlets.

Any clues?

Thanks. Kramer

If you are looking at high angles i.e. separated unsteady flows, then the steady state residuals will oscillate. Try a transient simulation? I've recently been using the SST on a 2D aerofoil and managed to get really good convergence for angles below stall.

Hi James,

Thanks for your help. But i my case i have a very moderate angle far away from stall. Do u think the meshing strategy with two inlets to create the angles with defined cart. vel. components in pre might be a problem. Is it better to rotate the airfoil? Or maybe timescale problems?

Thanks in advance.

Kramer

Kramer

I also used the inlet condition, it worked fine for me. I used a C-Grid. Inlets set on the C & bottom boundaries of the mesh. Pressure boundaries on the top & downstream boundaries.

It's worth checking to make sure your first node is correctly located off the section (y+=1.0).

James

Hi there,

I am new to this forum but I have hade quite some problems with oscillating convergence as well.

My experience is that the physical timescale you can select manually has a fairly large impact on convergence. I guess you were using the automatic timescale, so take the value from there and start reducing it. The smaller the timescale the more robust your solution should become. If you select the timescale to be very small, then you might slow down convergence.

What kind of boundary conditions are you using for outlet? If you are using the opening boundary condition, you might try switching from the "stable" to the "entrainment" condition.

As far as I remember also the reference pressure of the domain influenced convergence in my cases, but I am not too sure about that.

Hope that helps a bit, best

Sebastian

Hi Sebastian,

Thanks for your help. I tried to set the timescale manually, but again after about 200 Iterations the u-mom residuals begin to oscillate. Even if I choose 0.001 for timescale. I tested both BC opening (stable/entrainment) and outlet makes no difference. It's strange that it always begin to oscillate after about 200 Iterations no matter what timescale factor i use. Arrgh.

Maybe someone has any more hints.

Thanks in advance.

Sorry that didn't help.

Have you checked on the mesh, in particular on the y+ distribution as James suggested? If you are using the SST model, the implemented low-Re model requires y+ values along the walls to be smaller than 2 for accurate predictions and a proper functioning of the low-Re model. So maybe it is a grid issue? What are the RMS values when the oscillations begin?

Best, Sebastian