Oscillating solutions in steady state simulation
 

Hi everyone, I'm working on an hydraulic separator and my preliminary results are a bit confusing.
In order to decide if the simulation has converged, I use as reference the temperatures of the outputs. The only problem is that they oscillate even after thousands of iterations.
Is anybody able to explain this phenomenon?

Is this steady or transient model?
Please explain your problem.

It is a steady model. Two mass-flow intels and two outflows. The fluid travels from the inlets, in oppposite directions, to a chamber and then flow through the outputs. My problem is that the temperatures on the outputs never reach a costant value but oscillate in a 0.5 K interval.

There are several criteria for the convergence. And temperature fluctuation may be from many factors.
1. One possibility is an error in mesh or mesh is not much refined.
2. Check suitability of turbulence model.
3. Boundary condition.
4. Solution Methods.

best regards

I'm working on the mesh refinements.
I think that BC should be correct for this kind of problem, in fact I have to assure the same mass flow rate in both outputs, which must be the equal to the inlets.
The biggest problem is the turbulent model. Inside the chamber there are a lot of swirling regions and the fluxes smash against the walls when they flow inside. Do you have any advice regarding that?

Thanks a lot

Could it be the flow in your problem is actually oscillating? In that case, you will not find a steady state solution; the steady-state solver will not produce an average version of the actual transient flowfield, but it will cycle between the possible solutions of the transient problem.

As a thought experiment, consider a pendulum (frictionless, and already moving). If you were to simulate this with a transient solver, obviously the pendulum would swing between the "left" and "right" position with a certain frequency. If you were now to switch to steady-state, it would not give the central position as the solution of the moving pendulum because the pendulum has a non-zero velocity in this position. It would still oscillate, but now with some meaningless pseudo-frequency. After all, if it is in the "left" extreme position, the systems energy would lower by going right - and vice versa.

So yes, the factors above (mesh, model, etc.) may play a role - but first verify whether your problem actually has a steady-state solution, else all the refinement in the world won't help you find it.

Thank Chees, I'm considering the fact that my outfows boundaries might be too close to the chamber of the separator. In that the case, the cause of the oscillation could be that the zero diffusion flux condition is not satisfied.

Problem partially solved. I have extended the distance of the outflows and the mass flow inlets from the chamber, and now the temperatures reach a stable value. The great problem now are the pressure drops.