Learning Forum

[JD_JN]

Hello,

I am trying to simulate a relatively simple 2D problem of a shock tube inviscid flow. It is a transient problem.

I have created 2 cell zones for two sections in the geometry (which is a rectangle) and after providing boundary conditions, I have enabled adaptive mesh refinement.

I did AMR by creating 2 cell registers for refinement and coarsening, and provided the refinement and coarsening thresholds for gradients of pressure.

However, after several thousands of iterations the solution residuals stop converging. PFA image of a residual plot.

[DrAmine]

What do you mean stop converging? What does the console tell? From the plot it seem it does only one or maybe few iteration per time step.n

[JD_JN]

nThanks for the response!nI meant that the residuals do not converge iteratively per time step after some thousands of iterations.nThe residuals stay around a value and do not drop below the residual convergence value (as indicated by the horizontal parallel line towards the end in the residuals plot).nPer time step, it does around 150 iterations.n

[JD_JN]

Actually, when selecting global average for gradient adaption, it does not perform AMR at all. The mesh remains as it is.nI should also mention that I am trying to achieve Dynamic AMR.nI have enabled that option, but either the solution is not iteratively converging, or it does not perform AMR at all.nI use an explicit scheme, density based, transient solver with a second order implicit transient formulation.nI made sure that the boundary conditions are correct, density is polynomial, and an ideal gas fluid.nI have run out of ideas. I am stuck on this for the past couple of weeks now!n

[DrAmine]

The lines are at which level (looks already low). Please copy paste a couple of line from transcript or screenshot them.n150 iterations per time step is insane, bad ant not recommended.n

[JD_JN]

nI couldn't give the response right away since I had ended the session (as it was not iteratively converging) and so I had to re-start the calculation from the last save point.nThe absolute convergence criteria for the residuals is 1E-6.nHere is the screenshot from the console:nnn