Learning Forum

[Moussa]

Hi,

I am currently modeling transient natural convection of a liquid inside a tank but my simulation doesn't converge using SIMPLEC algorithm as mentioned in the article.

[Rob]

Do you have a gas phase present too? Mesh (and cell) quality and time step are more likely to be causing the problems. n

[Moussa]

Thank you, RobnI'm simulating the liquid phase only (H = 1m, W = 1m), the time step = 1s, the mesh is fine but my problem is the energy residual.nnn

[Moussa]

Results with the COUPLED algorithm, the simulation is turbulent, g = 9.81, the problem is symmetric and steady.nThank you in advance nThe results from the articlenn

[Draxler]

Did you change the Residual values to a smaller number than the default? nMost youtube videos change them to 10-6 and I was doing that as well. For very simple geometries, this works, but for complex ones, I was also having a problem with converging. I changed them back to 10-3 and the solution did converge. nSmaller residuals lead to more accurate results as far as I know. But in my comparison of the 2, there was very little difference.nn

[Rob]

Given the SIMPLEC has only done 140 iterations you may want to try running it on. Also note, the PBCS solver uses a courant number or time factor so may converge better if you've not adjusted any of the under relaxation defaults. n

[Moussa]

Thank younThe energy residual is falling under e-11, for the sufficient convergence of the iterative solutions, they use e-10 for the energy equation and e-0u00035 for the other conservation equations.nThe results at the initial state (t = 0) are obtained from the steady-state calculation at ptank = 1atm ( temperature contours posted earlier), there is a big difference between the two algorithms.nResiduals using SIMPLEC algorithm in steady-state simulation.nn

[Rob]

Which isn't converged. Try dropping the momentum under relaxation factor. What density model are you using?n

[Moussa]

Thank you, Rob nI'm using the Boussinesq model, steady-state, k-epsilon, g = 9.81, the liquid is methane.nMomentum under relaxation factor = 0.1nThe problem is the energy residual dropping I think nWhen I use the COUPLED algorithm with momentum under relaxation factor = 0.1nnn

[Moussa]

Any advice? Thank you in advance.n

[Rob]

I'd be more worried about the continuity equation in the first case. Given PBCS uses a time factor I suspect that's why the convergence is better. Converge the first case properly and then compare the results. n