Results do not converge for a large schale CFD model

Hi, I am doing a ventilation analysis for a neighbourhood scale (14 building blocks, each is over 60m high, and there are some details for each building), but after 1500 iterations my results do not have the trend to be converged. I used fine scale mesh and the mesh quality is around 2.5. This is the right view of my result


I have also attached the convergence plots (domin, inlet, outlet, walls and residues respectively)

Do I need to increase my external volume? Because it seems that my box is not enough to see the overall impact of the vortex. Or I need to improve my mesh quality?

Hello sc2280 ,
in regards to the residuals, i think if you run the simulation for around 500 more iterations they should converge, and maximum residuals look also okay.

Overall I would extend the fluid volume. the flow at the outlet should be constant and no influence from the buildings should be visible. But if you are only interested in the flow at the buildings, I would just run a quick test to see if the results are hugely different. If not you could stick with this method.

Best regards

Hi, many thanks for your help. The graph for residues seems will be converged after around 500 iterations, but other graphs seem will be continue to vibrate for a long period. Can I say it is converged just after the residues become smaller than 0.001 ?

oscillation in the results can mean different things.

#1 Small oscillation frequency with small amplitude → usually within a couple of iterations. Numerical Oscillation induced by the solver
#2 Bigger oscillations, oscillation within the flow field indicates a transient behavior, and maybe a transient analysis can bring better results.

For you, I’ll guess that you have some transient effects from the flow fields, maybe a transient analysis can help, but I would say that simply averaging the results of the las n*100 iterations is sufficient.

Best regards