How to improve convergence in transient model of a CSTR
 

I am trying to use Star-CCM+ to develop the transient model of a stirred-tank chemical reactor handling fast-reacting liquid species which generate vapor and gas products. The current version of the model appears to perform well, but the residuals of some variables only decrease to values in the range of 0.005 to 0.0005 –and remain in that range even after 200 iterations (and time steps as small as 0.001 s).

I had been considering that convergence to be adequate (particularly since the iterations take quite a long time), but have received comments stating that I should seek residuals of at most 1E-6. I have thus tried smaller time step sizes, more inner iterations, and a finer mesh, but in some cases the simulation diverges, and in others the residuals do not decrease much.

I would appreciate any guidance on the matter, or a suggested literature reference on how to achieve smaller residuals.

Thank you

Hi,
Is the residual you said is normalized..?