I am running Phoenics 3.3 on windows 2000 pro.

I am running a large model and Phoenics just stops in during the model run, at around ISWEEP = 9. When I enter ISWEEP less than 9 it works but the solution has not converged.

Has anyone got any advice on where to start to fix this problem.

Thankyou for your help.

Leon

(1). Is that the only parameter to stop the calculation? Are there other parameters involved to stop the calculation?

That is all that I am awear of.

Phoenics does not give any error messages at the end of the run, it just stops.

Dear Leon

I suppose this model was implemented into GROUND and if it is large (in code lines) the probability of human coding error are exponentially larger (even if you have several years working on PHOENICS).

It could have many reasons. What is the model for? fluid properties, interface properties, solid properties, radiation, chemical reaction... Have you solve the same case without the model activation? I mean, with constant fluid properties, nor radiation, nor chemical reaction.

You can check all the solved variables and it can be done in several ways. Nevertheless, it exist a graphical output interface which is simple and efficient for monitoring the variables.

Insert the following line into your Q1 file.

TSTSWP=-1

In such a way you will see a graphical windows during the calculation in which you can identify the variables behavior and their residuals every sweep (ABS(TSTSWP) = 1). The Q1 command above should be togheter with the declaration of:

IXMON = IX FOR VISUALIZATION CELL IYMON = IY FOR THAT CELL IZMON = IZ FOR THAT CELL

Try to locate the visualization cell inside the region where you have your model active. It will be easier to check if is there where the any "model variable" start to diverge. I think it is a good point to start.

Can you split your model into "steps or stages". If so you can check the result of every "step" activation. for instance, to consider surface tension I track the interface (first step), then I calculate interface normal vector components (second step), followwing the interface curvature (third step) and finally the source term into the momentum eqs. I can check how the result of every "step" influence the next "stage" in the model.

I hate large models, I think the larger the model the bigger its difference with reality.

Regards

Kike

Thanks for your help.

My model now works.