hi,all CFX4.3 or CFX4.4 can not compute the two phases flow with reaction directly. So I had to solve the problem using the user fortran subroutine, but I run into many difficulties. I used USRSRC to set source for mass fraction equation. Comparison with the CHEMISTRY module, I computed the single phase flow with the reaction using the USRSRC subroutine. The results(including single gas phase and single liquid phase) coincided with the result using the CHEMISTRY module. However, after I used the same way to add the subroutine into gas-liquid two phases flow, the result was incorrect and often overflow! Why? It had dazzled me for a week, I wish somebody to help me! Thank for your help in advance! Sorry for my poor english.

Hi, dingding

I suppose that this problem should be likely due to the non linear equations of mass fraction. In the CHEMISTRY module of CFX-4, a coupled solver is used to solve the chemical kinetics. When you use the subroutines (mainly the USRSRC)there is no treatment of the non linear terms of chemical kinetics and thus many problems arises such as numeircal overflow.

I suggest you to take care when you linearize the chemistry equations.

Regards

Hi Ribeiro and dingding I am dealing with similar system, chemical reactions and multiphase. And YES, I do get numerical overflow often, even when I just change the subroutine slightly. It's very frustrating. Could you explain a bit more on "linearizing" the chemical reactions, please? thank you.

Hi, all

The linearization is a mathematical procedure to trasform a non-linear equation into a linear equation in order to simplify this equation. There are many methods to do this but the most familiar is to use the expanding Taylor Series. That's I use.

In the case of chemical reaction kinetics equation, in general, these equations are non linear, i.e., they have terms with quadratic variables ([Ca]^2). Thus, it is necessary to become these terms linear ([Ca]).

This is because the equations solved into CFX-4 are linearised. Thus, the source terms (inclusive the kinetics equations) must be linear.

Most of Calculus books describe the linearization procedure.

I hope I help you

Regards

Hi people,
Daniel is right, care must be taken to linearize the source term. I'll give you a simple example, if you have a simple rate law, say k*Ca^2 (as Mr. Ribeiro mentioned), how to linearize it? In Patankar's book, he suggests a linearization expressed by this equation:
S = S' + (dS/dCa)' * (Ca - Ca')


where S' is your source term you want apply, but with exception that ' means values at previous iteration. Thus, performing the calculations:
S = k*Ca'^2 + 2*k*Ca' * (Ca - Ca')
S = k*Ca'^2 + 2*k*Ca'*Ca - 2*k*Ca'^2
S = -k*Ca' + 2*k*Ca'*Ca
Thus, SU = -k*Ca' and SP = 2*k*Ca' in your transport equation.


Hope this helps you. Regards,
cfd guy

hi cfd guy, Thank you for the example. It illustrates the method very clear. But isn't SU to be positive and SP negative or zero always?

Hi kahing,
I believe you're talking about the particular example given in the manual. Sometimes you can violate general rules but you must know exactly what you're doing. So, I suggest you to study a couple more what you want to do. There's a lot of cfd books which approach this subject, and in there you might find some valuable information to your work.
Good luck!
cfd guy