Learning Forum

[Weiqiang Liu]

Hi all,

I posted a thread about how to model porous soot cake combustion before. Now I finished all the code and tried to run this transient case. I have UDS which is local soot concentration in the case and a lot of source UDF to define source term.

I can get reasonable temperature contour during soot combustion. Soot is decreasing in the system and the maximum combustion temperature is almost the same with literature result. However, the velocity contour is very unreasonable. 

I found the continuity residual is never below 1e-2 during transient simulation. I think this is the reason why my velocity contour is not reasonable. Therefore, I tried to unload all my UDFs and remove UDS. Then I calculated a pure cold flow and very good velocity contour was obtained. Then I reload all my UDFs and UDS to continue calculation with this perfect initial flow field. However, velocity contour became unreasonable immediately after all UDFs and UDS were involved. I also tried to reduce URF of momentum. However, nothing worked.

I wanna ask how can I decrease continuity residual when running a complicated transient case with UDFs and UDS. 

Or are there any strategies I can utilize to get converged continuity.

Best regards

Weiqiang.

[DrAmine]

Smaller time step sizes are first thing to try. Check if your user defined source terms are linearized. 

Does the soot combustion affect the momentum field?

[Weiqiang Liu]

Hi Amine,

I just followed the time step setup in literature. I tried to decrease time step size. However, that did not work. 

In my source UDFs, I just let dS[eqn]=0. Because some source terms are in the form of Arrhenius's law. I don't know how to calculate the derivative of these terms to dependent variable. 

Yes, soot combustion affect momentum field.

Best.

Weiqiang

[DrAmine]

So it does affect I assume due to density changes? if yes consider the density URF.

Do you have good mesh? 

Which Solver are you using?

Regarding source term linearization: you should always try to initialize the source terms otherwise they are handled explicitly and this not a good idea.

[DrAmine]

All my suggestions are assuming that all UDF are properly set and defined and written.

[Weiqiang Liu]

Hi, Amine,

I have very good mesh just exactly the same with literature. I use pressure based solver. Yes, I'll try to linearize the term and check my UDF.

Thanks very much for your suggestions.

Best

Weiqiang

[Weiqiang Liu]

Hi Amine,

is the derivative of source term must be negative? Because I might have positive derivative for the source term.

Best

Weiqiang

[DrAmine]

Good point. Just provide the derivative. Fluent will take care if the sign is positive.

[Weiqiang Liu]

Thanks very much. I'll check if linearization will help.

Best 

Weiqiang

[Weiqiang Liu]

Hi Amine,

What if the derivative of source term is always positive. How fluent will deal with it?

Best 

Weiqiang

[DrAmine]

Sa I said Fluent will Deal with that as it does not make a sense using positive coefficient.

[Weiqiang Liu]

So I just need to provide dS. Whatever the value of dS is, fluent will transform it into negative?

Best.

Weiqiang

[DrAmine]

As I said Fluent will take care of it. I cannot tell more here on this open community.

[DrAmine]

If you have a university mail address I can tell you what does Fluent do with that.

[Weiqiang Liu]

Hi Amine,

My university mail address is w322liu@uwaterloo.ca

Thanks very very much!!

Best 

Weiqiang

[DrAmine]

Thanks. I sent you a reply.