I have a model with volumetric heat generation. There isn't much explanation of the "source coefficient" for volumetric heating and I was wondering if anyone knew how important this is to the final answer. At the moment, I'm getting more power out than I put in (m_dot*cp*dt) and, while it could be a number of things, I'd like to see if it could be that I need to enter a valid source coefficient to correlate the data.

-M

Hi Wooster,

In a converged solution, the source coefficient does not influence the final value of the generated heat. It only helps to get a converged solution by - more or less -underrelaxing the source term.

If you have a source A, you should define a source coefficient between -A*1e-10 and -A*1. The first implies a weak underrelaxation. Then in general you do not require a lot of iterations but the solution might become unstable and you won't get a converged solution at all. The latter implies a strong underrelaxation. Then you might end up with a converged solution but you will need a lot of iterations.

Just try to find out what suites you best. If you get more power out than you put in, you should monitor you imbalances. If this isn't equal to zero, you need more iterations.

Regs, Gert-Jan

www.bunova.nl

Thanks for the feedback. I wasn't 100% sure how the source coefficient worked and the manual is slightly confusing on how it works.

I think my problem with the power is due to the text file I was using had higher energy particles than what was measured in the benchmark, but I need to explore different reasons.

By imbalances, do you mean residuals? The residuals converge at about 26-30 iterations. The main residuals keeping it out that long are w-momentum and energy. w-momentum is just a function of my wild geometry.

-M

No, I do not mean residuals. In the solver manager you should create a new Monitor and then select that in submenu IMBALANCE you want to monitor mass, energy, momentum etc. These all should go to zero, meaning that what you put in comes out as well.

Gert-Jan

www.bunova.nl

Ok, thanks, I'll take a look at it and see what I can find.

-M

The source coefficient should just be an estimate of dS/dT.