I wrote a small code to calculate the temperature distribution in 1D heat flow. I now have to limit the number of iterations using a convergence criterion.
I calculate the residuals as the difference between temperatures of two iterations. I was asked to normalise the residuals by flux. (Calculated flux as k.gradT). I summed up the residuals of all cells and divided it by sum of fluxes of all cells. It worked for this case.
Could someone tell me is this the right way to do it?
|All times are GMT -4. The time now is 10:14.|