In particle tracking in CFX 10.0, why can the particle source coefficient affect the solution and not just convergence? I thought that in CFX-10.0, source coefficients only affected convergence.
1. The particle source coefficient affects the accuracy of the track calculation and can therefore change the solution.
This can be explained by the fact that particle momentum equation is solved as an ordinary differential equation (ODE). If we take a simple ODE like
du/dt = -k*u
Integrating it with a source coefficient is like solving it by passing the linearised term on the left hand side, ending up solving something like:
du/u = -kdt
which has the solution
u=u0 exp (-kt) (*)
Integrating it without linearising it (or setting the source coefficient to 0) is equivalent to solving the equation by keeping the source term on the right hand side. This way the solution obtained is:
u = -k*u*t + u0 (**)
Now the solution obtained without linearisation (**) is an approximation to the exact solution (*), which will only be accurate when using very small timesteps.
One way to test whether the linearisation is appropriate to get a good accuracy is to increase the parameter 'Number of Integration Steps per Element' under SOLVER CONTROL/PARTICLE CONTROL. The default value is 10. If this value is increased to e.g. 100 (this will increase CPU time!) and if the solution doesn't change much then the linearisation is reasonable.
2. However, for the other models in CFX-10.0, which are solved in residual mode, it is true that the source coefficients only affect the convergence and not the final solution.