Interpolation of results
 

Are we suppose to interpolate results from previous simulation if we change the boundary conditions of our domain. On what occasions we need to interpolate results and when can we simply re-start our simulation? Thank you

alpha

Re: Interpolation of results
 

You interpolate your results if you change the mesh. Changing boundary conditions has nothing to do with interpolation. Boundary conditions are just to close the set of transport equations appropriately.

For a steady state simulation as long as your results converge it doesn't matter whether you started midway or all over again. You care about the final results. Sometimes you can't get convergence with the conditions you are supposed to run, therefore you change parameters midway to get your results converged.

In transient simulations, it is good to start from zero, but if the computational cost is high you may want to store your data at certain time steps, and restart your runs next time using these stored data.

HTH

Ertan

Re: Interpolation of results
 

I agree that interpolation is usually made when there are grid interpolation. That is the interpolation of the computation domain, including the boundary surfaces.

If there are transient changes that you want to, say move on the boundary surface, then you could interpolate respectively, according to your specification. For example, a point source of heat that moves along a boundary surface, as a transient BC, then you specify the interpolation respectively to the distribution. If something changes the value on the field, you have to use interpolation, of course...

In unsteady simulations, you can start from a steady state computation. That will be the initial condition, unless you want to compute or define the initial conditions yourself, in that case, start with the first unsteady step. If you want to change the boundary condition you can change it to your specification.

regards Hafidz Rahmat

Re: Interpolation of results
 

Thank you guys that was helpful.

alpha