Hello!
I am currently doing my first big projekt in comsol in which I'm trying to simulate a cooling system using the conjugate heat transfer module. I'm starting out with laminar flow for the cooling fluid (water) but will later need to have it turbulent as the Reynolds number requires that.
However, I am having problem with convergence as I add a boundary layer to the cooling channel mesh. Without the layer it converges just fine but as soon as I add any BL using the built-in feature the solver does not converge and the temperature of the fluid changes drastically into the far negative range (-1e9). This happens even though I have no heat load applied.
My first thought was that it is the mesh that is causing problems and when I plot the mesh quality the BL sticks out as the worse part of the mesh. However, my testing does not indicate that this poor quality is causing the problems and I don't think that is should because the BL is supposed to be skinny and long. Am I wrong?
The solver uses the multigrid preconditioner which has been causing me some problems when it can't automatically generate a coarser mesh but I think that I've managed to fix that.
What do you think could be the problem?
Best regards,
Kasper Hörnkvist
I am currently doing my first big projekt in comsol in which I'm trying to simulate a cooling system using the conjugate heat transfer module. I'm starting out with laminar flow for the cooling fluid (water) but will later need to have it turbulent as the Reynolds number requires that.
However, I am having problem with convergence as I add a boundary layer to the cooling channel mesh. Without the layer it converges just fine but as soon as I add any BL using the built-in feature the solver does not converge and the temperature of the fluid changes drastically into the far negative range (-1e9). This happens even though I have no heat load applied.
My first thought was that it is the mesh that is causing problems and when I plot the mesh quality the BL sticks out as the worse part of the mesh. However, my testing does not indicate that this poor quality is causing the problems and I don't think that is should because the BL is supposed to be skinny and long. Am I wrong?
The solver uses the multigrid preconditioner which has been causing me some problems when it can't automatically generate a coarser mesh but I think that I've managed to fix that.
What do you think could be the problem?
Best regards,
Kasper Hörnkvist