Discussion Forum

Hello,

I am working on the modeling of the spray forming process. We use a molten metal spray to spray cylindrical, tubular or flat products on a substrate. In my case, it is a cylindrical shape. Well, not really cylindrical, especially its tip can be curved in different ways. Final dimensions are about (diameter) 200mm times (height) 500mm. I know the geometry of the product over spray time from image analysis and I am interested in the thermal history including solidifcation effects etc.

The current model I am using is working this way:
I have a matlab script that puts together the geometry from thin layers (about 800 layers). So the product is put together layer by layer with each layer having a certain thermal distribution. The matlab code also sets all the boundary conditions (convecntion, contact resistances etc.) so that I can do a calculation how that layer cools down until I put on the next layer. So far so good, it works pretty well, but a different formulation of the problem would a advantageous for us:

Instead of puting the product together layer by layer, I would like to move the surface boundary. For each boundary point, I know the trajectory. As a boundary condition, I would like to use the following:
- boundary velocity and direction (at each point)
- convection and radiation (no problem so far)
- heat input from deposition through heat flux (also no problem so far)

finally my question:
I used a comsol mode for the parametrized geometry and unfortunately, if the geometry has already a thermal distribution, no materials seems to be actually "added", but it seems the thermal distribution is just stretched to fit the new geometry. What I would like to do is NOT stretch the initial thermal distribution but add new "material" on top given by the boundary condition. Is this possible at all?