Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Problem: Laser-Welding + Marangoni

Please login with a confirmed email address before reporting spam

Hello everybody!

I have a problem with my model of a laser welding process for structured steel with marangoni convection.
The model is a 2D modell. In face the problem is symmetric but I simulate it without symmetry axis to elimante a source of error. I use a triangular mesh which is very dense at the surface and density decreases with distance to the heat source.
I use comsol 3.5 with the following moduls:

- Heat and Conduction
- Navier Stokes for the stream in the melted steel
- Boundary PDE with the equation of the marangoni sample model from the COMSOL Library
lm_test*(eta_ns*uy-d_gamma2*Tx)+u_test*lm

The Temperature dependent phase shift effects on the following parameters.

- heat conductivity
- heat capacity
- mass density
- viscosity

the heat conducitvity and capacity is a fit of single values which is quite smooth.
the mass density is decreasing with the flcsh2 and width of 50K.
the viscosity should drop from a very high value (I take 500 but in reality it's magnitudes higher) to 0.055.
Therefore I also used the flcsh2 function with a width of 50K.

The Laserspot is considered as a gauss heat source.

The model works fine for high activity values (0.1-0.5) which also effects d_gamma2

gamma2 = gamma_m-A*(T-T_m)-R*T*gs*(log(1+k*a*exp(-delta_H/(R*T)[1/kg])))[1/kg]
d_gamma2 = diff(gamma2,T)

For lower values of the activity my modell is not converging. The criticall point is for some values the point when it beings to melt for
other points the melting is allready excisting but the modell can't be solved until the end.

For higher end values of the viscosity the modell can be solved but the solutions are of course unrealistic.

I use a timdependet UMFPACK solver with Generalized alpha method, as BDF turned out to be less robust in my case.

I would be great if you have some advice for me how to make my modell more robust for example stabilization parameters for the single moduls or a better function for the phase shift.
As I don't know anything about the solver maybe you can tell me if these settings are right.
Would a moving mesh be advantageous?

Tanks for your help!
Lukas


1 Reply Last Post 2011年9月8日 GMT-4 23:06

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月8日 GMT-4 23:06
i encountered with the similar problem, which indicates that the non-linear solver does not converged. Could it results from the solver setting?

best regards
i encountered with the similar problem, which indicates that the non-linear solver does not converged. Could it results from the solver setting? best regards

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.