Phase Change+strong Marangoni, FIDAP - need a help

Vitaliy Pavlyk

Hello!

I am using FIDAP to model a 3D-fluid flow and temperature in a weld pool. The problem is strongly non-linear due to the temperature dependend material properties and especially due to the non-liner heat-flux boundary conditions ( radiation, convection and evaporation from the weld pool surface. In addition, the solid-liquid pase change and the strong Marangoni flow ( free surface) makes the problem extremely dificult to converge. The only segregated solver can be user due to the size of the mesh ( 50000-120000 nodes).

I wonder I somebody has experience with modelling such kind of problems?

my questions are :

1) which presure algorithm is the best

a) Presure-correction

b) Pressure- projection

c) Pressure -update ??

2) does upwinding help?

3) by using PRESSURE-DISCONTINUOUS ( 27 -node quadratic elements) which clipping is best

a) high - 1.0E-7

b) low - 1.0E-10 ?

4) wchich relaxation approach is the best?

I use currently RELAXATON( HYBRYD) with explicite factors for velocity components and pressure 0.95-0.98 / temperature 0.5-075/ Implicite factors for velocity components, pressure and temperature 0.1 0.05, 0.1 0.05, 0.1 0.05, 0.01 0.005, 0.05 0.0005 recpectively.

5) it lookes like only switching off the JACOBI Option ( NOJACOBI) and NON-SYMETRY REDUCED STENSIL (NONREDU) sometimes, also a poor, but convergency can be provided.

Usually it takes over 500 iterations to come to a more or less small solution change ( 1.0E-2 for velocities, and 1.E-3 for temperature) if at all. ( 2-3 days on a workstation I have).

Thanks a lot for the help,

Vitaliy.