Experiences from modeling very thin structures in …

1 Experiences from modeling very thin structures in comsol multiphysics Bergen comsol User Group workshop April 10th 2013 Jan Kocbach Note! This is a reduced version of the presentation without figures for the coil case 2 Introduction Generally challenging to model thin structures in FE software Selected cases thin coil (RF module) Wellbore heat exchanger (Subsurface module) thin elastic layers (Piezoelectric transducer) Comment on CAD models Our approach usually to redraw model in comsol multiphysics Should be possible with Virtual operations (not tested by us) 3 Advices: Sweeping the mesh, Virtual operations, Scaling the mesh 4 Case: Fields from thin coil (RF module) thin coil in air excited by voltage First attempt Change mesh parameters (resolution narrow regions, etc.) Limited success (meshing possible for somewhat thinner coil) Second attempt Exclude coil from simulation domain, impedance boundary condition Still not possible to get mesh set up Third attempt Convert coil to thin surface Transition boundary condition for losses Meshing possible for somewhat thinner coil, but now convergence problems in solution 5 Case.

2 Fields from thin coil (RF module) Fourth attempt Change meshing algorithm Start meshing the surfaces that have problems (coil) Then the small domains sorrounding (make a tube/donut around the coil) Finally mesh the rest of the structure Meshing successful for thin coil but convergence problem for solution Final solution Start meshing the surfaces that have problems Small non-zero conductivity (sigma = 0 gives convergence problems) Avoid metal in model ( subtract domains to avoid meshing inside metals) Use Magnetic fields physics interface (better convergence) 6 7 8 9 10 11 thin elastic layers 12 13 14 15