Discussion Forum

Hi All

Using an adapted mesh roughly numbering 2 million mesh elements, the solver calculates based on 14 million

degrees of freedom which is outside my RAM capabilities (48Gb). When using a normal mesh of 2 million elements, the solver only has around 2-2.5 million degrees of freedom. Why?

If it helps, here is a quick summary of the model: I am using the non-isothermal flow module with pressure, temperature and velocity as the dependent variables, there are also weak contributions to the velocity (marangoni forces) on a couple of boundaries modelled using two weak expressions (one each for x and y components of velocity). The model simulates the melt pool formed during laser melting of a solid. The adapted mesh is created using an identical technique to the one in the 'continuous casting' tutorial module. I have successfully generated an adaptive mesh reducing the width of the temperature peak (dT) from 500K to 75K. I can get a solution using a normal solver with the finest mesh setting (extremely fine) and a higher dT (around 300K). Trying to use the adaptive mesh on a 'continuation' solver generates a ridiculous number of DoF.

My main concerns are why? is this normal? what can I do to reduce the numer of DoF when using the adapted mesh?

Many Thanks

Dan