Hello, Im having some problems with COMSOL multiphysics 3.5a. I am using 2D weakly compressible flow with general heat transfer and am trying to model a natural convection loop. The Approach I have taken is to use properties of the material for a certain range of temperatures to model the variations in density with temperature which are responsible for the natural convection loop. I have got the model working only for very small temperature difference (1degK) but above this temperature difference the model goes haywire. I was wondering if anyone has any insights into this problem.