Discussion Forum

Hi,

I am running a model to test the thermal stress that appear at different zones of a solar thermal absorber made of prolypropylene. For this purpose i have sketched the axial section of this absorber, which is a pretty simple geometry, and the model is also simple as I only set fixed temperatures at the upper surface (temperature of irradiation), walls of the canals (temperature at which the water flows inside) and lower surface (5°C lower than the temperature in the canals).

I runned this model first of all without considering gravity effects, obtaining the results schown in "withoutsg.png", and I obtained the extrange result that the exagerated deformation shown while plotting the Von Mises stress, seems to be displaced to the right, like there had been any rotation effect, what I find senseless since all the settings of the model are simetrical in the x axis, so there shouldn´t be any displacement in this direction. More strange is that, when I add gravity to the component 1 (absorber), this rotation effect occurs in the opposite direction (anti-clockwise) as you can see in "gravity.png".

Later, to avoid the border effects in order to appreciate better the stress effects in the corners of the canals, I added symmetry to the vertical walls on the left and right borders (the 2 vertical lines limitating the 2D model). When I computed the study, without gravity, the effect of clockwise rotation could be even more noticeable (see "symmetry.png").

Anyway I can imagine that these results are not relevant, since I had to rise the relative error tolerance to 20 (by default it is 0.001 and I obtained allways a too high error to be able to run the study, so I just changed it to see what happened with the different options but now its too high), but it would be really helpfull if anyone knew why these rotation effects can occur in a model symmetrically built. Also if anyone knows how to decrease the error so that I can work with a lower relative tolerance, please tell me any idea.

I attach the pictures and also the model, with symmetry and no gravity but that can be simply changed at the structural mechanics section in the left (for Comsol 4.4)

Thank you very much