I am using the Modulus-sym module to solve for the three-dimensional steady-state temperature field. My model is quite complex as it is a multi-layered structure with different materials having different thermal conductivities used at various locations, and multiple heat sources are placed in different layers. Currently, I have set different result variables T1, T2, … for each material, and applied heat diffusion equation constraints internally for materials with different thermal conductivities. These are then linked through boundary constraints. In this scenario, my model converges after a period of training, and the error is quite small, at the order of e-11, but the model’s temperature field differs greatly from the actual situation. Therefore, I am wondering if I should apply only one heat diffusion equation constraint for the multi-layered structure. As you know, I can represent thermal conductivity as parameters related to (x,y,z), which is challenging but achievable. I am curious if this could be the main reason affecting the accuracy of the results. I am looking forward to everyone’s help and replies. Thank you!
Note: Since obtaining the actual T through simulation takes a long time, I have not included a data-driven section.