Learning Forum

[JIN LU]

The specific heat capacity of one of materials I used in the Transient-State Thermal module is temperature-dependent, and I input its tabular data of heat capacity vs temperture with temperature ranging from 15℃ to 200℃. I set the calculation time be 150s. First time, the solution gave the highest temperature 97℃, and then I change the tabular data of the material's heat capacity vs temperture to temperature range from 15℃ to 300℃, then I run the solution, but the solution gave the highest temperature 154℃, which is very different from the first time solution.

Both times, the highest temperature is blow 200℃，so the heat capacity of the material from 200℃ to 300℃ shouldn't take part in the calculations, and the highest temperature should almost be the same. I tried many time, but it's apparently not the case. Who could tell me the underlying reason? 

[Rob]

What was the initial condition in both cases? Did the solver converge every time step? 

[JIN LU]

The geometry was simply an aluminum alloy cubic box with length of sides 200mm, inside which was a phase change material with latent heat 100J/g. Heat flow of 1000W was applied to one of the box's six surfaces, lasting for 150s. The ambient temperature was set to 60℃.

Actually I don't know whether the solver converge or not. In the Steady-State Thermal module, the solving process can always be done and the solution gave an exact temperature contour result. But I did't find the residual curve of the solving process, which is very clear and apparent in Fluent module.

I understand that Fluent deals with fluid dynamics depicted by several partial differential equations, which makes it difficult to converge. In Fluent, very different results even under the same initial conditons is normal. But for steady-state thermal problem as my case, no eradiation, no convection, the solver only needs to deal with Fourier's heat conduction equation, non-convergence is rare.

Could you tell me how to judge whether the solver converge or not in steady-state thermal module?

 

[Rob]

Did you run this in Fluent or Mechanical? And if you're getting different results in Fluent for the same model something isn't set up correctly. 

[JIN LU]

I didn't run this in Fluent or Mechanical. But I always choose automatic meshing, if the meshing quality is poor, is it possible that getting different results for the same model?

[Rob]

If the mesh is poor, and there's no convergence, it is possible to get different results with a different mesh. The more physics you add into the model the larger the potential difference too.