[CFD-POST] How are gradients calculated?
I'm having trouble understanding how gradients are obtained by CFD-Post. I am running an adimensional problem in Fluent and I want to calculate the Nusselt number in CFD-Post. However, whenever I probe the temperature gradient on the pipe wall, Fourier's law is never obeyed. In other words:
probe(Wall Heat Flux)@wallPoint ≃ 1 (which is what is set up in Fluent, so it's OK), but -probe(Thermal Conductivity)@wallPoint*probe(Temperature.Gradient )@wallPoint ≃ 0.719, which contradicts Fourier's law. It's an axisymmetric problem (simple circular duct), so Temperature.Gradient Y ≃ Temperature.Gradient, but Fourier's law is not obeyed. |
CFD-Post uses shape functions to calculate gradients. They are detailed in the CFX documentation.
Fluent stores the results in the mesh elements/cells, whereas CFX stores them in the mesh nodes. So when CFD-Post read a Fluent result, it interpolates the cell values to nodes, and all calculations are then node-based (including the gradients based on shape functions). That is why you see a difference. Fluent, on the other hand, calculates them internally, but the temperature gradient calculated by Fluent is not available to CFD-Post by default. You should always trust the solver computed values (Wall Heat Flux, Wall Shear, etc), not the one you manually computed. Cheers. |
All times are GMT -4. The time now is 12:43. |