I tested on 2D and 3D models and the response was the same: temperature field not symmetric. This sounds a bit strange since Hgen applied in a node should provide spherical (3D) or cylindrical (2D) distribution. This happens only for transient hgen load.
What kind of mesh do you have? How big is the asymmetric behaviour? Please, describe your case better.
The boundary conditions are adiabatic walls (default). I apply a thermal load in just one node of the model and the load is transient (a short heat impulse). I would expect same temperature at nodes equally distant from the source but this is not the case and I do not understand why. If the load is stationary (not an impulse but a heat generation lasting all the observation time), the distribution is symmetric. The asymmetry is not negligible (more than 10%).
Is asymmetry always in the same direction? I mean, on the same angular coordinate of your mesh you still have same (let's say) defect of temperature?
If you have an unstructured grid, I suggest to decrease the tolerance on the time step of your transient solver, and enhance max iteration number on time step.
Applying a thermal load on just one node could produce a locally numerical defect on effective thermal conductivity (it happens to me some time ago).
Dear Scarselli,
It is not clear for me if your subject is heat convection problem? Fluid flow related heat transfer may exhibit symmetry breaking bifurcation above a certain threshold condition because fluid flow is a non-linear process.
Asymmetry of the computational mesh can also be a source of the asymmetric solution.
Regards, Ji.
Dear Cho,
the mesh is structural and the conditions should be linear. I am investigating the transient load application to understand if this could cause this unexpected behaviour.
Regards
Gennaro
I used PLANE77 instead of PLANE55 and I found the expected symmetry!!!
Dear Gennaro,
it is a clear problem of numerical diffusifivity. It depends on the unavoidable truncationl error that enhances in a transient process for a numerical (non physical) diffusion , strictly depending on the order of approximation. In detail, such an effect is much higher when the element has a low order (this is the caseof plane55). On the contrary, when the element order increases such an effect is reduced ( as you observe for plane 77that is an high order generalization of the element plane55).
- Badges
- Science topic