Learning Forum

[lancer]

Hi, I'm running a sofc module.

This is some I think is an important to let you know,

Type: Pressure-Based

Time: Steady

Velocity Formulation: Absolute

Energy > ON

Viscous>Laminar

Species Transport > ON

Boundary Condition =  Co-Flow, Inlet, Temperature, Species

Under-Relaxation Factors

 

MESHING

Skewness

How do I solve the floating exception problem.Divergence detected in AMG solver: species-2 Error at host: floating point  Which part should I focus? The floating exception occur when reaches 10-20 Iteration.

[Aleksey Gerasimov]

Hello Lancer,

There might be a number of reasons:

1) You have a nice hexahedral mesh, but the orthogonal quality drops down to 0.29. I cannot see the mesh distribution in the 3rd direction. I presume that you might have cell expansion in Z direction. If the orthogonal quality suffers, then I'd suspect that the bias has not been applied consistently on all edges, thus creating skewed cells that aren't visible from the outside. Please double check.

2) It is worth having some flow-development sections at inlets and outlets. The inlet relaxation section allows flow development before it reaches the active zone of the cells and thus avoiding sharp gradients there. The oultet flow relaxation sections are needed to limit the effects of a potential reversed flow that can create very sharp and unrealistic temperature and species gradients there.

3) The gas diffusion layers are very different in thickness. It is not unusual for them to be different, but I have never seen such a difference. However, the most important aspect there is that the thinnest layer is only 1-cell thick. The first near-electrolyte plane of cells in porous electrode is very source dominant area, Please consider having at least 3 cells across it.

4) When it comes to solver-settings, please remember that for segregated solver the energy, species and potential URFs need to be 1 at the end of the simulation, However, at the initial stages, if convergence is problematic, one can reduce these values temporarily to, say, 0.9 and then gradually increase them to 0.95, 0.98, 0.99 and finally to 1.

5) You can try turn F-cycle for all equations.

6) Proper initialization and even a simulation of the flow in the cell without electrochemical reactions is a good practice.

7) In the initial stages of the simulation, you can temporarily turn off "Enable Volumetric Energy Source" & "Enable Surface Energy Source" model options in the SOFC panel.

8) Please consider running potetiostatic boundary conditions (prescribed voltage) instead of galvanostatic boundary conditions (prescribed current) and start with high voltage (low current) and gradually decrease voltage (increase current) in order to reconstruct the polarization curve. For lower voltage simulation, please consider using the preceeding simulation at higher voltage as an initial guess.

Hope this helps and best regards,

Aleksey.