[Daveo643]

Hello everyone,

I have searched solutions to this problem and have not found anything to help, so I hope someone would be able to address this issue anew.

I'm doing a 2D transient simulation of hydrogen injected into a chamber. The injector is modeled simply as 2 jets with a pressure inlet boundary condition. Air simultaneously enters through a separate sets of channels also with profile-defined pressure inlet boundary condition that mimics an in-cylinder compression following a polytropic.

Here is a picture of the setup.



My problem is that a some timestep point of the simulation during injection, there is a spike in the residuals - with no good physical reason occurring at that moment - and after that, I get messages "Minimum PDF enthalpy exceeded in xxx cells" and my temperature monitor shows a zone where the temperature has dipped to 173 K - an impossible phenomenon. This zone at the minimum enthalpy eventually spreads until almost the entire area of the modeled chamber.

Strange thing is, on a very similar geometry but with identical boundary and initial conditions, I don't have this problem at all!!

Here are some settings and other info that might be germane to the problem:
- partially-premixed combustion model
- Automated Grid Refinement enabled
- There is an over-expanded injection jet of hydrogen using a pressure inlet BC from 10 bar to ~1 bar, thus the PDF minimum temperature is such that during injection, this remains below the calculated static temperature.
- Coupled P-V solver
- Body force-weighted pressure discretization; second-order upwind for everything else; high-order term relaxation enabled. Second-order blending ( solve > set > numerics = 0.75)
- F-cycle multigrid with BCGSTAB stabilization for all variables.

Based on search results for solutions, I have tried:
- Gradually reducing underrelaxation as the simulation runs
- define > models > species > partially-premixed-combustion-expert > Maximum Heat Loss Fraction [0.667] = 0.333
- One suggestion is to change the polynomial coefficients for the Thermal Conductivity and Viscosity, but I have no idea where the suggested values come from and I'd rather not change these from the defaults if they work for other cases.

Any insights and assistance would be greatly appreciated!!

[Daveo643]

FYI: Fixed with remeshing. There was an area poor mesh quality (that wasn't spotted because overall mesh quality statistics were excellent) where the "minimum PDF enthalpy exceeded" warning initiates and the error appears to propagate into the zone.