Learning Forum

[abtharpe42]

I made a post on here recently asking for help on setting up a journal file. I am trying to recreate the results from a paper that tested and simulated an nh3 microcombustor with LES. I'm using a different chemical mechanism from the paper, and my steady RANS simulations shows that the nh3 flame blows out in conditions where the paper's experiments and LES simulations showed stable flames. I need to figure out if running an LES simulation with my mechanism shows that I do in fact get a stable flame, proving this hypothesis that I have that steady RANS and LES simulations of the same conditions can produce wildly different combustion behavior.

The journal file I am assembling will read and run a pre-made steady state case file made with the Fluent GUI. This case uses the CHEMKIN-CFD solver because I've found it to be way more stable and reliable than the Stiff-Chemistry solver in turbulent combustion (persistent DASAC errors a lot of the time). The journal then swaps over to an LES transient setup using pure nh3 fuel to match the paper, and the simulation progresses from there for certain amount of total time. The steady flame is used as the starting point for the pure nh3 flame.

The issue that I am having is that I am now learning that the CHEMKIN-CFD solver is apparently really bad at transient simulations for some reason. After a few iterations in even the first time step (1e-6 sec), the residuals explode hard and never come back down, leading to a crash. After some testing on the side with a simpler case, I've found that the CHEMKIN-CFD solver diverges hard when any sort of ignition event occurs, but now the Stiff Chemistry solver takes the crown for maintaining really good stability.

Now I have to alter my journal file to switch to the Stiff Chemistry solver before running the transient case. Why is the CHEMKIN-CFD so aggressively unstable while the Stiff Chemistry solver is stable in transient combustion simulations while in steady state it's vice versa, albeit a bit less aggressive with the instability?

[Richie Mathew]

Hello,

If you are using EDC or the Finite Rate model, I would suggest starting with the Eddy Dissipation model to obtain an initial flame and then switching to the respective model. Also, initially use the mechanism referenced in the paper. If you are still experiencing blow-off, then the solver setup needs to be checked. However, if you obtain the desired flame, then the mechanism should be reviewed.

Chemkin-CFD has the advantage of converging faster compared to the Stiff Chemistry Solver. However, the solution may initially diverge but can eventually converge. The Stiff Solver is inherently stable but slower to converge. If the Chemkin-CFD Solver is diverging, try changing the default mixture properties (Cp, thermal conductivity, and viscosity) to the mixing law, and set (mass and thermal diffusion) to kinetic theory.

You can find more information in the documentation manual (Chemkin-CFD API). You can also observe from the residuals that, in the Stiff Chemistry Solver, the residuals for products do not appear at the beginning; they show up as the reaction progresses. In contrast, in the Chemkin-CFD Solver, all residuals appear at once and then stabilize, leading to faster convergence.

Additionally, consider using the FGM combustion model. It is very useful for LES simulations and performs better in predicting NOx emissions compared to the Steady Flamelet Diffusion model.