Learning Forum

[m.d.lasica]

I'm currently simulating a steady combustion chamber for a hydrogen-powered gas turbine (a small section of it), using non-premixed burners where hydrogen and air are injected separately. For species transport, I'm using the EDC model with the CHEMKIN solver and imported reaction mechanisms from DRM, along with thermo and transport data from GRI3.0.

The simulation domain is relatively small: 11.8 mm width x 17.5 mm height x 210 mm length, with a global mesh size of 0.32 mm (5 mln cells). There are two bodies of influence in the mixing areas, with local mesh sizes of 0.08 mm and 0.16 mm. Mesh has a minimum orthogonal quality of around 0.4 and a maximum aspect ratio of about 30 (due to the inflation layer).

When running simulations with relatively low injection velocities for air and hydrogen, everything converges nicely. However, when I increase the injection velocities, I start encountering convergence issues. The continuity residual doesn’t drop below 5e-3, despite multiple adjustments.

Here's what I've tried so far:

Switching to 1st order schemes.
Using adaptive mesh refinement.
Lowering URFs gradually—starting from the default values, then reducing to 0.25 for all, and even down to 0.1.
Lowering the URFs seems to help but only marginally. The lowest continuity residual I've reached is 1.8e-3. I also track the average outlet temp and NOx and they do not vary much, from 5e-3 to 1.8e-3 they remain almost constant. The areas of highest mass imbalance are within the flame and at the flame/air interface.

Has anyone experienced similar issues or have suggestions on what might be causing the convergence problem at higher velocities (hydrogen injection velocity ~ 200 m/s)? Any tips would be greatly appreciated!

[Essence]

Hello,

I would suggest you to first use EDM with RTE and converge it. Then move on to EDC since it is a bit heavy model. Select High Order Relaxation term and select Warped Face correction. Also please use psuedo time step. Lastly, please do not judge the convergence solely based on the scaled residuals. Check the net mass and heat flow rate. Also check for the report definitions.

[m.d.lasica]

Hello, 
Thank you for your answer. I will do as you say.