Hybrid CRN-CFD model of a micro Gas Turbine combustor fueled with methane/hydrogen mixture for emission prediction

As renewable energy grows, storing excess power as green hydrogen is a promising solution for grid balanc-
ing. Micro gas turbines (mGTs) can be a viable option due to its several advantages, such as low maintenance
costs, lower noise, and possibility of fuel flexibility and lower emissions for using this hydrogen in decentralized
CHP systems, but achieving stable combustion with low NOx emissions remains a key challenge. Chemical
Reactor Networks (CRNs) offers a inexpensive modeling method to predict combustion behavior and emissions,
but their accuracy depends highly on the design method and is often limited by simplifying assumptions such
as reduced dimensionality, idealized flow assumptions, and the limited representation of complex physical and
chemical processes compared to detailed CFD or experiments. Therefore, there is a need for a rigorous hybrid
CFD-CRN approach to model a real mGT combustor that addresses the overlook impact of combined aerodynamic
and chemical features of the flow especially for hydrogen enriched fuel conditions. This study evaluates a quasi-
automatic CRN construction approach by defining different zones inside the combustor using aerodynamic and
chemical criteria applied to high-fidelity numerical data on a swirl-stabilized combustor of 20 kWth mGT fueled
with hydrogen/methane mixtures. The considered burner is a real-world industrial implemented in an mGT used
for small-scale domestic heating and electricity production. Therefore, we are limited in controlling inlet condi-
tions as the inlet parameters cannot be controlled independently of each other. High-fidelity RANS simulations
are performed for pure methane and a fuel mixture (10%volH2 added to CH4). After a 2-D domain selection, it is
clustered into zones with similar thermo-chemical features based on several flow and mixing quantities. The clus-
tered zones are used to construct CRN models to simulate the temperatures and emissions of the combustor. The
methodology, validated against experimental data and CFD, demonstrates a good prediction of NOx and captures
the main combustion regions and trends.