Comparative analysis of detailed and reduced kinetic models for CH4 + H2 combustion. (15th June 2019)

• Record Type: Journal Article
• Authors: Li, Rui; He, Guoqiang; Qin, Fei; Pichler, Christoffer; Konnov, Alexander A.
• Is Part Of: Fuel. Volume 246(2019)
• Page Start: 244
• Page End: 258
• Publication Date: 2019-06-15
• Subjects: Chemical mechanism reduction -- Directed relation graph method -- Hydrogen enrichment -- Turbulent JHC flame -- CFD; Fuel -- Periodicals; Coal -- Periodicals; Coal; Fuel; Periodicals; 662.6
• DOI: 10.1016/j.fuel.2019.02.132
• Languages: English
• ISSNs: 0016-2361

Abstract

Highlights: Two reduced kinetic models for CH4 + H2 combustion. The models are validated over the range typical for gas turbines. Two models behave differently in CFD. The differences are traced back to important chemical reactions. Abstract: Directed relation graph with error propagation (DRGEP) method combined with extensive validation for 0D, 1D and 2D CFD modeling supported by sensitivity and Rate-Of-Production (ROP) analyses are implemented for comparative study of detailed and reduced kinetic mechanisms for CH4 + H2 combustion. To this end, two detailed kinetic mechanisms, namely AramcoMech 2.0 and recently updated Konnov mechanism, were validated using available measurements of ignition delay times and laminar burning velocities for hydrogen, methane and hydrogen + methane fuel mixtures. For all experimental conditions visited, both detailed mechanisms demonstrated good and close to each other performance. Two-stage DRGEP method and reaction reduction based on computational singular perturbation (CSP) were then implemented to achieve two skeletal models: 25 species and 105 reactions for AramcoMech 2.0 and 27 species and 107 reactions for the Konnov model. The conditions for skeletal models generation cover ɸ = 0.5–2.0, temperature 900–2000 K, and pressure 1–50 bar. Turbulent non-premixed flames of CH4 + H2 in the Jet in Hot Co-flow (JHC) burner for two different oxygen concentrations in a co-flow were modeled using both skeletal models. 2-D RANS simulations with OpenFOAM code of the flame structure using the two skeletal kinetic mechanisms are similar except for the mass fraction of OH and CO. To elucidate the differences between two skeletal mechanisms generated using the same reduction method, extensive validation for 0D, 1D and 2D CFD modeling were supported by sensitivity analysis for detailed and skeletal reaction models. Good agreement between the skeletal and detailed mechanisms was found in top reactions as well as their sensitivity coefficients, which affect auto-ignition process and laminar flame propagation. Further chemical and sensitivity analysis of the structure of laminar flames demonstrate that three important reactions, i.e. CO + OH = CO2 + H, H2 + OH = H + H2 O, and CH4 + OH = CH3 + H2 O have different rate constants in the Aramco and Konnov models that may contribute to the differences in the prediction of CO concentration profiles. The simulation predictions for CO concentrations are improved for laminar flames and JHC flame by using a 25-species modified version in which these rate constants were taken from the Konnov mechanism. It was noted that DRGEP method applied to different detailed kinetic schemes generate skeletal models with different, non-overlapping lists of retained species. The presence of CH2 CHO in the Aramco 25-species skeletal mechanism and its absence in the Konnov 27-species mechanism, and the presence of CH, CH2, CH2 CO in the latter and their absence in the former mechanism were analysed and explained using Rate-Of-Production analysis for conditions found in the CFD simulations.