Sensitivity analysis of CFD-chemical reactor network approach with reactor numbers for Sandia flames D and E

• CFD-CRN Sensitivity Analysis : Evaluates the effect of varying reactor counts (5, 50, 250, and 1250) in a coupled Computational Fluid Dynamics (CFD) and Chemical Reactor Network (CRN) approach, achieving mean NOx formation rate deviations of less than 10% for 1250 reactors compared to CFD predictions. • Trade-off Between Accuracy and Efficiency : Demonstrates that higher CRN resolution (1250 reactors) enhances thermal field and species concentration accuracy for Sandia flames D and E, with a 75% reduction in computational cost compared to fully resolved CFD simulations. • Application to Sandia Flames D and E : Employs Code_Saturne 8.2 with Reynolds Stress Model (RSM) and Cantera 3.0 with the GRI-Mech 3.0 CH 4 mechanism to predict species concentrations (CO 2 , CH 4 , H 2 O, N 2 , NO, O 2 ), NOx formation pathways, and thermal profiles, validated against experimental data. • Optimized Combustion Modeling : Establishes the CFD-CRN approach as a computationally efficient method for large-scale combustion studies, achieving prediction errors below 5% for species and NOx with 1250 reactors, suitable for industrial applications. Large-scale combustion is strongly influenced by turbulence–chemistry interactions, but resolving these processes with detailed chemistry in CFD is computationally expensive. The chemical reactor network (CRN) approach addresses this limitation by capturing detailed kinetics at reduced cost. In this study, CFD simulations were conducted using Code_Saturne 8.2, and results were post-processed in Cantera 3.0 to solve chemical kinetics using the GRI-Mech 3.0 CH 4 mechanism (53 species, 325 reactions). We investigated Sandia flames D and E by evaluating temperature fields, species concentrations, and NOx formation pathways, including thermal, prompt, N 2 O intermediate, and reburning mechanisms. A quantitative sensitivity analysis shows that a CRN with 1250 reactors closely matches CFD predictions, with NOx formation rates differing by less than 10%. In contrast, low-resolution CRNs (5 and 50 reactors) underestimate emissions by over 50%, with reburning pathway deviations reaching 80%. High-resolution CRNs better capture localised variations in species and temperature, while coarser networks result in deviations of up to 60%. Notably, the 1250-reactor CRN reduces computational cost by 75% while maintaining NOx prediction errors under 5%. Unlike prior studies, this work integrates detailed chemistry with RSM turbulence and validation against both CFD and experimental data. It provides the first systematic sensitivity analysis of reactor number in CFD–CRN coupling for Sandia flames D and E, showing that high-resolution CRNs deliver accurate NOx predictions while reducing computational cost by over 70%.