To achieve sustainable combustion in scramjets, struts or cavities are typically introduced as flame holders, but they also contribute to complex combustion instabilities. This study employs a hybrid Large Eddy Simulation (LES) and Reynolds-Averaged Navier-Stokes (RANS) technique to investigate the German Aerospace Center’s (DLR) strut-based supersonic combustor, providing detailed insights into the combustion processes with a focus on the near-field flame development and overall combustion oscillations within the combustor. The high-resolution simulation reveals that two pairs of recirculation zones emerge behind the strut. The expansion of the main recirculation zone induces a forward movement of the high-temperature region, triggering auto-ignition and causing a flame reverse movement phenomenon. The interaction between turbulence and chemical reactions results in two types of reverse flame movement in the near-field region, collectively forming a complex flame-switching behavior to induce upstream combustion oscillations. The pressure oscillation frequencies in the near- and far-field regions are essentially identical and exhibit strong cross-correlation characteristics, suggesting a potential influence of near-field combustion oscillations on the far-field combustion oscillations.

The principal component transport model has been successfully used for large eddy simulations (LES) of turbulent flames. However, recent experimental–computational studies of KAUST piloted NH3/H2/N2 flames revealed large discrepancies in flame height, mixture-fraction decay, and local extinction, which are attributed to the errors in presumed subgrid-scale (SGS) probability density function (PDF) closures that do not fully incorporate multi-component transport and differential diffusion effects. To address this issue, the present study employs the construction of the principal component (PC) scores with Varimax rotation to simplify the closure of the transport terms and to enhance the statistical independence among different PC scores in the PDF closure models. An a priori analysis using direct numerical simulation (DNS) data of a planar NH3/H2/N2-air flame at a Reynolds number of 36,000 demonstrates that adaptive local Varimax-rotated PCs achieve the lowest statistical dependence and highest reconstruction accuracy compared with global Varimax-rotated PCs and conventional mixture fraction and progress variable variables (Z,c). Subsequent a posteriori LES of the KAUST NH3/H2/N2 flame using the global Varimax-rotated PCs framework with SGS closures demonstrated accurate prediction of mixing, temperature, and major species, and suggests that the local Varimax-rotated PCs representation further improves minor-species predictions, enhancing the fidelity of the PCA-based model as an efficient and reliable predictive tool for ammonia–hydrogen combustion simulations.Novelty and significance statement: This work presents the first DNS of a partially premixed cracked ammonia–air turbulent flame at a high Reynolds number (36,000) with detailed chemistry and transport, which is used for an a priori analysis to improve LES subgrid models. The analysis demonstrates that adaptive local principal component analysis (PCA) with Varimax rotation enhances statistical independence and improves manifold reconstruction relative to conventional Z–c formulations and global PCA approaches. The resulting data-driven framework enables more robust presumed SGS PDF closures without relying on predefined controlling variables. At high Reynolds numbers, accurate SGS modeling is essential for capturing local extinction phenomena associated with combustion instability and engine failure. By addressing modeling limitations identified in recent joint experimental–numerical investigations of KAUST piloted NH3/H2/N2 flames, this work establishes a unique benchmark dataset and a data-driven pathway for improving reduced-order models and predictive simulations of ammonia–hydrogen combustion.

One of the challenges in using hydrogen to replace fossil fuels in industrial high-temperature processes is the reduction in thermal radiation due to the absence of CO2 and soot particles. Blending hydrocarbons with hydrogen offers a practical solution. In this study, high-fidelity numerical simulations of pure hydrogen flames and hydrogen/hydrocarbon blended flames were conducted to investigate radiation characteristics under different blend ratios, with a pure methane flame used as a reference. Detailed chemical kinetics were incorporated, and soot evolution was modelled using the method of moments with polycyclic aromatic hydrocarbons (PAH)-based soot kinetics models. Radiation heat loss from H2O, CO2, and soot particles was calculated using finite angle method combined with the full-spectrum correlated k-distribution radiative property model. The predicted temperature and radiation fractions showed good agreement with experimental measurements. Results indicate that, the radiation heat loss fraction of the pure H2 flame is only one-third that of the pure CH4 flame. The addition of 20 % C3H8 increases the flame radiation fraction from 3.2 % to 7.0 %, reaching approximately 75 % of the CH4 flame’s value, primarily due to the combined effects of CO2 and soot, with CO2 as the dominant contributor. Further increasing the C3H8 ratio to 35 % raises the radiation fraction from 7.0 % to 8.0 %, mainly due to increased soot, as the enhancement of CO2 radiation is offset by a reduction in H2O radiation. Although the peak soot volume fraction reaches 1.89 ppm in 35 % C3H8-blended flame, its distribution is limited with the flame interior. This limited spatial distribution restricts the contribution of soot to the overall radiative heat transfer.

Soot formation in methane reforming reactors can deactivate catalysts and obstruct flow. This study applies an inverse diffusion flame (IDF), representative of such reactors, to decouple chemical and physical effects on soot size evolution. Combined light extinction and angular light scattering with numerical simulations were applied to quantify the soot volume fraction (fv), primary particle diameter (dp), number density (np), and surface area per unit volume (S). Physical effects were examined by varying pressure (3–5 bar) in nitrogen (N2) diluted flames, while chemical effects were assessed by comparing N2 and carbon dioxide (CO2) dilution at 5 bar. fv increases with pressure, and the soot region moves toward the flame centerline with both height and pressure. CO2 dilution suppresses fv by ∼ 62% and reduces S by ∼ 40% relative to N2 at 5 bar, indicating restricted surface growth. dp increases with both pressure and height, reaching ∼30 nm. A pronounced rise in dp for the N2-diluted flame at 4 bar indicates a possible transition in pressure-dependent growth. Simultaneous dp increase and np decrease suggests particle coalescence, whereas larger dp values at higher pressures confirm enhanced surface growth. Numerical simulations reproduce the experimental trends despite overpredicting dp. While predicted temperature differences at the peak fv locations remain small, CO2 dilution decreases calculated acetylene (C2H2) concentration, and reduces peak H radical concentration by >50%, identifying chemical effects as the dominant soot-suppression mechanism. At elevated pressure, the dominance of the Hydrogen-Abstraction-Carbon-Addition (HACA) mechanism relative to the adsorption of polycyclic aromatic hydrocarbons (PAHs) further enhances the impact of H-radical concentration. These results offer insights into soot size evolution under different conditions in IDFs and enable validation of soot models based on a hybrid method of moments in methane-reforming applications.Novelty and significance statement This study presents the first in situ characterization of soot size evolution in methane inverse diffusion flames at elevated pressures under CO2 and N2 dilution, conditions relevant to methane reforming. By combining light extinction and angular scattering with numerical predictions of temperature, chemical species, and soot growth rates, this study decouples chemical (diluent-driven) and physical (pressure-driven) effects on soot volume fraction and primary particle size. A transition in particle growth is identified at 4 bar. The competing effects of CO2 dilution and pressure provide insights into their roles in soot formation and growth. Pressure enhances surface growth, whereas increasing height promotes soot coalescence. CO2 dilution reduces H and C2H2 concentrations, suppressing HACA-driven surface growth and decreasing soot volume fraction (by 62%) and particle size. These fundamental insights provide guidance for soot mitigation strategies in industrial methane reforming reactors.

Co-firing ammonia with hydrocarbon fuels is a promising strategy for achieving partial decarbonization while maintaining compatibility with existing combustion facilities. In this study, flame stability, NO emissions, and flame characteristics of CH4/NH3 non-premixed flames were investigated under oxygen-enriched O2/CO2 environments. The NH3 blend ratio ranged from 0 % to 100 %, and the O2 mole fraction in the O2/CO2 mixture varied from 30 % to 70 %. The effects of NH3 blend and O2 mole fraction on OH and NO distributions were visualized using Planar Laser-Induced Fluorescence (PLIF). High-fidelity simulations incorporating detailed reaction chemistry were performed and validated against experimental measurements. The measured flame stability diagrams indicate that a higher O2 mole fraction is required to sustain a stable flame at higher NH3 blend ratios. Flame lengths decrease linearly with increasing NH3 content and O2 mole fraction. The peak NO concentration occurs at the flame base, where the net production rate of NO is an order of magnitude higher than that along the centerline. The conversion ratio of fuel-bound nitrogen to NOx decreases significantly with increasing NH3 content. When the NH3 blend ratio increases from 51 % to 66 %, NO emissions rise by only 2.5 %. Moreover, compared to O2/N2 environments, NO emissions are reduced by 26 % in O2/CO2 environments, accompanied by a shift in the dominant NO formation pathway from NH-* N-* NO to NH-* HNO-* NO.

Radiative heat transfer is important in NH3-H2 combustion, yet the contribution of NH3 radiation remains unclear. This study quantitatively evaluates the effect of NH3 radiation in laminar premixed and non-premixed NH3-H2 flames using high-fidelity simulations with and without NH3 radiation. Detailed chemistry, the finiteangle method, and a full-spectrum k-distribution model are employed to ensure the accuracy. The results show that, in the non-premixed flame under study, NH3 radiation reduces the flame temperature by up to 8 K, which is much lower than the 53 K reduction caused by H2O radiation. Radiation of H2O and NH3 reduces the peak OH mole fraction along the flame centerline by 12% and 0.5%, respectively. In contrast, in the premixed flame under study, NH3 radiation leads to a slight temperature increase of less than 1 K. Compared with H2O radiation, the effect of NH3 radiation is minor in the laminar premixed flame considered in this study. Under similar flame size and fuel/oxidizer conditions, the effect of NH3 radiation is stronger in laminar non-premixed flames than in laminar premixed flames.

Inverse diffusion flame (IDF) configuration is widely employed in high-temperature industrial processes. While oxy-fuel combustion is considered a promising technology for CO2 capture, the high CO2 concentrations involved can significantly suppress soot formation and potentially reduce radiative heat transfer. In this study, ethylene IDFs with varying levels of CO2 dilution in the oxidizer are investigated through high-fidelity numerical simulations. Detailed soot kinetic models and non-gray radiative property models are employed to ensure close agreement with experimental measurements, including flame temperature, flame height, and soot volume fraction. A fictitious species strategy is employed to isolate the thermal, radiative, transport, and chemical effects of CO2 on soot formation. Additionally, the individual radiative contributions of CO2, H2O, CO, and soot particles are quantitatively evaluated. Results reveal that the thermal and chemical effects of CO2 are the most significant in suppressing soot formation, primarily by lowering flame temperature and reducing the concentration of soot-forming species. The chemical effect is dominant at a 50% dilution level, while the thermal effect becomes more important at 70%. The transport effect of CO2 primarily leads to an increase in flame height, but has a negligible impact on peak soot volume fraction. Moreover, the overall radiative capability of CO2-diluted flames is consistently higher than that of N2-diluted flames at equivalent dilution levels, with the difference becoming more pronounced at higher dilution levels. In N2-diluted flames, soot radiation dominates but decreases sharply with increasing dilution. In contrast, CO2-diluted flames exhibit dominant CO2 radiation, which remains largely unaffected by further dilution.

Partial cracking of ammonia (NH3) into hydrogen (H2) and nitrogen (N2) enhances flame stability in carbon-free combustion by increasing fuel reactivity. However, modeling turbulent jet flames fueled by such multicomponent mixtures remains challenging due to complex turbulence-chemistry interactions (TCI), localized extinction, and the role of differential diffusion. This study develops a reduced-order large eddy simulation (LES) framework that couples principal component (PC) transport with deep neural networks (DNNs), incorporating differential diffusion and subgrid-scale (SGS) closures for efficient and accurate simulation of NH3/H2/N2-air jet flames. A mixture-averaged (MA) transport model is formulated using a Varimax-rotated PC basis, resulting in a diagonally dominant diffusion matrix that enables species-specific transport. SGS closures for PC variance and scalar fluxes are integrated to improve representation of unresolved turbulence. The framework is validated against experimental data for two piloted jet flames from the KAUST NH3 flame series, referred to as flames D and F, with Reynolds numbers of 24,000 and 36,000, corresponding to 59% and 89% of the blowoff threshold, respectively. The proposed MA-Varimax-SG model captures key features including mixture fraction decay, temperature profiles, and the distributions of major and intermediate species (e.g., OH, NH2), while accurately predicting localized flame holes and extinction-reignition dynamics. Compared to the PC-DNN model with unity Lewis number, the proposed approach improves extinction prediction in highly strained regions with less than 10% additional computational cost. These findings highlight the critical role of differential diffusion and SGS modeling in partially cracked ammonia flame simulations and demonstrate the scalability and robustness of the proposed framework for complex combustion systems.

Partial cracking of ammonia presents a practical solution to the challenges posed by the low laminar burning velocity and high autoignition energy of pure ammonia. On the other hand, the presence of hydrogen in the fuel can induce significant differential mass diffusion in turbulent non-premixed flames. In this study, a direct numerical simulation (DNS) of a temporally evolving planar jet NH3/H2/N2-air non-premixed flame was conducted to investigate the characteristics of differential diffusion in partially cracked ammonia flames. A recently proposed extended mixture fraction definition was employed to consider the effect of nitrogen element. The results indicate that the conditional mean temperature and mass fractions of major species consistently fall between those predicted by the two laminar flamelets, calculated using the mixture-averaged diffusion model and the unity Lewis number. Differential diffusion was quantitatively characterized by the difference in the mixture fractions of hydrogen and nitrogen elements. In the turbulent non-premixed flame with a high Reynolds number of 23,000, the extent of differential diffusion still reaches 60% of that observed in laminar condition. Moreover, a spatial filtering analysis in the context of large eddy simulation (LES) confirmed that differential diffusion is not reflected in the subgrid species flux, as it arises from the convection term rather than the diffusion term.

Blending ammonia with hydrogen and partially cracking ammonia are promising strategies to enhance the combustion performance of ammonia. Accurate prediction of ammonia/hydrogen blend combustion behavior requires careful consideration of differential diffusion effects associated with hydrogen. In this study, differential diffusion effects are assessed considering various flamelet-based modeling approaches: the unity Lewis number flamelet/progress variable (ULF) model, variable Lewis number flamelet/progress variable (VLF) model, and species-weighted flamelet/progress variable (SWF) model. The latter one incorporates weighting between two flamelet datasets based on the unity Lewis number assumption and the mixture-averaged diffusion models. An a priori analysis based on direct numerical simulation (DNS) data and an a posteriori analysis involving large eddy simulation (LES) of turbulent partially premixed NH3/H2/N2-air jet flame are conducted. The analysis confirms the presence of strong differential diffusion in turbulent partially premixed NH3/H2/N2-air flames and demonstrates the feasibility of weighted flamelet models for properly capturing the differential diffusion effects in different levels of turbulence. Moreover, due to the longer chemical time of NO formation on the fuel-lean side, adding the NO mass fraction in the definition of the progress variable effectively improves NO predictions. In the LES simulations, it is found that the SWF model performs well by considering both turbulent diffusion and molecular diffusion. The predictions of the SWF model fall between those of the ULF and VLF models and align closely with the measurements on fuel-rich side, indicating the significant roles of both turbulent diffusion and molecular diffusion in these regions. Novelty and Significance Statement: This study innovates by conducting comprehensive a priori and a posteriori analyses on modeling differential mass diffusion using flamelet-based models. The a priori analysis based on the DNS data confirms the strong differential diffusion in turbulent partially premixed NH3/H2/N2-air flames, as well as the feasibility of using weighted-based flamelet models for the modeling of differential diffusion. LES simulations further reveal that differential diffusion is more pronounced on the fuel-rich side than on the fuel-lean side.

This study presents the integration and application of a generalized soot model within a pressure-based, variable-density Navier-Stokes solver in OpenFOAM for multidimensional large eddy simulations of turbulent, pressurized sooting flames. The model, previously validated in laminar configurations with enhanced sensitivity to strain rate and pressure, is further evaluated against experimental data encompassing a broad range of operating pressures and turbulence intensities. Combustion closure is achieved via a radiative flamelet-progress-variable approach with a dedicated flamelet generation framework. The generalization of the soot model primarily focuses on soot inception and oxidation processes, allowing improved representation of the complex coupling between turbulence, pressure, and soot chemistry. Model predictions are assessed in terms of spatial distributions of OH, polycyclic aromatic hydrocarbons (PAHs), and soot volume fraction (SVF), with a focus on the effects of pressure and turbulence. The results demonstrate strong predictive capabilities across varying thermodynamic and flow regimes. At atmospheric pressure, the limitations of the flamelet approach in representing multiple characteristic chemical timescales are more evident, due to strong turbulence-chemistry interaction effects. In contrast, at elevated pressures (3 and 5 bar), the model provides more accurate predictions of SVF profiles. The pressure scaling behavior is captured with reasonable accuracy, offering improvements over conventional soot modeling approaches. A comparative analysis suggests that the proposed approach, featuring tabulated PAHs mass fractions and heterogeneous collision based inception, more effectively represents key soot formation characteristics than a transported lumped PAH model relying on homogeneous dimerization. An in-depth analysis of soot source terms shows a transition from turbulence-dominated to chemistry-controlled soot formation mechanisms as pressure increases.Copyright (c) 2025 American Association for Aerosol Research

Large eddy simulation / filtered mass density function (LES-FMDF) is applied for the first time to a supersonic reacting mixing layer to assess its capability and superiority for supersonic combustion. LES-FMDF uses a hybrid Eulerian-Lagrangian approach for the solution of flow parameters and chemical reaction source terms, respectively. To deal with the strong compressibility effects in supersonic flows, a conservative high-speed source term model proposed in our previous work is adopted. First, in the non-reacting flow field, temperature results align well with those from direct numerical simulation (DNS), validating the accuracy of the current high-speed source term model. The FMDF method also yields passive scalar variance results that closely match DNS, while traditional LES overestimates this variance by 15 % and 45 %. Subsequently, in the turbulent reacting flow field, comparisons with DNS and the well-stirred reactor (WSR) model highlight the superiority of the Lagrangian particle approach in reacting simulations. This advantage is particularly evident in the simulation of critical reactive radicals such as OH and HO2, which are key indicators of autoignition processes. The mean and variance predictions of OH and HO2 by FMDF show significant improvements over WSR, accurately capturing the location of autoignition. Finally, the capability of LES-FMDF to capture autoignition behavior in supersonic flows is further evaluated. Through scatter plots of reactants and products, it is demonstrated that the compressible FMDF approach effectively predicts the complex interactions between turbulence and chemical reactions in supersonic combustion.

Inverse diffusion flame (IDF) configuration, where the oxidizer is surrounded by fuel, is commonly used in reforming of hydrocarbon fuels for hydrogen production through the autothermal reforming and partial oxidation processes. Understanding the mechanism of soot formation in IDF is crucial for achieving efficient and environmentally friendly hydrogen production. In this study, high-fidelity numerical simulations were conducted to investigate the effects of pressure and gravity on the soot formation in a laminar IDF configuration at pressures up to 20 bar. The chemical kinetic models with detailed polycyclic aromatic hydrocarbons (PAH) pathways and an empirical reactive soot inception model are employed. The simulation results agreed well with experimental measurements, showing consistency in flame height, PAH concentration, and soot volume fraction. The simulations accurately reproduced the spatial distributions of PAHs and soot in the IDF, and quantitatively captured the linear increase in peak soot volume fraction with pressure. The linear relationship is mainly attributed to the linear increase in the PAH concentration, driven by changes in density due to pressure increase. Moreover, compared to zero gravity condition, higher flame temperature and radical concentrations were observed in normal gravity, leading to higher soot formation rates. However, buoyancy accelerates fluid movement, reducing residence time and ultimately suppressing soot formation in normal gravity conditions.

In this study, high-fidelity simulations are conducted on a 35 MW pulverized coal oxy-fuel boiler to investigate NOx formation characteristics under staged oxy-fuel combustion, including both fuel-staged and oxygen-fuel two-way staged modes. Detailed volatile components are considered by using the chemical percolation devolatili-zation model. A skeletal reaction mechanisms consisting of 35 species developed for fuel-NOx formation under oxy-fuel conditions are employed for the gas phase combustion and NOx modeling. The radiative property models are also optimized for oxy-fuel combustion. The results show that in the fuel-staged mode, a reduction atmosphere is established in the reburning zone to reduce the NOx concentration in the flue gas. A reduction of 5.6% in the average NO concentration at the furnace outlet is observed relative to the baseline oxy-fuel case. Moreover, the application of oxygen-fuel two-way staged combustion leads to a further reduction in NOx for-mation. The average NO concentration at the furnace outlet is decreased by 17% compared to the baseline oxy-fuel case. Pathway analysis indicates a reduction in the NO formation pathway involving HCN-* NCO-* NO, while the NO reduction pathway of NO-* HCN is enhanced.

To improve the NO modelling in turbulent flames, the flamelet/progress variable (FPV) model is extended by introducing NO mass fraction into the progress variable and incorporating an additional NO transport equation. Two sets of flamelet databases are tabulated with progress variables based on major species and NO mass fraction, respectively. The former is used for the acquisition of the main thermochemical variables, while the latter is employed for NO modelling. Moreover, an additional transport equation is solved to obtain the NO mass fraction, with the source term corrected using the scale similarity method. Model assessments are first conducted on laminar counterflow diffusion flames to identify lookup-related errors and assess the suitability of progress variable definitions. The results show that the progress variables based on major species and NO could correctly describe the main thermochemical quantities and NO-related variables, respectively. Subsequently, the model is applied to the large eddy simulation (LES) of Sandia flames. The results indicate that the extended FPV model improves the NO prediction, with a mean error for NO prediction at 55%, significantly lower than those of existing FPV models (130% and 385%). The LES with the extended FPV model quantitatively captures NO suppression in the mid-range of Reynolds numbers from 22 400 (Flame D) to 33 600 (Flame E), but underestimates the NO suppression at higher Reynolds numbers from 33 600 to 44 800 (Flame F). This underprediction is primarily attributed to the underestimation of local extinction levels in flames with high Reynolds numbers.

Cracking NH3 3 into a mixture of NH3/H2/N2 3 /H 2 /N 2 effectively addresses its low reactivity when used as a fuel. In this study, large eddy simulation (LES) and experiments are conducted for non-premixed NH3/H2/N2-air 3 /H 2 /N 2-air jet flames with simulated cracking ratios of 14 % and 28 % at an elevated pressure of 5 bar. Detailed experimental data on the flow field are provided for assessing turbulence models. A recently proposed species-weighted flamelet/ progress variable (SWF) model considering differential diffusion is adapted for simulating cracked NH3 3 flames. The effects of differential diffusion on scalar structures are further analyzed. The SWF model, incorporating differential diffusion, achieves good agreement with experiments in predicting velocity, mean temperature, major species mass fractions, differential diffusion parameters, and NO formation. The model also qualitatively captures variations in localized extinction along the flame height. In contrast, the unity Lewis number flamelet/ progress variable (ULF) model predicts the occurrence of localized extinction too upstream, significantly overestimating its occurrence in the near field where differential diffusion is significant. Moreover, NO formation in the cracked NH3 3 flame is well predicted by the SWF model using flamelet tables, and slightly overpredicted by the ULF model without differential diffusion.

The co-firing of ammonia (NH3) with hydrocarbon fuels is an effective strategy for reducing CO2 emissions and addressing the inherent low reactivity of NH3. However, soot formation presents a potential challenge in this context. In this study, experimental and numerical investigations are conducted to understand the chemical effect of NH3 on soot formation. Measurements of polycyclic aromatic hydrocarbons (PAHs) and soot volume fraction are performed in coflow diffusion flames with varying NH3 blending ratios (0-25 %), and this data was compared against numerical simulations. To improve the prediction of reduced PAH with NH3 substitution, an additional reaction of C3H3 and HCN is incorporated into the gas-phase mechanism. Moreover, a recently proposed reactive soot inception model is employed and the soot surface growth model is improved to account for the blocking of active sites on the soot surface by NH3 decomposition products (such as NH2). Simulations were conducted for counterflow and coflow diffusion flames. The results demonstrate that NH3 substitution leads to a decrease in H radical and an increase in H2. The improved models, which include the action of the H radical and nitrogen-containing species, provide better predictions of how soot volume fraction and mean particle diameter change with NH3 substitution, demonstrating the importance of C-N chemistry in both gas-phase and solid-gas reactions on PAHs and soot formation. Future work is needed to develop a more comprehensive C-N chemical pathways for predictions of co-firing of ammonia with hydrocarbon fuels.