Accurate kinetic modeling of hydrocarbon–NOx interactions is critical for predicting emissions and reactivity in exhaust gas recirculation (EGR) and reburning-based NOx-reduction systems. Ethane (C2H6) is the second-most abundant component in natural gas. Its coupling with NOx is therefore relevant to practical natural gas combustion systems. This study presents the first shock-tube investigation of C2H6 with nitric oxide (NO) in the absence of O2, covering fuel-lean to fuel-rich conditions. Experiments were performed near-atmospheric pressure over a temperature range of 1700–2800 K. Time-resolved CO mole fractions were measured using a quantum cascade laser near 4.85 µm. The measurements were used to evaluate recent detailed kinetic mechanisms incorporating NOx and C–N sub-mechanisms. Systematic comparisons of the measured data revealed significant discrepancies between the experiments and predictions of the mechanisms, especially under fuel-lean and low-temperature conditions. CO and OH rate-of-production analyses and pathway analyses show that CO is formed mainly through HCO decomposition at early times and through an HNCO-mediated route at later times, with a smaller direct contribution from C2H+NO. Sensitivity analyses identified reactions involving CH3+NO as key reactions controlling CO formation. Updating the rate coefficients of CH3+NO⇄H2CN+OH and CH3+NO⇄HCN+H2O, using literature values, yielded improved agreement with the measured CO across all equivalence ratios. This study indicates that uncertainty in CH3+NO reaction channels can still limit predictive capability for C2H6/NO interactions. Overall, this work provides the first CO benchmark dataset for the C2H6/NO system and identifies the CH3+NO channels as a limiting source of uncertainty in current mechanisms, supporting future refinement of hydrocarbon/NOx kinetic mechanisms.
The early-stage reaction evolution of confined explosives under non-shock initiation is critically influenced by the gap between the charge and the confinement structure. However, the quantitative experimental study and understanding remain limited. This study, for the first time under macro-scale engineering assembly gap conditions, experimentally quantifies the effects of gap configurations on convective combustion during the early non-shock initiation of pressed PBX charges under long thick-walled cylinder confinement. Three distinct gap configurations are examined: normal-gap, joint-coating, and radial-sealing. Various synchronized diagnostics including high-speed imaging, photonic Doppler velocimetry (PDV), pressure sensors, and electric probes were used to obtain a comprehensive analysis and comparison. Two key novel findings are reported. First, it is demonstrated that convective combustion through the charge-shell gap dominates the initial reaction propagation. A distinct two-stage pressurization process was identified in both unsealed configurations: an initial slow pressurization stage followed by a fast pressurization stage. Crucially, the duration of the slow stage shortens progressively along the axial direction, which is a direct manifestation of accelerating convective flame spread through the charge-shell gap. In contrast, the radial-sealing condition eliminates this slow pressurization stage, resulting in a direct and rapid pressure rise once the sealing layer fails. Second, it is revealed that gap conditions fundamentally govern the reaction pathway and the resulting structural failure mode—most notably, radial sealing induces preferential mid-shell rupture, indicating an altered spatial distribution of energy release. Although reaction intensities varied to some extent among configurations, the differences were not substantial, and no transition to detonation was observed in any test. These findings provide mechanistic insights and quantitative data for assessing the safety of energetic materials.
This study investigates the combustion behaviour of a single full-size wood briquette under realistic operating conditions in a residential wood stove. Spatially resolved temperature measurements were performed inside the briquette during combustion to characterise the internal heating process and the evolution of thermal gradients within the fuel. The experimental data from a single combustion experiment were combined with thermogravimetric analysis (TGA) and a non-stationary heat conduction model to reconstruct the devolatilization dynamics of the briquette and the temporal release of major fuel components.The model describes the temperature-dependent mass loss of water, carbon, hydrogen, and oxygen and enables the estimation of the evolution of briquette density, heat accumulation, and volatile release during combustion. The results show that strong radial temperature gradients develop inside the briquette, leading to a delayed heating of the core region where pyrolytic processes dominate under oxygen-deficient conditions. Thermogravimetric analysis performed under an inert atmosphere indicated that approximately 80% of the original combustible mass may be released as volatile gaseous products during the devolatilization phase, while the remaining solid fraction consists predominantly of fixed carbon.The presented approach provides a physically consistent framework linking internal heat transfer, fuel decomposition, and gas release in densified biomass fuels. The results contribute to improved understanding of briquette combustion behaviour and may support the development and validation of combustion models for residential biomass heating appliances.
Predicting the transient penetration of the liquid phase is critical for optimising fuel-air mixing and preventing wall wetting in modern combustion systems. This study performs a comprehensive meta-analysis of 17 test fuels—including diesel, n-dodecane, and n-heptane—to evaluate the universality of spray correlations published between 1998 and 2025. We demonstrate that foundational heat-transfer-based models (the Higgins-Siebers lineage), when refined with specific thermodynamic adjustments, provide the most robust fit across all fuels and conditions to date. The equation avoids the tendency to overfit or to produce dimensionally non-homogeneous equations that are valid only for specific experiments.A core contribution of this work is the development of a generalised transient model for liquid-phase penetration (S_L). This model tracks the liquid phase as it follows the vapour-phase penetration through distinct regimes: growth is governed by the spray breakup time (t_b), and a transition to a steady-state liquid length (LL) occurs. We propose a unifying expression, S_L=d_e∙max(8, min(40, LL⁄d_e, S_L⁄d_e)), which effectively bounds penetration between the breakup length and a characteristic maximum of ∼ 40 equivalent diameters. The model is shown to be valid for almost all known test fuels under typical operating conditions, with a predictive accuracy of ±16%, making it a reliable tool for engine designers. This study reconciles the scatter in the historical literature by quantitatively comparing diagnostic discrepancies among Mie scattering, DBI, shadowgraphy, and X-ray radiography. We map the parametric sensitivities of the limiting liquid length, illustrating how fuel properties (boiling point, latent heat, and specific heat) and ambient conditions (density and temperature) shift the evaporation boundaries. This work provides a definitive framework for standardising transient liquid spray behaviour, suitable for both analytical design and CFD validation.
This study provides a direct comparison of cassava rhizome pyrolysis under direct- and indirect-heating modes using an integrated fixed-bed carbonizer and cross-draft biomass gasifier–burner system. Process performance, product distribution, product quality, and greenhouse gas emissions were evaluated. At similar average reactor temperatures of 553 and 550°C, direct heating reduced the processing time from 370 to 240 min, corresponding to a 35.1% reduction, and decreased firewood consumption from 56.00 to 15.00 kg, equivalent to a 73.2% reduction. Biochar productivity increased from 1.13 to 2.82 kg h⁻¹, while biochar yield remained comparable at 24.3–25.0 wt.%. Direct heating increased pyroligneous-acid yield from 34.8 to 43.8 wt.%, an increase of 9.0 percentage points, and reduced pyrolysis gas yield from 40.2 to 31.9 wt.%. The produced biochar contained 76.01–77.00 wt.% carbon and exhibited iodine numbers of 195–220 mg g⁻¹, indicating potential for deodorizing and other non-fuel applications, although its ash content exceeded the requirements for cooking and grilling charcoal. Direct heating also reduced total greenhouse gas emissions from 101.99 to 27.32 kg CO₂-eq and specific emissions from 9.19 to 2.85 kg CO₂-eq kg⁻¹ biochar, corresponding to a reduction of approximately 69.0%. Compared with traditional kilns requiring 72–84 h, the direct-heating system completed the process within 4 h while maintaining a comparable biochar yield. Overall, direct heating improved process intensity, fuel efficiency, and environmental performance under the tested system configurations.
This study experimentally and numerically investigates passive pre-chamber combustion of pure ammonia in an internal combustion engine. The goal of the work is to explore how pre-chamber geometry ameliorates combustion characteristics of ammonia, including low flame speed and low reactivity, under low- and mid-load conditions. The results demonstrate that very stable combustion (CoVIMEPg < 1%) at 25% relative indicated load (near idle) is possible with an appropriate pre-chamber geometry at 1200 and 1800 RPM, respectively. An 8-hole, 1.1 mm diameter nozzle with A/V ratio of 2.4 was shown to be too small for low-load at 1200 RPM and exhibited two separate instability mechanisms. Poor pre-chamber gas exchange led to variable jet magnitude and timing that was nearly twice as variable as the other configurations. The 1.1 mm nozzle also produced a turbulent jet quenching effect in the main-chamber that caused poor re-ignition stability. The numerical results confirm a stronger presence of highly turbulent thickened flame regime in the main-chamber with the 1.1 mm nozzle with an average Karlovitz number of 320.4 compared to the 178.6 of the 1.6 mm nozzle. At 1800 RPM, the 8-hole, 1.6 mm diameter nozzle was no longer stable, but only produced instability linked to turbulent jet quenching. The results demonstrate that hole size and A/V ratio must both be considered when choosing pre-chamber configurations and that properly sized nozzle geometry can enable low load operation.
Aluminum-lithium (Al-Li) alloys are promising candidates for solid propellant applications due to their high specific energy and favorable combustion characteristics. However, the high chemical reactivity introduced by Li raises concerns regarding its stability and compatibility with propellant ingredients. This study evaluates the stability of Al-10Li particles containing 10 wt.% Li and their compatibility with representative propellant components. Results indicate that exposure to ambient air promoted the formation of hydroxyl, carbonate, Al-O, and Li-O species on the particle surfaces. These changes were accompanied by surface cracking and attenuation of the AlLi phase. After aging in ambient air, the rate of decrease in heat of combustion and the rate of increase in particle mass both rose with increasing relative humidity (RH). X-ray photoelectron spectroscopy (XPS) and Fourier transform infrared (FTIR) measurements imply that H2O and CO2 play important roles in the structural damage after aging in ambient air. Compatibility tests of Al-10Li with key components in solid propellant at 60°C show that Al-10Li particles are unreactive with hydroxyl-terminated polybutadiene (HTPB). But it shows mild reactivity with toluene diisocyanate (TDI) and obvious reactivity with dioctyl sebacate (DOS). In contrast, mixing Al-10Li particles with ammonium perchlorate (AP) at room temperature produces a much stronger reactivity, including severe agglomeration and the release of alkaline gaseous products.
Predicting whether short-circuit abuse will trigger thermal runaway (TR) in lithium-ion batteries requires an explicit criticality criterion, rather than trajectory-resolved simulations alone. Here, we develop a Semenov-type analytical criterion by reducing the external- and internal-short-circuit (ESC/ISC) stage to an equivalent thermal impulse and evaluating the resulting post-abuse state with an extended thermal-explosion condition. A coupled lumped electro-thermal-kinetic benchmark model, assembled from established submodels and compared with published abuse-test data, shows that the dominant electrical heat release is completed on a shorter time scale than subsequent TR induction for the tested 18650 lithium nickel–manganese–cobalt oxide (NMC) baseline. Together with a posteriori trajectory comparisons initialized from the analytical post-abuse state, this hierarchy supports a sequential short-circuit-to-criticality mapping. The criterion incorporates linearized radiative heat loss and a first-order reactant-consumption correction for parallel decomposition reactions, yielding explicit onset boundaries in ambient temperature, initial state of charge, and external short-circuit resistance. Relative to the benchmark, the analytical boundary captures the critical-surface topology and is conservative in two representative external-resistance slices. Closure-uncertainty analysis identifies the heat-release partitioning between negative- and positive-electrode decomposition reactions as the dominant source of boundary variability. Additional lithium iron phosphate (LFP) tests, together with extended global sensitivity screening, show that transfer beyond the NMC baseline requires chemistry- and pathway-specific recalibration, particularly of TISC and η. The formulation provides a transparent cell-scale tool for rapid safety screening within calibrated lumped battery models.
This study employs optical emission spectroscopy to experimentally assess thermal non-equilibrium (NEQ) in stoichiometric H2 detonations at 293 K under different initial pressures (30–100 kPa), diluents (Ar/N2), dilution levels (40%–70%), and positions behind the shock. In total, more than 150 single-shot spectra are collected, and the study is conducted in three main parts. Firstly, the study focuses on collecting broadband emission spectra (200–700 nm) to identify the contributing species to each spectrum. The results indicate that only the hydroxyl radical (OH), near 310 nm, contributes significantly to all the spectra. Hence, this first result indicates that emission spectroscopy cannot be employed to measure the thermal non-equilibrium of the reactants (H2, O2, or N2) in the induction zone for any studied condition. Secondly, the study specifically characterizes the OH⋆ spectra, produced in the exothermic reaction zone, from which vibrational NEQ is observed. For most of the tested conditions, vibrational NEQ of OH(A2Σ+) is detected, with a vibrational temperature 150–450 K above the gas temperature. In addition, our measurements indicate that the vibrational NEQ of OH (i) decreases with pressure, (ii) is stronger in N2-diluted than Ar-diluted mixtures, and (iii) increases with the dilution levels. Thirdly, the study is dedicated to the characterization of the relaxation of OH behind the detonation waves by measuring the evolution of the OH⋆ spectra behind the shock. Our measurements evidence that OH relaxes to equilibrium faster at high pressure and low dilution, and that the relaxation is faster in Ar-diluted than N2-diluted mixtures. The experimental OH relaxation times agree with the predicted ones from the well-established Millikan and White correlations, within 5% or up to a factor of 3 on average, when including or neglecting the experimentally determined OHH2O contribution (i.e., determined from experimental fitting), respectively.
Ammonia has gained considerable interest from the combustion community as an alternative fuel. The chemical promises to remove carbon entirely from heating, power and propulsion processes without incurring in extensive infrastructure developments, an important parameter for isolated regions located far from electrical or natural gas grids. As such, the work on ammonia combustion expands from fundamental analyses to giga-Watt practical applications in power generation. However, experimental research usually lacks further information about radical interactions, blends reactivity, emissions abatement, flame stability, and scalability parameters. Therefore, numerical modelling via Computational Fluid Dynamics (CFD) not only offers detailed insights into these combustion challenges using ammonia but also ensures that by applying novel methodologies ammonia resolution is accurate, relevant and affordable for a topic with such a large impactful potential. Therefore, this review addresses a large literature survey from researchers and technologist across the globe currently assessing ammonia as an energy vector. The review approaches the current situation of ammonia continuing with a brief description of the importance of CFD to analyse ammonia flames. After this section the review addresses relevant CFD research on topics that progress from the most fundamental combustion parameters to innovative applications such as liquid injection, plasmas and porous media systems. Finally, major challenges and future developments are presented, with conclusions that address the versatility of studies and most interesting areas of further research relevant to the use of CFD modelling in the resolution of fundamental parameters in ammonia combustion systems. The novelty of the review lays on the discussion of the latest research, via computational fluid dynamics, intended to delve new technologies such as small burners, liquid atomization, plasma systems and porous media to deploy ammonia energy concepts to new fields that include medium heating, small transport, domestic applications and even aerospace systems.
Methanol is an alternative fuel gaining traction in the maritime sector. Its direct adoption, however, is accompanied by a unique set of technical challenges, such as low cetane number and high latent heat of vaporization. An approach to overcome these challenges is being developed at the US Department of Energy’s Oak Ridge National Laboratory, where onboard generation of dimethyl ether (DME) via catalytic dehydration of methanol can be used to assist in the mixing controlled combustion of direct-injected (DI) methanol. The generated mixture from this dehydration process can be premixed with intake air to condition the cylinder via. homogeneous charge compression ignition (HCCI) for subsequent DI methanol. In this preliminary work, various catalyst or reactor conversion efficiencies were simulated (using bottles) at constant DME and water flow at low load on a single-cylinder marine-variant of a CAT® C18 18 L engine with a 145 mm bore. To substantiate the experimental findings, a zero-dimensional engine model was developed in Cantera using a DME mechanism with 79 species and 658 reactions. Results presented include experimental and simulation heat release rate comparisons, species evolution information, and constant volume ignition delay (ID) for DI methanol with and without background species from HCCI of the premixed products from different reactor efficiencies. The results suggest that thermal effects dominate the DI methanol ignition process, and this work provides a chemical kinetic foundation or guideline for developing future control schemes.
Multi-flame fires represent one of the most challenging fire scenarios to extinguish, and the intense injection of suppression agent may cause secondary ignition. This work investigates the performance of air vortex rings in extinguishing a line of up to 7 candle flames (38 W each in normal gravity) and the effects of buoyancy-driven disturbances. To decouple the buoyancy effect, microgravity experiments are conducted in a 3.6-s drop tower, and a piston-tube apparatus is developed to generate stable vortex rings with energies up to 1 mJ for extinguishing multiple candle flames. Results show an 11% relative increase in vortex extinguishing capacity in microgravity, based on which the mechanism for vortex-induced multiple flame extinction is revealed using flame stretch rate. The rotation of the vortex ring contributes strongly to flame stretch and entrainment, while the translation laterally blows the flame away. In microgravity, the vortex ring requires a lower stretch rate to extinguish the flame than in normal gravity, and the elimination of buoyancy disturbances allows it to maintain a high rotational kinetic energy. In multi-flame scenarios, the vortex ring effectively utilizes its energy, achieving up to 3 times the efficiency compared to extinguishing a single flame. This efficiency can be further improved in microgravity or by optimizing the kinetic energy structure of the vortex ring. This work provides a fundamental understanding of vortex-induced extinction, offering insights for future firefighting systems for spacecraft and extraterrestrial applications.
While large-scale waste incineration offers benefits such as volume reduction and energy generation, domestic combustion, driven by heating cost savings, is illegal in the EU due to uncontrolled air pollution. Although plastics have heating values comparable to those of coal, data on their household-level emission factors remain scarce. This study investigates the co-combustion of polyethylene terephthalate (PET) and polystyrene (PS) at 5 wt% and 20 wt% concentrations in a 25 kW automatic pellet-fired boiler. Experiments were conducted under two conditions: optimal air intake (8–11% O2 in dry flue gas) and low air intake (1–4% O2 in dry flue gas).Results showed that under optimal air conditions, carbon monoxide (CO) and organic gaseous compounds (OGC) emissions from plastic incineration were comparable to or only slightly higher than those from wood pellet combustion. However, low air intake led to substantial increases, with CO levels rising 10–41 times and OGC levels 37–97 times higher than those for the same fuel combusted under optimal air intake. Increased plastic content generally led to higher CO, OGC, and total suspended particles (TSP), particularly under oxygen-limited conditions. Under low air intake, PS produced higher gaseous and TSP emissions than PET. Finally, thermochemical analysis (thermogravimetric analysis with mass spectroscopy, TG-MS) was used to characterize gas phase profiles. Overall findings underscore the significant environmental risks of domestic plastic-waste combustion, particularly when air-supply regulation is inadequate.
Quantitative measurements of NH2 radicals are essential for validating ammonia combustion kinetics but remain experimentally challenging in low-pressure, confined environments. This current study presents the first spatially resolved, absorption-based quantification of ground-state NH2 radicals in three premixed NH3/O2/N2 flat flames (ϕ = 0.87–1.30; P=75 Torr) using cavity ring-down spectroscopy (CRDS). The widely used A˜2A1(0,9,0) ← X˜2B1(0,0,0) PQ1,N(7) transition at 16739.9 cm⁻¹ was found unreliable under these conditions due to strong H₂O broadband absorption and multi-exponential time-decay of CRDS signal from excessive optical depth. A weaker transition within the A˜2A1(0,9,0) ← X˜2B1(0,0,0) band, PP₄,₅(8) at 16764.97 cm⁻¹, combined with localized N2 purging, ensured optically thin absorption and clean mono-exponential time-decays. Absolute mole fractions were obtained through oscillator-strength anchoring at a validated reference location, simulation-derived effective path lengths, and Boltzmann correction from independently measured temperatures. Peak NH2 mole fractions reached approximately 5500, 4200, and 7000 ppm for ϕ = 0.87, 1.10, and 1.33 respectively, with peak locations shifting downstream and profiles broadening as the mixture became richer. Simulations reproduce peak positions well but systematically underpredict mole fraction magnitudes, particularly in the rich flame. The reported NH2 profiles provide new benchmark data for ammonia oxidation and NOₓ formation mechanisms, and the methodology offers a transferable diagnostic framework for radical quantification in systems with presence of broadband absorbers.
Difluoromethane (R32) is a mildly flammable refrigerant widely adopted in low-GWP air-conditioning systems. Understanding its forced ignition characteristics is essential for assessing safety risks associated with accidental ignition. In this study, the forced ignition behavior of stoichiometric R32/air mixtures is investigated numerically using the in-house code INSFLA, which solves one-dimensional conservation equations with detailed molecular transport and a newly developed skeletal CH2F2 kinetic mechanism. The mechanism, derived from an optimized detailed model, includes 39 species and 75 reactions and accurately reproduces the key fluorine-abstraction and chain-branching pathways governing R32 oxidation. The effects of ignition energy, ignition radius, ignition duration, and geometry on flame-kernel formation, quenching, and propagation are examined. Localized ignition may occur at moderate energies but often quenches; and only sufficiently high energies produce self-sustained flames. The minimum ignition energy for a successful flame propagation, MIEprop, scales approximately with rs3 for large ignition radii but becomes nearly constant for small rs. Combustion of R32 produces substantial amounts of HF, which becomes a dominant reaction product during sustained propagation. These findings provide mechanistic insight into R32 ignition behavior and safety implications for mildly flammable refrigerants.
With an increasing focus on adopting biofuels in the transportation sector, methanol is rapidly emerging as one of the most promising options globally. For the development of their engine technology, fuel spray characterisation is crucial, as it determines charge formation and combustion quality. This study characterises methanol spray droplets from a multi-hole gasoline direct injection (GDI) injector at a wide range of engine-relevant ambient pressure (Pa) conditions, ranging from 0.5 to 16 bar. The results were compared with baseline gasoline. Phase Doppler Interferometry (PDI) was employed to determine the velocity and size distributions of spray droplets at various axial and transverse locations within the spray plume. The morphological characteristics were determined by spray imaging using diffused backlit illumination (DBI) technique. Increasing the ambient pressure to 8 bar decreased the peak droplet velocity by ∼50 %. The peak droplet velocities were higher for methanol at ambient pressures of 0.5 and 1 bar than for baseline gasoline. Methanol exhibited a lower arithmetic mean diameter (AMD) and Sauter mean diameter (SMD) than baseline gasoline at 0.5 bar ambient pressure. However, it showed higher AMD and lower SMD than baseline gasoline at both atmospheric and elevated pressures. Gasoline’s non-uniform droplet size distribution, characterised by a higher number of finer droplets and fewer larger ones, decreased the AMD and increased the SMD of the fuel spray. M100 spray showed lower AMD at the centre and higher AMD at the periphery of the spray plume. The overall morphological characteristics of methanol were quite similar to baseline gasoline. These findings indicate that methanol can be effectively utilised in GDI engines with minor modifications to the fuel injection system. The spray database thus generated can support injector design, validation of numerical spray models, and the optimisation of methanol-fuelled direct-injection engines for obtaining improved efficiency and superior emissions performance.
Accurate and quantitative detection of metallic species in the exhaust plumes of liquid rocket engines is essential for assessing component wear and enabling early fault diagnosis. Aiming to establish a deep‑learning‑based inversion model for plume metal concentrations, this paper presents three main contributions. First, a corrected line‑by‑line (LBL) spectral simulation method is developed. The correction model, which accounts for self‑absorption effects, is derived from empirical relationship observed in experimental data. Although calibrated using copper (Cu) measurements, the correction model is experimentally shown to be equally applicable to iron (Fe) and nickel (Ni). Second, by combining the corrected LBL simulations with a minimal set of experimental measurements (e.g., only two concentration levels per metal), the proposed approach generates a comprehensive and accurate training dataset without extensive experimental effort. Third, the inversion model itself is a fully connected feedforward neural network whose hyperparameters are optimized by a multi‑strategy Grey Wolf Optimizer (MS‑GWO). Experimental validation under unseen operating conditions demonstrates that the model achieves prediction accuracies exceeding 90% for Cu, Fe, and Ni, thereby confirming its strong generalization capability. The performance of the proposed MS‑GWO‑FNN model is further benchmarked against a genetic‑algorithm‑optimized back‑propagation (GA‑BP) network, a standard FNN, and a radial basis function neural network (RBFNN). The comparative results demonstrate that the MS‑GWO‑FNN achieves superior prediction accuracy relative to the other three models, confirming its strong potential to meet the stringent requirements of plume‑based health monitoring and diagnostics.
This study investigates nanosecond repetitively pulsed discharge (NRPD) enhancement of NH₃ combustion characteristics including ignition delay time (IDT), laminar flame speed, flame thickness, heat release rate, and NOx emissions in various oxidizer compositions across lean to rich conditions by utilizing a combined 0D and 1D numerical framework with the ChemPlasKin and Cantera solver. The model was validated against temperature measurements and time-resolved H₂O/OH/NO laser diagnostics and prior ZDPlasKin-Chemkin modeling. Parametric studies were undertaken systematically to isolate the thermal and kinetic effects of plasma, with the use of 30% argon dilution to minimize NOx formation. The results show that plasma activation significantly reduces the IDT by approximately two orders of magnitude, consistent with earlier reports, and further clarify the relative roles of thermal and kinetic plasma effects. IDT decrease is primarily driven by kinetic radical production (H, O, OH, NH₂) at lower temperatures, whereas thermal gas-heating effects become more comparable only at high inlet temperatures. In contrast, 1D simulations reveal that laminar flame speed is predominantly controlled by the thermal effects of the discharge, which also results in significantly thinner flame fronts and increased peak heat release rates under lean conditions, with NH₃/O₂ mixtures exhibiting superior reactivity over air-based systems. Shifts in the upstream peak and enhanced concentrations were also observed with plasma activation, while the effects were diminished under rich conditions and argon dilution. Regarding emissions, plasma reduces the peak NO formation because IDT occurs at lower temperatures, which slows the H/OH + HNO reactions that contribute significantly to thermal NO production. At the same time, plasma enhances the NO consumption pathway through NH2 + NO, enabling alternative reaction routes that can help reduce the final NO emissions. The results also showed that argon dilution not only lowers flame temperatures but also facilitates electron-impact energy transfer, allowing diluted mixtures under plasma to achieve ignition characteristics comparable to undiluted cases.
This study explores the use of machine learning (ML) techniques, specifically artificial neural networks (ANN) and gradient boosting(GB), to replace conventional flamelet look-up tables in turbulent combustion modeling. The approach is based on the unsteadyflamelet/progress variable (UFPV) formulation, which serves as the foundation for the underlying flamelet manifolds. A central focusis placed on the transformation of high-dimensional tabulated chemistry data into efficient ML surrogates suitable for integrationinto computational fluid dynamics (CFD) simulations. Accordingly, various data pre-processing strategies and training methodologiesare evaluated, and the predictive performance of the resulting ML models is thoroughly assessed. To demonstrate the generalityand robustness of the proposed framework, the methodology is applied to multiple fuels, including n-dodecane (C12H26), n-heptane(C7H16), and oxymethylene ether (OME34). Beyond predictive accuracy, the study emphasizes key practical advantages of theML-based approach, namely reduced memory footprint and lower computational overhead by eliminating runtime interpolation,quantifying its potential to enhance the efficiency of combustion simulations.
