Liming Cai

Oxymethylene ethers (OMEs) are promising alternative fuels. A comprehensive evaluation of the prediction accuracy of their chemical kinetic models is essential for their computational fluid dynamics simulations. Thus, this study evaluates comprehensively the performance of ten recently proposed kinetic models of OME_1–4 against 43 experimental datasets, including ignition delay times, laminar burning velocities, and species concentrations. The investigated models are different in terms of their development methods and chemical natures. While all ten models include the chemistry of OME₁, seven, five, and three models of them can describe the combustion of OME₂, OME₃, and OME₄, respectively. For OME₁, the models of Li et al. (2021), Cai et al. (2020), and Jacobs et al. (2019) demonstrate prediction advantages over other models. The model of Shrestha et al. (2022) provides the lowest prediction uncertainties for the combustion of OME₂, while the reduced model of Dinelli et al. (2024) matches the data of OME₃ and OME₄ best. However, none of these models achieves very high prediction accuracy for all considered targets and, more importantly, for all interested OMEs. The results of sensitivity and reaction flux analyses reveal that, for all OMEs, the fuel decomposition reactions via unimolecular H-transfer influence their auto-ignition at high temperatures and pyrolysis behaviors significantly. The auto-ignition behavior of long-chain OMEs over a wide temperature range is highly sensitive to the reactions of smaller OMEs, as their primary fuel radicals are produced via the β-scission of radicals of longer OMEs. These findings provide insights into improved chemical models of OMEs.

The safety performance of liquid hydrogen storage has a significant influence on its large-scale commercial application. Due to the complexity and costs of experimental investigation, computational fluid dynamics (CFD) simulations have been extensively applied to investigate the dynamic behaviors of liquid hydrogen release. The involved physical and chemical models, such as models of species thermodynamic properties and liquid–gas mass transfer, play a major role for the entire CFD model performance. However, comprehensive investigations into their impacts remain insufficient. In this study, CFD models of liquid hydrogen release were developed by using two widely used commercial simulation tools, Fluent and FLACS, and validated against experimental data available in the literature. Comparisons of the model results reveal strong discrepancies in the prediction accuracy of temperature and hydrogen volume fraction between the two models. The impact of the models of thermodynamic properties and liquid–gas mass transfer on the prediction results was subsequently explored by incorporating the FLACS sub-models to Fluent and evaluating the resulting prediction differences in temperatures and hydrogen volume fractions. The results show that the models of thermodynamic properties and liquid–gas mass transfer used in FLACS underestimate the vertical rise height and the highest hydrogen volume fraction of the cloud. Sensitivity analyses on the parameters in these sub-models indicate that the specific heats of hydrogen and nitrogen, in conjunction with the mass flow rate and outflow density of the mass transfer model, have a significant influence on model prediction of temperature.

The sooting characteristics of fuels are of high relevance for their application in advanced combustion devices. While practical petroleum fuels are often mixtures of various components, most literature studies have focused on the numerical and experimental investigation of the sooting behaviors of neat components, often in terms of the so-called yield sooting index (YSI). Recently, a YSI database was established for 21 neat fuel components and 151 mixtures of them. An artificial neural network (ANN) model was developed to predict the YSI of fuel mixtures with reasonable accuracy by only using the mass fractions of mixture components as input features. However, pioneering studies have demonstrated that if quantitative structure-property relationship (QSPR) is incorporated, the ANN models can match the measured physical and chemical properties of fuels very well and provide insights into the dependence of properties on fuel structures. Thus, in this study, in order to predict the YSI of fuel mixtures with improved accuracy and to understand the impact of functional groups on the sooting tendency of fuel blends, an artificial neural network (ANN)-based QSPR model is derived by taking functional group descriptors as input features, in conjunction with the mass fractions of mixture components. The model is cross-validated successfully and gives more satisfactory results than the literature ANN models, which consider only mass fractions as input features. In addition, the influence of functional groups on the sooting tendency of fuel mixtures is evaluated for various blending ratios and fuel classes by means of sensitivity analysis.

For net‐zero carbon emissions, hydrogen/ammonia blends have drawn considerable attention for the application in industrial combustion devices. Various chemical mechanisms have been developed to describe the oxidation and combustion of hydrogen/ammonia mixtures at certain conditions. A comprehensive evaluation and comparison of the performance of these mechanisms is thus of high interest, especially in terms of their application for particular computational studies. Thus, this work aims to compare the existing chemical mechanisms in terms of their performance for the combustion of hydrogen/ammonia mixtures over a wide range of experimental conditions. In addition to previous literature studies, the model performance is evaluated by using two different methods for the assessment of prediction accuracy. Besides the conventional measure of point‐wise differences between model and data, the curve‐matching method is also applied, which quantifies the dependence of model response on physical conditions additionally, by comparing the similarity between the curve shapes of the predicted and measured results. Extensive experimental data are taken into account in the model evaluation, including 136 datasets obtained from various facilities in the past 10 years. Nineteen mechanisms are compared, which were published in recent five years. It is revealed that these models give strongly different numerical results for combustion targets, such as laminar burning velocities, ignition delay times, and species concentrations. The chemical mechanisms of Zhang et al. (2021), Han et al. (2023), Mei et al. (2019), Li et al. (2019), and Stagni et al. (2020) show relatively satisfactory performance over the entire investigated domain. Moreover, it is found that the estimated prediction accuracy of chemical mechanisms is highly sensitive to model evaluation methods.

Reynolds-averaged Navier–Stokes (RANS) three-dimensional (3D) computational fluid dynamics (CFD) simulations of gaseous emissions from combustion engines are very demanding due to the complex geometry, the emissions formation mechanisms, and the transient processes inside the cylinders. The validation of emission simulation is challenging because of modeling simplifications, fundamental differences from reality (e.g., fuel surrogates), and difficulty in the comparison with measured emission values, which depend on the measuring position. In this study, detailed gaseous emission data were acquired for a spark ignition (SI) direct-injection (DI) single-cylinder engine (SCE) fueled with a toluene reference fuel (TRF) surrogate to allow precise comparison with simulations. Multiple devices in different sampling locations were used for the measurement of average emission concentration, as well as hydrocarbon (HC) cycle- and species-resolved values. A RANS 3D-CFD methodology to predict gaseous pollutants was developed and validated with this experimental database. For precise validation, the emission comparison was performed in the exact same locations as the pollutants were measured. Additionally, the same surrogate fuel used in the measurements was defined in the simulation. To focus on the emission prediction, the pressure and heat release traces were reproduced by calibrating a G-equation flame propagation model. The differences of simulation results with measurements were within 4% for CO₂, while for O₂ and NO, the deviations were within 26%. CO emissions were generally overestimated probably because of inaccuracies in mixture formation. For HC emissions, deviations up to 50% were observed possibly due to inexact estimation of the influence of the piston-ring crevice geometry. The reasonable prediction accuracy in the RANS context makes the method a useful framework for the analysis of emissions from SI engines, as well as for mechanism validation under engine relevant conditions.