Digital twin of a combustion furnace operating in flameless conditions: reduced-order model development from CFD simulations

The present paper presents the first-of-its-kind digital twin for a furnace operating in flameless combustion conditions. A methodology combining data compression, by means of Proper Orthogonal Decomposition (POD), and interpolation, using Kriging, was developed to design physics-based, reduced-order models (ROMs) for the prediction of combustion data at unexplored operating conditions. Three-dimensional simulations with detailed chemistry were carried out, spanning a wide range of operating conditions in terms of fuel composition (methane-hydrogen mixtures from pure methane to pure hydrogen), equivalence ratio (from 0.7 to 1) and air injector diameter (to adjust the air jet entrainment). Based on the available simulations, a ROM was developed, to predict both spatial fields, local and integral values of thermochemical variables at working conditions not included in the ROM development. Results showed that the developed ROM could reliably predict the temperature and main chemical species distribution in the furnace with an overall error below 10%, proving the effectiveness of the approach for the development of digital twins of combustion systems. A remarkable accuracy was observed for the prediction of specific quantities, including wall temperatures, OH decay length, OH peak value and location and exhaust gas composition, including pollutants, with prediction errors always below 5%, showing the potential of the approach to develop soft sensors.     

6110柴油机喷雾过程的多维瞬态数值模拟研究

将柴油机全体燃烧室部件(气缸盖-气缸套-活塞组)作为一个耦合体,在对耦合体进行传热数值模拟的基础上得到喷雾过程缸内计算的壁面边界条件.利用大型通用CFD软件STAR-CD及ES-ICE,在进气压缩过程多维瞬态数值模拟基础上,对6110柴油机喷雾过程进行多维瞬态数值模拟研究.通过计算,着重分析缸内两相流动,燃油喷射、雾化以及喷雾粒子的空间分布等.     

直喷式柴油机燃烧数值计算研究

以KIVA-3为计算平台,对4气门直喷式柴油机的早喷燃烧、预喷燃烧和传统燃烧三种燃烧方式工作过程进行了模拟计算,得以下结论:早喷燃烧方式因喷到缸壁的油量较多,受进气涡流影响较少,同时利用喷射效应改善缸内氧气分布,因此早喷燃烧比较完全。预喷燃烧方式仍以扩散燃烧为主,预喷油燃烧改变了主喷初期油注周围的组分分布与温度分布,从而改变主喷燃油的燃烧历程。进气涡流造成的缸内流场不均匀性是影响传统燃烧方式燃烧不完全的主要因素之一。     

Axis switching in turbulent buoyant diffusion flames

Dynamics of buoyant diffusion flames from rectangular, square, and round fuel sources were investigated using direct numerical simulation (DNS). Fully three-dimensional simulations were performed employing high-order numerical methods and boundary conditions to solve governing equations for variable-density flow and finite-rate Arrhenius chemistry. Significant differences among the different cases were revealed in the vortex dynamics, entrainment rate, small-scale mixing, and consequently flame structures. Mixing and entrainment enhancement in non-circular flames in comparison with circular ones was explained using the Biot–Savart instability theory, which relates vortex dynamics to the local azimuthal curvature. An extension of the theory elucidated why rectangular flames entrain more efficiently and spread wider than square ones, although both configurations have corners. It also provided an explanation for the aspect ratio effects in the near field. In the far field, nonlinear effects were dominant and the general transport equations for vorticity were analyzed in detail. The corner effects and aspect ratio effects were shown to be augmented by the intricate interactions among vortex dynamics, combustion, and buoyancy through the various terms in the equations. The presence of corners in non-circular flames led to concentrated regions of fine-scale mixing and intense reactions centered around the corners. Moreover, the rectangular flames exhibited a different dynamic behavior from even the square one, by creating discrepancies in entrainment, mixing, and combustion between the minor and major axis directions. Increasing the aspect ratio exacerbated such directional discrepancies, and ultimately led to axis switching. It was the first time that axis switching was observed by DNS in a rectangular flame of aspect ratio 3, which raised further questions in combustion prediction and control. Finally, a unified explanation for corner and aspect ratio effects was given on the basis of the Biot–Savart instability theory and the vorticity transport equations.     

几何参数对喷射器性能的影响分析

运用计算流体动力学软件Fluent对喷射器进行数值模拟,研究了工作喷嘴喉部直径、等截面混合室直径和长度等几何参数对喷射器性能的影响,并对数值模拟结果进行了分析。研究表明:在一定的工况下,工作喷嘴及等截面混合室对喷射器性能有极大的影响,存在最佳的喉部和混合室直径使得喷射系数最大;随着混合室长度的增加,喷射器的喷射系数将逐渐降低。     

Visualization of secondary flow choking phenomena in a supersonic air ejector

The present study deals with the visualization of the air flow inside a supersonic ejector. Our attention is more precisely focused on the choked flow phenomenon which occurs along the mixing chamber of the secondary nozzle and which can be visualized by CFD. Laser tomography visualizations are used to validate the CFD model. The evolution of flow configuration in the ejector with the primary stagnation pressure is examined both in the case of zero secondary flow and in the case of free entrainment of induced air.     

Influence of pilot-fuel mixing on the spatio-temporal progression of two-stage autoignition of diesel-sprays in low-reactivity ambient fuel-air mixture

The spatial and temporal locations of autoignition for direct-injection compression-ignition engines depend on fuel chemistry, temperature, pressure, and mixing trajectories in the fuel jets. Dual-fuel systems can provide insight into both fuel-chemistry and physical effects by varying fuel reactivities and engine operating conditions. In this context, the spatial and temporal progression of two-stage autoignition of a diesel-fuel surrogate, n-heptane, in a lean-premixed charge of synthetic natural-gas (NG) and air is imaged in an optically accessible heavy-duty diesel engine. The lean-premixed charge of NG is prepared by fumigation upstream of the engine intake manifold. Optical diagnostics include high-speed (15kfps) cool-flame chemiluminescence-imaging as an indicator of low-temperature heat-release (LTHR) and OH* chemiluminescence-imaging as an indicator high-temperature heat-release (HTHR). NG prolongs the ignition delay of the pilot fuel and increases the combustion duration. Zero-dimensional chemical-kinetics simulations provide further understanding by replicating a Lagrangian perspective for mixtures evolving along streamlines originating either at the fuel nozzle or in the ambient gas, for which the pilot-fuel concentration is either decreasing or increasing, respectively. The zero-dimensional simulations predict that LTHR initiates most likely on the air streamlines before transitioning to HTHR, either on fuel-streamlines or on air-streamlines in regions of near-constant ?. Due to the relatively short pilot-fuel injection-durations, the transient increase in entrainment near the end of injection (entrainment wave) is important for quickly creating auto-ignitable mixtures. To achieve desired combustion characteristics, e.g., multiple ignition-kernels and favorable combustion phasing and location (e.g., for reducing wall heat-transfer or optimizing charge stratification), adjusting injection parameters could tailor mixing trajectories to offset changes in fuel ignition chemistry.     

LES study of diesel flame/wall interaction and mixing mechanisms at different wall distances

In this paper, the flame-wall interaction of reacting diesel spray under engine like conditions is investigated using large eddy simulations. The aim of this study is to understand the influence of the distance between the wall and the spray nozzle on the air entrainment rate, which is a key variable in formation/oxidation process of soot. Three experimental cases are investigated, a free jet case and two wall impingement cases with a distance from nozzle to wall of 30 mm and 50 mm, which are considered as characteristic wall impingement distances for light- and heavy-duty bores in diesel engines, respectively. The optical soot measurements imply a positive influence of wall on the rate of soot oxidation. Numerical simulations are employed to elucidate importance of different mechanisms for the air entrainment, i.e., air entrainment prior to flame lift-off position, enhanced mixing due to the wall impingement and enhanced mixing by the entrainment wave. The results show that oxidation process after the end of injection is driven by a different mixing mechanism depending on the distance to the wall. The 30 mm case resulted in a “mixing boost”, where the dominant mixing mechanism is the wall impingement vortex mixing, which gives rise to the fastest soot decay among the cases. The mixing in the 50 mm case is governed by a late wall impingement vortex mixing, giving rise to a low, but a constant air entrainment rate, i.e., a “mixing plateau”. The free jet case resulted in mixing governed by the entrainment wave mechanism. Both wall impingement cases have faster soot oxidation rate compared with the free jet case, but due to a different underlying mixing process. LES is shown to be able to replicate the line-of-sight measurements of natural OH* chemiluminescence and distribution of soot region from the optical soot diagnostics.     

燃料敏感性对内燃机部分预混燃烧(PPC)光谱特性的影响研究

在一台光学发动机上,利用火焰高速成像技术和自发光光谱分析法,研究了燃料敏感性(S)为0和6时对发动机缸内火焰发展和燃烧发光光谱的影响。试验过程中,通过改变喷油时刻 (SOI=-25,-15和-5°CA ATDC) 使燃烧模式从部分预混燃烧过渡到传统柴油燃烧模式。通过使用正庚烷、异辛烷、乙醇混合燃料来改变燃料敏感性。结果表明,在PPC模式下(-25°CA ATDC),火焰发展过程是从近壁面区域开始着火,而后向燃烧室中心发展,即存在类似火焰传播过程,同时在燃烧室下部未燃区域也形成新的着火自燃点。敏感性对燃烧相位影响较大,对缸内燃烧火焰发展历程影响较小;高敏感性燃料OH和CH带状光谱出现的时刻推迟,表明高敏感性燃料高温反应过程推迟,且光谱强度更低,表明碳烟辐射强度减弱。在PPC到CDC之间的过渡区域(-15°CA ATDC),燃烧火焰发光更亮,燃烧反应速率比-25°CA ATDC时刻的反应速率更快。高、低敏感性燃料对缸压放热率的影响规律与-25°CA ATDC相近,此时的燃烧反应更剧烈,放热率更高,碳烟出现时刻更早。该喷油时刻下的光谱强度高于PPC模式下的光谱强度,说明此时的CO氧化反应与碳烟辐射更强。在CDC模式下(-5°CA ATDC),由于使用的燃料活性较低,燃烧放热时刻过于推迟,放热量很小,缸内燃烧压力低,因此燃料敏感性对缸压和放热率的影响不明显,但从燃烧着火图像中可以看到高敏感性燃料的火焰出现时刻较低敏感性燃料推迟。低敏感性燃料的燃烧初期蓝色火焰首先出现在燃烧室中心,着火火焰出现时刻更早,之后蓝色火焰从中心向周围扩散,呈现火焰传播为主导的燃烧过程;燃烧后期,局部混合气过浓区导致亮黄色火焰面积逐渐增大并向周围扩散。高敏感性燃料的火焰发展趋势与低敏感性燃料类似,黄色火焰的亮度与面积更小。尽管高、低敏感性燃料的OH和CH带状光谱的出现时间相近,但高敏感性燃料的光谱强度仍更低。综合分析,火焰发展结构与自发光光谱特征主要受喷油时刻的影响,燃料的敏感性主要影响着火时刻和火焰自发光光谱强度,且高敏感性燃料的光谱强度更低。     

Analysis of model dimensionality,particle shrinkage,boundary layer reactions on particle-scale modelling of biomass char conversion under pulverized fuel combustion conditions

In this work, the effects of model dimensionality, particle shrinkage, and boundary layer reactions on particle-scale modelling of biomass char conversion under pulverized fuel combustion conditions have been analysed by using six models: zero-dimensional models with constant particle size (0D_Cons) or shrinking particle size (0D_SPM), one-dimensional models with/without considering particle shrinkage (1D_Cons/1D_SPM), and 1D_Cons and 1D_SPM with considering boundary layer reactions (1D_Cons_BH and 1D_SPM_BH). A comparison with existing experimental data shows that the 1D_SPM_BH model with consideration of intra-particle heat and mass transfer, particle shrinkage, and boundary layer reactions is an appropriate model to describe biomass char conversion over a wide range of conditions. The 0D_Cons model is a good approximation for the conditions of small particle size (< 1 mm) at 1273–1473 K, but overestimates the char conversion rate for larger biomass char particle or at high temperatures (regime III). The 0D_SPM model gives a reasonable prediction on char conversion time but predicts a larger contribution of reaction between char and O₂ as compared to the 1D_SPM_BH model. The consideration of intra-particle heat and mass transfer in particle-scale modelling (1D_Cons and 1D_SPM) is beneficial to improving the model prediction of char conversion time and the contributions of char oxidation and gasification reactions. The boundary layer reactions have a significant effect on the prediction of char conversion for large particles (> 1 mm) and high temperatures (> 1473 K). An implication for the selection of a particle-scale model in CFD modelling is also given.     

JF12激波风洞高Mach数超燃冲压发动机实验研究

针对高Mach数（Ma ≥ 7）超燃冲压发动机高气动阻力下的燃烧组织问题,提出一种双突扩燃烧室结构方案.使用数值模拟方法考察了射流与双突扩燃烧室组合方式的混合燃烧特性.设计了双突扩超燃冲压发动机模型,在力学研究所JF12长试验时间激波风洞内,开展了Ma=7.0和Ma=9.5的氢燃料点火和燃烧试验对比.在风洞有效试验时间100 ms内,实现了Ma=7.0和Ma=9.5超燃冲压发动机的成功点火与稳定燃烧.在Ma=7.0情况下,进气道采用三维压缩,燃烧室入口设计Mach数Ma_c=2.5,壁面压力分布实验结果显示燃烧放热靠近燃烧室扩张段上游;在Ma=9.5情况下,进气道采用二维压缩,燃烧室入口设计Mach数Ma_c=3.5,由于燃烧室流动速度特别高,燃烧放热靠近燃烧室扩张段下游.      

3-D simulation of soot formation in a direct-injection diesel engine based on a comprehensive chemical mechanism and method of moments

Here, we propose both a comprehensive chemical mechanism and a reduced mechanism for a three-dimensional combustion simulation, describing the formation of polycyclic aromatic hydrocarbons (PAHs), in a direct-injection diesel engine. A soot model based on the reduced mechanism and a method of moments is also presented. The turbulent diffusion flame and PAH formation in the diesel engine were modelled using the reduced mechanism based on the detailed mechanism using a fixed wall temperature as a boundary condition. The spatial distribution of PAH concentrations and the characteristic parameters for soot formation in the engine cylinder were obtained by coupling a detailed chemical kinetic model with the three-dimensional computational fluid dynamic (CFD) model. Comparison of the simulated results with limited experimental data shows that the chemical mechanisms and soot model are realistic and correctly describe the basic physics of diesel combustion but require further development to improve their accuracy.     

液体火箭推力室面板发汗冷却与燃烧耦合数值模拟

对推力室的喷嘴多孔面板的发汗冷却和燃烧室内的燃料燃烧过程进行了耦合数值计算,建立了一个带燃烧的三维、真实气体、变物性的推力室CFD计算模型。利用UDF编写了CH₄、O₂、CO₂、H₂O气体的实际气体状态方程,并根据NIST物性数据拟合了不同温度和压力下各气体的比热容、扩散系数、黏性系数和导热系数等物性多项式。基于EDC模型建立了甲烷-氧燃烧的多步反应机理。计算了三种厚度的面板和多种燃料进口工况下的推力室内的发汗冷却和燃烧过程,研究了面板厚度、冷却剂进出条件等因素对发汗冷却和燃烧过程的影响规律。     

高温空气燃烧NOx排放特性的试验研究

通过两种结构烧嘴的热态燃烧试验对比,研究了烧嘴结构、燃气射流速度、过量空气系数对高温空气燃烧过程氮氧化物排放的影响特性。研究结果认为:在燃气喷口两侧布置两个矩形空气喷口的烧嘴,氮氧化物排放量低于圆形空气喷口烧嘴;随着燃气射流速度的提高,高温空气燃烧过程排放的氮氧化物逐渐减少。与普通燃烧过程不同的是,随着过量空气系数的提高,在一定范围内高温空气燃烧的氮氧化物排放量不断增加。分析认为,高温空气燃烧氮氧化物排放量与火焰体积、炉内氧气与燃气混合过程以及燃气射流和空气射流对炉内烟气的卷吸量有关。     

Investigations of soot formation in an optically accessible gasoline direct injection engine by means of laser-induced incandescence (LII)

This study presents the results of laser-induced incandescence (LII) measurements in an optically accessible gasoline direct injection engine. The focus was to evaluate LII as a particle measurement technique which is able to provide a deeper understanding of the underlying reaction and formation processes of soot in order to optimize the injection system to reduce exhaust gas emissions. A comparison of time-resolved LII, based on the model described by Michelsen, with an Engine Exhaust Particle Sizer (EEPS) was performed. In this context, the air–fuel ratio, the injection pressure and the injection timing have been varied while applying the measurement techniques in the exhaust system. In case of a variation of the air–fuel ratio, two-dimensional LII has been performed in the combustion chamber additionally. For each measurement, the Filter Smoke Number (FSN) was taken into account as well. Finally, a good agreement of the different techniques was achieved. Moreover, we found that by combining time-resolved LII and EEPS a differentiation of primary particles and agglomerates is possible. Consequently, a determination of the processes in the combustion chamber and agglomeration in the exhaust gas is feasible.     

Numerical study of premixed HCCI engine combustion and its sensitivity to computational mesh and model uncertainties

This study used a numerical model to investigate the combustion process in a premixed iso-octane homogeneous charge compression ignition (HCCI) engine. The engine was a supercharged Cummins C engine operated under HCCI conditions. The CHEMKIN code was implemented into an updated KIVA-3V code so that the combustion could be modelled using detailed chemistry in the context of engine CFD simulations. The model was able to accurately simulate the ignition timing and combustion phasing for various engine conditions. The unburned hydrocarbon emissions were also well predicted while the carbon monoxide emissions were under predicted. Model results showed that the majority of unburned hydrocarbon is located in the piston-ring crevice region and the carbon monoxide resides in the vicinity of the cylinder walls. A sensitivity study of the computational grid resolution indicated that the combustion predictions were relatively insensitive to the grid density. However, the piston-ring crevice region needed to be simulated with high resolution to obtain accurate emissions predictions. The model results also indicated that HCCI combustion and emissions are very sensitive to the initial mixture temperature. The computations also show that the carbon monoxide emissions prediction can be significantly improved by modifying a key oxidation reaction rate constant.     

Computational Modeling of Boundary Layer Flashback in a Swirling Stratified Flame Using a LES-Based Non-Adiabatic Tabulated Chemistry Approach

When operating under lean fuel–air conditions, flame flashback is an operational safety issue in stationary gas turbines. In particular, with the increased use of hydrogen, the propagation of the flame through the boundary layers into the mixing section becomes feasible. Typically, these mixing regions are not designed to hold a high-temperature flame and can lead to catastrophic failure of the gas turbine. Flame flashback along the boundary layers is a competition between chemical reactions in a turbulent flow, where fuel and air are incompletely mixed, and heat loss to the wall that promotes flame quenching. The focus of this work is to develop a comprehensive simulation approach to model boundary layer flashback, accounting for fuel–air stratification and wall heat loss. A large eddy simulation (LES) based framework is used, along with a tabulation-based combustion model. Different approaches to tabulation and the effect of wall heat loss are studied. An experimental flashback configuration is used to understand the predictive accuracy of the models. It is shown that diffusion-flame-based tabulation methods are better suited due to the flashback occurring in relatively low-strain and lean fuel–air mixtures. Further, the flashback is promoted by the formation of features such as flame tongues, which induce negative velocity separated boundary layer flow that promotes upstream flame motion. The wall heat loss alters the strength of these separated flows, which in turn affects the flashback propensity. Comparisons with experimental data for both non-reacting cases that quantify fuel–air mixing and reacting flashback cases are used to demonstrate predictive accuracy.     

燃油分级多点喷射低污染燃烧室的化学反应网络模型分析

本文采用基于详细化学反应机理的化学反应网络模型分析了航空发动机燃油径向分级多点喷射低污染燃烧室的NO_x排放特性。该分级燃烧室不同于传统燃烧室,头部由值班区和主燃区两个不同的燃烧区域,根据CFD得到的流场特性和当量比的分布特性对燃烧室进行分区构建化学反应器网络模型,研究了值班级当量比以及值班级和主燃级两级供油比例对排放的影响。同时,还分析了空气进口温度对NO_x排放的影响。得到了较为合理的变化趋势,为低污染燃烧室的初步设计提供了有益的指导。     

Visualization of Crevice Flow in an Engine Using Laser-Induced Fluorescence

A simultaneous visualization technique of reacting and unburned zones using laser-induced fluorescence (LIF) was applied to a high-pressure combustion field in an engine cylinder. Crevice flow from a crevice between a piston and a cylinder wall of a spark ignition gas engine was visualized by LIF of OH and acetone. OH was excited simultaneously with acetone that was seeded into fuel as a tracer by an excitation light at 283.92 nm. Fluorescence signals from each species were detected individually by two intensified CCD cameras using optical band-pass filters which transmit fluorescence wavelength of OH and acetone, respectively. Pressure- and temperature-dependence of LIF signals from each species were evaluated. From the visualized images, it was clarified that oxidation of the crevice flow is stopped at the time of exhaust valve opening. Effects of exhaust port pressure on the oxidation process were investigated.     

Flow Mechanisms and Diffusion Combustion of Turbulent Jets

The problem of flow and combustion of turbulent jets of fuel gas in the external medium of an oxidant (air) is solved with regard to the existence of the actual boundary of the turbulent flow region of a jet. Based on the ideas of the friction force of the external flow acting on the boundary of a jet, the entrainment equation for the external medium is derived that closes the system of equations of motion of turbulent jets. The physical meaning of the dissipation rate of the turbulent energy of a jet is interpreted as the work of the friction force. To describe the combustion kinetics, the limit of instantaneous reactions corresponding to the diffusion combustion mode is used. Calculations of the effective reaction rates for reactants and the volumes occupied by them are based on the representation of a turbulent medium as an aggregation of independent turbulent particles—vortices—whose random contacts lead to the mixing and combustion of reacting substances [31]. The concomitant phenomena of flow and combustion are analyzed, including radiation effects. In particular, it is shown that the apparent increase in the combustion temperature with increasing Reynolds number is in fact attributed to the relative decrease of thermal radiation losses. Qualitative agreement is obtained between the results of the theoretical calculations of the length of a combustion torch and experimental data.