The high cost of high-resolution computational fluid/flame dynamics (CFD) has hindered its application in combustion related design, research and optimization. In this study, we propose a new framework for turbulent combustion simulation based on the deep learning approach. An optimized deep convolutional neural network (CNN) inspired from a U-Net architecture and inception module is designed for constructing the framework of the deep learning solver, named CFDNN. CFDNN is then trained on the simulation results of hydrogen combustion in a cavity with different inlet velocities. After training, CFDNN can not only accurately predict the flow and combustion fields within the range of the training set, but also shows an extrapolation ability for prediction outside the training set. The results from CFDNN solver show excellent consistency with the conventional CFD results in terms of both predicted spatial distributions and temporal dynamics. Meanwhile, two orders of magnitude of acceleration is achieved by using CFDNN solver compared to the conventional CFD solver. The successful development of such a deep learning-based solver opens up new possibilities of low-cost, high-accuracy simulations, fast prototyping, design optimization and real-time control of combustion systems such as gas turbines and scramjets.
Eulerian-Lagrangian simulations are conducted for two-dimensional Rotating Detonative Combustion (RDC) fueled by partially prevaporized n-heptane sprays. The influences of droplet diameter and total equivalence ratio on detonation combustion and droplet dynamics are studied. It is found that small n-heptane droplets (e.g. 5 um) are completely vaporized around the detonation wave, while intermediate n-heptane droplets (e.g. 20 um) are consumed in or behind the detonation wave, with the escaped ones be continuously evaporated and deflagrated. The droplet distributions in the RDE combustor are si…
In the application of the combustion models based on low-dimensional manifolds (for instance flamelet models) to large-eddy simulation (LES) of reactive turbulent flows, the modeling simplifications of the combustion process is a critical source of uncertainty in addition to those due to the turbulent closure model and numerical discretization. The ability to quantitatively assess this uncertainty in absence of the reference result is vital to the reliable and predictive simulations of practical combustion devices.In the present study, the Pareto-efficient combustion (PEC) framework is extende…
A new computational methodology, termed "PeleLM-FDF" is developed and utilized for high fidelity large eddy simulation (LES) of complex turbulent combustion systems. This methodology is constructed via a hybrid scheme combining the Eulerian PeleLM base flow solver with the Lagrangian Monte Carlo simulator of the filtered density function (FDF) for the subgrid scale reactive scalars. The resulting methodology is capable of simulating some of the most intricate physics of complex turbulence-combustion interactions. This is demonstrated by LES of a non-premixed CO/H$_2$ temporally evolving jet fl…
We propose a model for assessing the unresolved wrinkling factor in LES of turbulent premixed combustion. It relies essentially on a power-law dependence of the wrinkling factor to the filter size and an original expression for the active corrugating strain rate. The latter is written as a product of an efficiency function which accounts for viscous effects and the kinematic constraint of Peters, by a recent expression for the turbulent strain intensity. Yields functional expressions for the fractal dimension and the inner cut-off length scale, the latter being (i) filter-size independent and …
A novel jet-stirred combustion chamber is designed to study turbulent premixed flames. In the new approach, multiple impinging turbulent jets are used to stir the mixture. It is well known that pair of counterflowing turbulent jets produces nearly a constant intensity along the jet axes. In this study, different numbers of impinging jets in various configurations are used to produce isotropic turbulence intensity. FLUENT simulations have been conducted to assess the viability of the proposed chamber. In order to be able to compare different configurations, three different non dimensional indic…
We investigate and model pressure effects on the turbulent burning velocity over a wide range of pressures and turbulence intensities with the direct numerical simulation (DNS) of statistically planar turbulent premixed flames for lean hydrogen/air mixture. DNS results indicate that the stretch factor has an impact on the turbulent burning velocity and flame surface area at elevated pressures. In particular, the enhanced stretch factor at high pressures increases the ratio of turbulent and laminar burning velocities, diminishing the "bending" effect. Based on a good consistency between turbule…
Transported probability density function (PDF) methods are widely used to model turbulent flames characterized by strong turbulence-chemistry interactions. Numerical methods directly resolving the PDF are commonly used, such as the Lagrangian particle or the stochastic fields (SF) approach. However, especially for premixed combustion configurations, characterized by high reaction rates and thin reaction zones, a fine PDF resolution is required, both in physical and in composition space, leading to high numerical costs. An alternative approach to solve a PDF is the method of moments, which has …
Experimental multi-scalar measurements in laboratory flames have provided important databases for the validation of turbulent combustion closure models. In this work, we present a framework for data-based closure in turbulent combustion and establish an a priori validation of this framework. The approach is based on the construction of joint probability density functions (PDFs) and conditional statistics using experimental data based on the parameterization of the composition space using principal component analysis (PCA). The PCA on the data identifies key parameters, principal components (PC…
In this work, we demonstrate a framework for developing closure models in turbulent combustion using experimental multi-scalar measurements. The framework is based on the construction of conditional means and joint scalar PDFs from experimental data based on the parameterization of composition space using principal component analysis (PCA). The resulting principal components (PCs) act as both conditioning variables and transport variables. Their chemical source terms are constructed starting from instantaneous temperature and species measurements using a variant of the pairwise mixing stirred …
This paper provides an improved flamelet/progress variable (FPV) model for the simulation of turbulent combustion, employing the statistically most likely distribution (SMLD) approach for the joint probability density function (PDF) of the mixture fraction, Z, and of the progress parameter, Λ . Steady-state FPV models are built presuming the func- tional shape of the joint PDF of Z and Λ in order to evaluate Favre-averages of thermody- namic quantities. The mixture fraction is widely assumed to behave as a passive scalar with a mono-modal behaviour modelled by a \b{eta} -distribution. Moreover…
The theory of inertial manifolds (IM) is used to develop reduced-order models of turbulent combustion. In this approach, the dynamics of the system are tracked in a low-dimensional manifold determined in-situ without invoking laminar flame structures or statistical assumptions about the underlying turbulent flow. The primary concept in approximate IM (AIM) is that slow dominant dynamical behavior of the system is confined to a low-dimension manifold, and fast dynamics respond to the dynamics on the IM instantaneously. Decomposition of slow/fast dynamics and formulation of an AIM is accomplishe…
Assessment of turbulent fire dynamics and combustion instabilities using a flamelet model
University of Science and Technology of China（中国科学技术大学）；State Key Laboratory of Fire Science（火灾科学国家重点实验室）；University of Wuppertal（乌珀塔尔大学）；Technical University of Darmstadt（达姆施塔特工业大学）
The Lagrangian point of view is adopted to study turbulent premixed combustion. The evolution of the volume fraction of combustion products is established by the Reynolds transport theorem. It emerges that the burned-mass fraction is led by the turbulent particle motion, by the flame front velocity, and by the mean curvature of the flame front. A physical requirement connecting particle turbulent dispersion and flame front velocity is obtained from equating the expansion rates of the flame front progression and of the unburned particles spread. The resulting description compares favorably with…
In practical combustion systems, the high nonlinearity of the turbulent combustion process can result in unexpected behavior even for nominal operating conditions. This includes flame flashback in premixed gas turbines, engine unstart in scramjets, and failure of engines to ignite at high altitudes. Many of these events have catastrophic consequences. Hence, understanding the nature of such irregular excursions of combustion systems is important to ensure robust operation. The focus of this work is to identify the sources of such rare excursions from a dynamical systems perspective. Specifical…
The flamelet approach offers a viable framework for combustion modeling of homogeneous charge compression ignition (HCCI) engines under stratified mixture conditions. Scalar dissipation rate acts as a key parameter in flamelet-based combustion models which connects the physical mixing space to the reactive space. The aim of this paper is to gain fundamental insights into turbulent mixing in low temperature combustion (LTC) engines and investigate the modeling of scalar dissipation rate. Three direct numerical simulation (DNS) test cases of two-dimensional turbulent auto-ignition of a hydrogen-…
Distributed combustion, often associated with the low-oxygen condition, offers ultra-low NOX emission. However, it was recently achieved without combustion air dilution or internal flue gas recirculation, using a distinct approach called Mixture Temperature-Controlled combustion. Here, the fuel-air stream is cooled at the inlet to delay ignition and hence foster homogeneous mixture formation. The aim of this numerical study aims to understand the operation of this combustion concept better and present a robust framework for distributed combustion modeling in a parameter range where such operat…
This study performs a linear analysis of a turbulent reacting methane-air jet flame, with the goal of predicting the response of the reacting flow to upstream acoustic actuation. Accounting for heat release fluctuations is a vital component when investigating thermoacoustic instabilities and flame noise in a linearized framework. Unlike previous studies this work develops and applies an active flame approach, meaning the heat release oscillations of the flame resulting from the acoustic fluctuations are taken into account. To yield an active flame approach in the linear framework, a combustion…
The present work reports on LES of turbulent reacting flow of the Delft-Jet-in-Hot-Coflow (DJHC) burner, emulating MILD combustion, with transported PDF based combustion models using ANSYS FLUENT 13.0. Two different eddy viscosity models for LES (Dynamic Smagorinsky and Kinetic Energy Transport) along with two solution approaches for PDF transport equation, i.e. Eulerian and Lagrangian, have been used in the present study. Moreover, the effects of chemical kinetics and the micro-mixing models have also been investigated for two different fuel jet Reynolds number (Re = 4,100 and Re = 8,800). Th…
In order to explore the influence of combustion-induced thermal expansion on turbulence, a new research method is introduced. The method consists in jointly applying Helmholtz-Hodge decomposition and conditioned structure functions to analyzing turbulent velocity fields. Opportunities offered by the method are demonstrated by using it to process Direct Numerical Simulation data obtained earlier from two statistically 1D, planar, fully-developed, weakly turbulent, single-step-chemistry, premixed flames characterized by two significantly different (7.52 and 2.50) density ratios, with all other t…
Ignition and combustion behavior in the second stage of a sequential combustor are investigated numerically at atmospheric pressure for pure CH4 fueling and for a CH4/H2 fuel blend in 24:1 mass ratio using Large Eddy Simulation (LES). Pure CH4 fueling results in a turbulent propagating flame anchored by the hot gas recirculation zone developed near the inlet of the sequential combustion chamber. Conversely, CH4/H2 fueling results in a drastic change of the combustion process, with multiple auto-ignition kernels produced upstream of the main flame brush. Chemical Explosive Mode Analysis indicat…
LES/PDF approaches can be used for simulating challenging turbulent combustion configurations with strong turbulence chemistry interactions. Transported PDF methods are computationally expensive compared to flamelet-like turbulent combustion models. The pre-partitioned adaptive chemistry (PPAC) methodology was developed to address this cost differential. PPAC entails an offline preprocessing stage, where a set of reduced models are generated starting from an initial database of representative compositions. At runtime, this set of reduced models are dynamically utilized during the reaction frac…
This paper provides an extension of the standard flamelet progress variable (FPV) approach for turbulent combustion, applying the statistically most likely distribution (SMLD) framework to the joint PDF of the mixture fraction, Z, and the progress variable, C. In this way one does not need to make any assumption about the statistical correlation between Z and C and about the behaviour of the mixture fraction, as required in previous FPV models. In fact, for state-of-the-art models, with the assumption of very-fast-chemistry,Z is widely accepted to behave as a passive scalar characterized by a …
Flamelet Progress Variable (FPV) combustion models allow the evaluation of all thermo chemical quantities in a reacting flow by computing only the mixture fraction Z and a progress variable C. When using such a method to predict a turbulent combustion in conjunction with a turbulence model, a probability density function (PDF) is required to evaluate statistical averages (e.g., Favre average) of chemical quantities. The choice of the PDF is a compromise between computational costs and accuracy level. The aim of this paper is to investigate the influence of the PDF choice and its modeling aspec…
This paper experimentally analyzes the simultaneous influence of heat release and coherent vortex structures on the transfer of kinetic energy across scales around the laminar flame thickness $δ^0_\mathrm{L}$ in a turbulent premixed swirl flame and a non-reacting swirl flow. High-resolution tomographic particle image velocimetry and formaldehyde planar laser induced fluorescence measurements are used to obtain 3D velocity fields and estimates of the progress variable and density fields. The kinetic energy transfer across a filter scale of $Δ= 1.5 δ^0_\mathrm{L}$ was then quantified using physi…
Analysis of near wall flame and wall heat flux modeling in turbulent premixed combustion
UM-SJTU Joint Institute, Shanghai Jiao Tong University（上海交通大学密西根学院）；State Key Laboratory of Clean Energy Utilization, Zhejiang University（浙江大学清洁能源利用国家重点实验室）
This paper extends the methodology to use physics-informed enhanced super-resolution generative adversarial networks (PIESRGANs) for LES subfilter modeling in turbulent flows with finite-rate chemistry and shows a successful application to a non-premixed temporal jet case. This is an important topic considering the need for more efficient and carbon-neutral energy devices to fight the climate change. Multiple a priori and a posteriori results are presented and discussed. As part of this, the impact of the underlying mesh on the prediction quality is emphasized, and a multi-mesh approach is dev…
Jülich Supercomputing Centre, Forschungszentrum Jülich GmbH（于利希研究中心于利希超级计算中心）；RWTH Aachen University（亚琛工业大学）
Models for finite-rate-chemistry in underresolved flows still pose one of the main challenges for predictive simulations of complex configurations. The problem gets even more challenging if turbulence is involved. This work advances the recently developed PIESRGAN modeling approach to turbulent premixed combustion. For that, the physical information processed by the network and considered in the loss function are adjusted, the training process is smoothed, and especially effects from density changes are considered. The resulting model provides good results for a priori and a posteriori tests o…
This paper describes the numerical simulation of the NASA Langley Research Center supersonic H2 -Air combustion chamber performed using two approaches to model the presumed probability density function (PDF) in the flamelet progress variable (FPV) framework. The first one is a standard FPV model, built presuming the functional shape of the PDFs of the mixture fraction, Z, and of the progress parameter, Λ. In order to enhance the prediction capabilities of such a model in high-speed reacting flows, a second approach is proposed employing the statistically most likely distribution (SMLD) techcni…
The present paper deals with the numerical study of high pressure LOx/H2 or LOx/hydrocarbon combustion for propulsion systems. The present research effort is driven by the continued interest in achieving low cost, reliable access to space and more recently, by the renewed interest in hypersonic transportation systems capable of reducing time-to-destination. Moreover, combustion at high pressure has been assumed as a key issue to achieve better propulsive performance and lower environmental impact, as long as the replacement of hydrogen with a hydrocarbon, to reduce the costs related to ground …
This work describes and validates an approach for autonomously bifurcating turbulent combustion manifolds to divide regression tasks amongst specialized artificial neural networks (ANNs). This approach relies on the mixture of experts (MoE) framework, where each neural network is trained to be specialized in a given portion of the input space. The assignment of different input regions to the experts is determined by a gating network, which is a neural network classifier. In some previous studies, it has been demonstrated that bifurcation of a complex combustion manifold and fitting different A…
A hybrid lattice Boltzmann/finite-difference solver for low Mach thermo-compressible flows developed in earlier works is extended to more realistic and challenging configurations involving turbulence and complex geometries in the present article. The major novelty here as compared to previous contributions is the application of a more robust collision operator, considerably extending the stability of the original single relaxation time model and facilitating larger Reynolds number flow simulations. Additionally, a subgrid model and the thickened flame approach have also been added allowing for…
Nearly homogeneous and isotropic, highly turbulent flow, generated by an original multi-scale injector is experimentally studied. This multi-scale injector is made of three perforated plates shifted in space such that the diameter of their holes and their blockage ratio increase with the downstream distance. The Multi-Scale Turbulence Injector (hereafter, MuSTI) is compared with a Mono-Scale Turbulence Injector (MoSTI), the latter being constituted by only the last plate of MuSTI. This comparison is done for both cold and reactive flows. For the cold flow, it is shown that, in comparison with …
Since detonation is the only established theory that allows rapid burning producing a high pressure that can be sustained in open areas, the generally accepted opinion was that the mechanism explaining the high rate of combustion in dust explosions is deflagration-to-detonation transition. We propose a theoretical substantiation of an alternative mechanism explaining the origin of the secondary explosion producing high speeds of combustion and high overpressures in unconfined dust explosions. We show that the clustering of dust particles in a turbulent flow ahead of the advancing flame gives r…
This numerical study investigates the combustion modes in the second stage of a sequential combustor at atmospheric and high pressure. The sequential burner (SB) features a mixing section with fuel injection into a hot vitiated crossflow. Depending on the dominant combustion mode, a recirculation zone assists flame anchoring in the combustion chamber. The flame is located sufficiently downstream of the injector resulting in partially premixed conditions. First, combustion regime maps are obtained from 0-D and 1-D simulations showing the co-existence of three combustion modes: autoignition, fla…
The simulation of turbulent combustion phenomena is still an open problem in modern fluid dynamics. Considering the economical importance of hydrocarbon combustion in energy production processes, it is evident the need of an accurate tool with a relatively low computational cost for the prediction of this kind of reacting flows. In the present work, a comparative study is carried out among large eddy simulations, performed with various grid resolutions, a Reynolds averaged Navier-Stokes simulation, and experimental data concerning the well-known Sandia D flame test case. In all the simulations…
In this paper we report results of a numerical investigation of turbulent natural gas combustion for jet in a coflow of lean combustion products in the Delft-Jet-in-Hot-Coflow (DJHC) burner which emulates MILD (Moderate and Intense Low Oxygen Dilution) combustion behavior. The focus is on assessing the performance of the Eddy Dissipation Concept (EDC) model in combination with two-equation turbulence models and chemical kinetic schemes for about 20 species (Correa mechanism and DRM19 mechanism) by comparing predictions with experimental measurements. We study two different flame conditions cor…
North Carolina State University（北卡罗来纳州立大学）；ANSYS INC.（ANSYS公司）
Probability density function (PDF) based turbulent combustion modelling is limited by the need to store multi-dimensional PDF tables that can take up large amounts of memory. A significant saving in storage can be achieved by using various machine-learning techniques that represent the thermo-chemical quantities of a PDF table using mathematical functions. These functions can be computationally more expensive than the existing interpolation methods used for thermo-chemical quantities. More importantly, the training time can amount to a considerable portion of the simulation time. In this work,…
A purely data-driven approach using deep convolutional neural networks is discussed in the context of Large Eddy Simulation (LES) of turbulent premixed flames. The assessment of the method is conducted a priori using direct numerical simulation data. The network has been trained to perform deconvolution on the filtered density and the filtered density- progress variable product, and by doing so obtain estimates of the un-filtered progress variable field. Any non-linear function of the progress variable can then be approximated on the LES coarse mesh and explicitly filtered to advance the LES s…
Significant progress has been made on the model development for simulating turbulent reacting flows. As a consequence, we are currently in a position where key-physical aspects of fairly complex combustion processes are well understood at a qualitative and -- in many cases -- also at a quantitative level. Examples are the prediction of temperature and major species, statistically stationary flames, gas-phase combustion, turbulent transport, and turbulence/flame coupling. However, current challenges lie in capturing transient processes and stability boundaries, minor species and emissions, mult…
The complex interactions existing between turbulence and mixing in a bluff-body stabilised flame configuration is investigated by means of a wavelet cross-correlation analysis on Large Eddy Simulations. The combined approach allows to better point out typical features of unsteady turbulent flows with mixing through the characterisation of the processes involved both in time and scales. The wavelet cross-correlation analysis of the time signals of velocity and mixture fraction fluctuations can be an an effective tool to study the processes involved in turbulent mixing flows which are of great i…
This paper presents a load balancing strategy for reaction rate evaluation and chemistry integration in reacting flow simulations. The large disparity in scales during combustion introduces stiffness in the numerical integration of the PDEs and generates load imbalance during the parallel execution. The strategy is based on the use of the DLB library to redistribute the computing resources at node level, lending additional CPU-cores to higher loaded MPI processes. This approach does not require explicit data transfer and is activated automatically at runtime. Two chemistry descriptions, detail…
