Premixed turbulent combustion modeling using tabulated detailed chemistry and PDF

Tabulated chemistry and presumed probability density function (PDF) approaches are combined to perform RANS modeling of premixed turbulent combustion. The chemistry is tabulated from premixed flamelets with three independent parameters: the equivalence ratio of the mixture, the progress of reaction, and the specific enthalpy, to account for heat losses at walls. Mean quantities are estimated from presumed PDFs. This approach is used to numerically predict a turbulent premixed flame diluted by hot burnt products at an equivalence ratio that differs from the main stream of reactants. The investigated flame, subjected to high velocity fluctuations, has a thickened-wrinkled structure. A recently proposed closure for scalar dissipation rate that includes an estimation of the coupling between flame wrinkling and micromixing is retained. Comparisons of simulations with experimental measurements of mean velocity, temperature, and reactants are performed.     

Large eddy simulation of turbulent premixed combustion using tabulated detailed chemistry and presumed probability density function

A method of chemistry tabulation combined with presumed probability density function (PDF) is applied to simulate piloted premixed jet burner flames with high Karlovitz number using large eddy simulation. Thermo-chemistry states are tabulated by the combination of auto-ignition and extended auto-ignition model. To evaluate the predictive capability of the proposed tabulation method to represent the thermo-chemistry states under the condition of different fresh gases temperature, a-priori study is conducted by performing idealised transient one-dimensional premixed flame simulations. Presumed PDF is used to involve the interaction of turbulence and flame with beta PDF to model the reaction progress variable distribution. Two presumed PDF models, Dirichlet distribution and independent beta distribution, respectively, are applied for representing the interaction between two mixture fractions that are associated with three inlet streams. Comparisons of statistical results show that two presumed PDF models for the two mixture fractions are both capable of predicting temperature and major species profiles, however, they are shown to have a significant effect on the predictions for intermediate species. An analysis of the thermo-chemical state-space representation of the sub-grid scale (SGS) combustion model is performed by comparing correlations between the carbon monoxide mass fraction and temperature. The SGS combustion model based on the proposed chemistry tabulation can reasonably capture the peak value and change trend of intermediate species. Aspects regarding model extensions to adequately predict the peak location of intermediate species are discussed.     

Effects of subgrid scale and combustion modelling on flame structure of a turbulent premixed flame within LES and tabulated chemistry framework

The effects of combustion and SubGrid Scale (SGS) modelling on the overall flame characteristics of a turbulent premixed flame are investigated. This is achieved in terms of mean flow statistics, variances and flame surfaces. In particular, the chemical flame structure is analysed and compared. The Artificially Thickened Flame (ATF) approach coupled with the Flamelet Generated Manifolds (FGMs) and Filtered TAbulated Chemistry for LES (F-TACLES) approaches are used for this investigation. A Germano like procedure for dynamical calculation of SGS wrinkling is used which ensures the conservation of the total flame surface for both models. It turns out that using the dynamic SGS wrinkling model improves the results. Although the results of both combustion models in terms of statistics, mean and variances show very good agreement, the resolved flame surfaces hide different dynamic behaviour.     

Modelling compression ignition engines by incorporation of the flamelet generated manifolds combustion closure

Tabulated chemistry models allow to include detailed chemistry effects at low cost in numerical simulations of reactive flows. Characteristics of the reactive fluid flows are described by a reduced set of parameters that are representative of the flame structure at small scales so-called flamelets. For a specific turbulent combustion configuration, flamelet combustion closure, with proper formulation of the flame structure can be applied. In this study, flamelet generated manifolds (FGM) combustion closure with progress variable approach were incorporated with OpenFOAM® source code to model combustion within compression ignition engines. For IC engine applications, multi-dimensional flamelet look-up tables for counter flow diffusive flame configuration were generated. Source terms of non-premixed combustion configuration in flamelet domain were tabulated based on pressure, temperature of unburned mixture, mixture fraction, and progress variable. A new frozen flamelet method was introduced to link one dimensional reaction diffusion space to multi-dimensional Computational Fluid Dynamics (CFD) physical space to fulfill correct modelling of thermal state of the engine at expansion stroke when charge composition was changed after combustion and reaction rates were subsided. Predictability of the developed numerical framework were evaluated for Sandia Spray A (constant volume vessel), Spray B (light duty optical Diesel engine), and a heavy duty Diesel engine experiments under Reynolds averaged Navier Stokes turbulence formulation. Results showed that application of multi-dimensional FGM combustion closure can comprehensively predict key parameters such as: ignition delay, in-cylinder pressure, apparent heat release rate, flame lift-off , and flame structure in Diesel engines.     

Modelling alkali metal emissions in large-eddy simulation of a preheated pulverised-coal turbulent jet flame using tabulated chemistry

The numerical modelling of alkali metal reacting dynamics in turbulent pulverised-coal combustion is discussed using tabulated sodium chemistry in large eddy simulation (LES). A lookup table is constructed from a detailed sodium chemistry mechanism including five sodium species, i.e. Na, NaO, NaO₂, NaOH and Na₂O₂H₂, and 24 elementary reactions. This sodium chemistry table contains four coordinates, i.e. the equivalence ratio, the mass fraction of the sodium element, the gas-phase temperature, and a progress variable. The table is first validated against the detailed sodium chemistry mechanism by zero-dimensional simulations. Then, LES of a turbulent pulverised-coal jet flame is performed and major coal-flame parameters compared against experiments. The chemical percolation devolatilisation (CPD) model and the partially stirred reactor (PaSR) model are employed to predict coal pyrolysis and gas-phase combustion, respectively. The response of the five sodium species in the pulverised-coal jet flame is subsequently examined. Finally, a systematic global sensitivity analysis of the sodium lookup table is performed and the accuracy of the proposed tabulated sodium chemistry approach has been calibrated.     

LES based investigation of autoignition in turbulent co-flow configurations

The impact of turbulence on the autoignition of a diluted hydrogen jet in a hot co-flow of air is studied numerically. The LES combustion model used is successfully validated against experimental measurements and 3D DNS. Parametric studies are then carried out by separately varying turbulent intensity and integral length scale in the co-flow, while keeping all other boundary conditions unchanged. It is found that the impact of turbulence on the location of autoignition is non-trivial. For weak to mild turbulence, with a turbulent time scale larger than the minimum ignition delay time, autoignition is facilitated by increased turbulence. This is due to enhanced mixing between fuel and air, creating larger most reactive mixture fraction regions. On the other hand, for turbulent time scales smaller than the ignition delay time, the increased scalar dissipation rate dominates over the effect of increased most reactive mixture fraction regions, which leads to a rise in the autoignition length. Turbulence–chemistry interaction mechanisms are analysed in order to explain these observations.     

A comparison of different approaches to integrate flamelet tables with presumed-shape PDF in flamelet models for turbulent flames

Flamelet models for turbulent combustion modelling make use of presumed-shape probability density functions (PDFs) for integrating laminar flamelet solutions to obtain an integrated flamelet table that can readily be used for turbulent flame calculations. The existence of non-unique approaches for such an integration has rarely been investigated before. For the first time, this work studies systematically the non-uniqueness of the flamelet table integration approaches. A flamelet model called the flamelet/progress variable model is used in the study, although the issue exists generally in many other flamelet models. Two classes of table integration approaches are investigated, one preserving the laminar flamelet structures during integration and the other not. Three different table integration approaches are examined and compared in detail to provide a thorough understanding of the different approaches. A partially stirred reactor is used as a test case for examining the different approaches. A method based on the transported PDF method is also employed to provide a reference for the assessment of the different flamelet table integration approaches. It is found in general that the flamelet preserving integration approach yields a more reasonable joint PDF of the mixture fraction and the progress variable, and the prediction results are closer to the referenced transported PDF results.     

Turbulent flame simulation taking advantage of tabulated chemistry self-similar properties

Predicting the flame shape, its stabilization process, and pollutant emissions in practical combustion devices requires to incorporate complex chemistry features. As detailed chemical schemes are too voluminous for practical numerical simulations, tabulated chemistry techniques have been proposed to account for the complexity of kinetics in turbulent flame simulations. Unfortunately, the size of these databases may become a crucial issue for efficient implementation on massively parallel computers. A reduction strategy that takes advantage of self-similar properties of tabulated chemistry is proposed for turbulent combustion modeling. A reduction of the database size by a factor of 1000 is achieved. This procedure is successfully applied to a RANS simulation of a turbulent jet flame.     

An abstraction layer for efficient memory management of tabulated chemistry and flamelet solutions

A large number of methods for simulating reactive flows exist, some of them, for example, directly use detailed chemical kinetics or use precomputed and tabulated flame solutions. Both approaches couple the research fields computational fluid dynamics and chemistry tightly together using either an online or offline approach to solve the chemistry domain. The offline approach usually involves a method of generating databases or so-called Lookup-Tables (LUTs). As these LUTs are extended to not only contain material properties but interactions between chemistry and turbulent flow, the number of parameters and thus dimensions increases. Given a reasonable discretisation, file sizes can increase drastically. The main goal of this work is to provide methods that handle large database files efficiently. A Memory Abstraction Layer (MAL) has been developed that handles requested LUT entries efficiently by splitting the database file into several smaller blocks. It keeps the total memory usage at a minimum using thin allocation methods and compression to minimise filesystem operations. The MAL has been evaluated using three different test cases. The first rather generic one is a sequential reading operation on an LUT to evaluate the runtime behaviour as well as the memory consumption of the MAL. The second test case is a simulation of a non-premixed turbulent flame, the so-called HM1 flame, which is a well-known test case in the turbulent combustion community. The third test case is a simulation of a non-premixed laminar flame as described by McEnally in 1996 and Bennett in 2000. Using the previously developed solver ‘flameletFoam’ in conjunction with the MAL, memory consumption and the performance penalty introduced were studied. The total memory used while running a parallel simulation was reduced significantly while the CPU time overhead associated with the MAL remained low.     

Non-premixed turbulent combustion modeling based on the filtered turbulent flamelet equation

In turbulent combustion simulations, the flow structure at the unresolved scale level needs to be reasonably modeled. Following the idea of turbulent flamelet equation for the non-premixed flame case, which was derived based on the filtered governing equations(L. Wang, Combust. Flame 175, 259(2017)), the scalar dissipation term for tabulation can be directly computed from the resolved flowing quantities, instead of solving species transport equations. Therefore, the challenging source term closure for the scalar dissipation or any assumed probability density functions can be avoided;meanwhile the chemical sources are closed by scaling relations. The general principles are discussed in the context of large eddy simulation with case validation. The new model predictions of the bluff-body flame show sufficiently improved results, compared with these from the classic progress-variable approach.     

The impact of reduced chemistry on auto-ignition of H2 in turbulent flows

The auto-ignition behaviour of hydrogen in a turbulent flow field has been studied through a combination of detailed and systematically reduced chemistry with a transported PDF approach closed at the joint-scalar level. Radiation is accounted for through the RADCAL method and the inclusion of enthalpy into the joint-scalar PDF. Molecular mixing is closed using the modified Curl's model with the mixing frequency accounted for via two algebraic closures. The main aim of the work is to compare the impact of alternative chemical mechanisms on auto-ignition and to explore the accuracy that can be expected when reactive scalars are sequentially removed through the application of quasi-steady-state approximations (QSSAs). Two different detailed mechanisms were tested to establish the effects of intrinsic uncertainties in the detailed chemistry and to provide reference points to past work. The mechanisms feature nine solved species and 19 or 20 reversible chemical reactions. The chemical mechanisms were subsequently systematically reduced to five, four and three independent scalars through the successive introduction of QSSAs for H₂O₂, HO₂ and O. Resulting inaccuracies were quantified following each simplification step with reference to experimental data obtained in shock tubes and under turbulent flow conditions in the Cabra burner configuration. A sensitivity analysis was also performed to identify the relative impact of uncertainties in key reactions as compared to systematic simplification process. It was found that alternative recommended rates for the O + H₂ = OH + H reaction have an impact on the point of flame stabilization that is similar to that observed as a consequence of the simplification process. The work also shows that realistic results can be obtained with simplified chemistry. However, it is also concluded that the temporal evolution of the radical pool and the point of stabilization is affected by the introduction of a QSSA for the O radical. Furthermore, it is shown by comparisons with time resolved OH radical data obtained in shock tubes that the progressive elimination of species via QSSA leads to a shortening of ignition delay times and that the same effects are present, but less severe, in turbulent flow fields.     

An a priori study of different tabulation methods for turbulent pulverised coal combustion

In many practical pulverised coal combustion systems, different oxidiser streams exist, e.g. the primary- and secondary-air streams in the power plant boilers, which makes the modelling of these systems challenging. In this work, three tabulation methods for modelling pulverised coal combustion are evaluated through an a priori study. Pulverised coal flames stabilised in a three-dimensional turbulent counterflow, consisting of different oxidiser streams, are simulated with detailed chemistry first. Then, the thermo-chemical quantities calculated with different tabulation methods are compared to those from detailed chemistry solutions. The comparison shows that the conventional two-stream flamelet model with a fixed oxidiser temperature cannot predict the flame temperature correctly. The conventional two-stream flamelet model is then modified to set the oxidiser temperature equal to the fuel temperature, both of which are varied in the flamelets. By this means, the variations of oxidiser temperature can be considered. It is found that this modified tabulation method performs very well on prediction of the flame temperature. The third tabulation method is an extended three-stream flamelet model that was initially proposed for gaseous combustion. The results show that the reference gaseous temperature profile can be overall reproduced by the extended three-stream flamelet model. Interestingly, it is found that the predictions of major species mass fractions are not sensitive to the oxidiser temperature boundary conditions for the flamelet equations in the a priori analyses.     

A mixing controlled direct chemistry (MCDC) model for diesel engine combustion modelling using large eddy simulation

A mixing controlled direct chemistry (MCDC) combustion model with sub-grid scale (SGS) mixing effects and chemical kinetics has been evaluated for Large Eddy Simulation (LES) of diesel engine combustion. The mixing effect is modelled by a mixing timescale based on mixture fraction variance and sub-grid scalar dissipation rate. The SGS scalar dissipation rate is modelled using a similarity term and a scaling factor from the analysis of Direct Numerical Simulation (DNS) data. The chemical reaction progress is estimated from a kinetic timescale based on local internal energy change rate and equilibrium state internal energy. An optical research engine operating at conventional operating conditions and Low Temperature Combustion (LTC) conditions was used for evaluation of the combustion model. From the simulation results, the effect of SGS scalar mixing is evaluated at different stages of combustion. In the context of LES, the new approach provides improved engine modelling results compared to the Direct Chemistry Solver (DCS) combustion model.     

Modelling of the correlation between velocity and reactive scalar gradients in turbulent premixed flames based on DNS data

The interaction between turbulence and reactive scalar fields is discussed for the wrinkled flamelets regime of turbulent premixed combustion. Emphasis is placed on the effects associated with the turbulent straining term. In the regime of turbulent combustion under consideration, which corresponds to Karlovitz and Damköhler numbers such that Ka < 1 and Da > 1, a clear and simple formulation is proposed to explain and to model the influence of the correlation between velocity and reactive scalar gradients. This formulation is based on the conservative variables budget across one-dimensional premixed laminar flamelets. The analysis firmly confirms the dependence on both the Damköhler number and the expansion factor, a feature already foreseen in recent studies. Nevertheless, in contrast with previous work, (i) the scaling arguments used in the present contribution are different from those used in other recent proposals, and (ii) the proposed closures are not only deduced from dimensional arguments but also from the consideration of conservative variable budgets across laminar flamelets. The resulting functional dependence on the expansion factor is found to be influenced by the underlying one-dimensional flamelet representation and two possible closures are put forward to take this dependence into account. (iii) The two closures do not exhibit a proportionality to the mean scalar dissipation rate as suggested in previous studies but to the square of this quantity. This results in the presence of a second contribution proportional to in the modelled transport equation for the mean scalar dissipation rate, in addition to the modelled molecular dissipation term. (iv) Since previous Direct Numerical Simulation (DNS) studies have been essentially devoted to the influence of the Damköhler number, the present DNS validation step is focused on the effects of the expansion rate. To this purpose, the proposed models are validated against three available DNS databases obtained for turbulent premixed flames with different values of the density ratio between unburned and fully burned gases.     

Evaluation of the flame propagation within an SI engine using flame imaging and LES

This work shows experiments and simulations of the fired operation of a spark ignition engine with port-fuelled injection. The test rig considered is an optically accessible single cylinder engine specifically designed at TU Darmstadt for the detailed investigation of in-cylinder processes and model validation. The engine was operated under lean conditions using iso-octane as a substitute for gasoline. Experiments have been conducted to provide a sound database of the combustion process. A planar flame imaging technique has been applied within the swirl- and tumble-planes to provide statistical information on the combustion process to complement a pressure-based comparison between simulation and experiments. This data is then analysed and used to assess the large eddy simulation performed within this work. For the simulation, the engine code KIVA has been extended by the dynamically thickened flame model combined with chemistry reduction by means of pressure dependent tabulation. Sixty cycles have been simulated to perform a statistical evaluation. Based on a detailed comparison with the experimental data, a systematic study has been conducted to obtain insight into the most crucial modelling uncertainties.     

Application and theoretical analysis of the flamelet model for supersonic turbulent combustion flows in the scramjet engine

In the framework of Reynolds-averaged Navier–Stokes simulation, supersonic turbulent combustion flows at the German Aerospace Centre (DLR) combustor and Japan Aerospace Exploration Agency (JAXA) integrated scramjet engine are numerically simulated using the flamelet model. Based on the DLR combustor case, theoretical analysis and numerical experiments conclude that: the finite rate model only implicitly considers the large-scale turbulent effect and, due to the lack of the small-scale non-equilibrium effect, it would overshoot the peak temperature compared to the flamelet model in general. Furthermore, high-Mach-number compressibility affects the flamelet model mainly through two ways: the spatial pressure variation and the static enthalpy variation due to the kinetic energy. In the flamelet library, the mass fractions of the intermediate species, e.g. OH, are more sensible to the above two effects than the main species such as H₂O. Additionally, in the combustion flowfield where the pressure is larger than the value adopted in the generation of the flamelet library or the conversion from the static enthalpy to the kinetic energy occurs, the temperature obtained by the flamelet model without taking compressibility effects into account would be undershot, and vice versa. The static enthalpy variation effect has only little influence on the temperature simulation of the flamelet model, while the effect of the spatial pressure variation may cause relatively large errors. From the JAXA case, it is found that the flamelet model cannot in general be used for an integrated scramjet engine. The existence of the inlet together with the transverse injection scheme could cause large spatial variations of pressure, so the pressure value adopted for the generation of a flamelet library should be fine-tuned according to a pre-simulation of pure mixing.     

Acceleration of the chemistry solver for modeling DI engine combustion using dynamic adaptive chemistry (DAC) schemes

Acceleration of the chemistry solver for engine combustion is of much interest due to the fact that in practical engine simulations extensive computational time is spent solving the fuel oxidation and emission formation chemistry. A dynamic adaptive chemistry (DAC) scheme based on a directed relation graph error propagation (DRGEP) method has been applied to study homogeneous charge compression ignition (HCCI) engine combustion with detailed chemistry (over 500 species) previously using an R-value-based breadth-first search (RBFS) algorithm, which significantly reduced computational times (by as much as 30-fold). The present paper extends the use of this on-the-fly kinetic mechanism reduction scheme to model combustion in direct-injection (DI) engines. It was found that the DAC scheme becomes less efficient when applied to DI engine simulations using a kinetic mechanism of relatively small size and the accuracy of the original DAC scheme decreases for conventional non-premixed combustion engine. The present study also focuses on determination of search-initiating species, involvement of the NO_x chemistry, selection of a proper error tolerance, as well as treatment of the interaction of chemical heat release and the fuel spray. Both the DAC schemes were integrated into the ERC KIVA-3v2 code, and simulations were conducted to compare the two schemes. In general, the present DAC scheme has better efficiency and similar accuracy compared to the previous DAC scheme. The efficiency depends on the size of the chemical kinetics mechanism used and the engine operating conditions. For cases using a small n-heptane kinetic mechanism of 34 species, 30% of the computational time is saved, and 50% for a larger n-heptane kinetic mechanism of 61 species. The paper also demonstrates that by combining the present DAC scheme with an adaptive multi-grid chemistry (AMC) solver, it is feasible to simulate a direct-injection engine using a detailed n-heptane mechanism with 543 species with practical computer time.     

Computational study of lean premixed turbulent flames using RANSPDF and LESPDF methods

A computational study is performed on a series of four piloted, lean, premixed turbulent jet flames. These flames use the Sydney Piloted Premixed Jet Burner (PPJB), and with jet velocities of 50, 100, 150 and 200 m/s are denoted PM150, PM1100, PM1150 and PM1200, respectively. Calculations are performed using the RANSPDF and LESPDF methodologies, with different treatments of molecular diffusion, with detailed chemistry and flamelet-based chemistry modelling, and using different imposed boundary conditions. The sensitivities of the calculations to these different aspects of the modelling are compared and discussed. Comparisons are made to experimental data and to previously-performed calculations. It is found that, given suitable boundary conditions and treatment of molecular diffusion, excellent agreement between the calculations and experimental measurements of the mean and variance fields can be achieved for PM150 and PM1100. The application of a recently developed implementation of molecular diffusion results in a large improvement in the computed variance fields in the LESPDF calculations. The inclusion of differential diffusion in the LESPDF calculations provides insight on the behaviour in the near-field region of the jet, but its effects are found to be confined to this region and to the species CO, OH and H₂. A major discrepancy observed in many previous calculations of these flames is an overprediction of reaction progress in PM1150 and PM1200, and this discrepancy is also observed in the LESPDF calculations; however, a parametric study of the LESPDF mixing model reveals that, with a sufficiently large mixing frequency, calculations of these two flames are capable of yielding improved reaction progress in good qualitative agreement with the mean and RMS scalar measurements up to an x/D of 30. Lastly, the merits of each computational methodology are discussed in light of their computational costs.