Ammonia (NH3) combustion is an attractive energy source with zero carbon dioxide (CO2) emissions and a potential hydrogen-combustion enabler. In addition, the availability of ammonia and its amenability to transportation make it appealing to future energy supply. To provide fundamental insights into the heat release characteristics of ammonia in moderate or intense low oxygen dilution (MILD) combustion, one-dimensional laminar and two-dimensional turbulent premixed flames are studied numerically. In particular, the flame structure and heat release distribution of ammonia and ammonia-hydrogen premixed flames are examined under conventional and MILD, and preheated conditions, respectively. The results show that stoichiometric ammonia MILD flames differ from normal ammonia flames producing NH2 earlier, which results in the formation of a radical pool that shifts the chemical pathway towards hydrogen-related chemistry. The NH2 radical serves as a precursor for the ammonia flame and characterizes the reaction zone, and plays a crucial role in bringing the explosive nature into the system. For fixed ratios of the root-mean-square turbulent velocity fluctuation to the laminar flame speed and integral length scale to the laminar flame thickness, the flame surface wrinkling is comparable between the stoichiometric ammonia flame at MILD and non-MILD conditions. Furthermore, the hydrogen enrichment of the lean ammonia flame results in a significant increase in heat release rate due to the preferential diffusion effect. The study also examines chemical markers to identify heat-releasing regions of the flames and proposes the use of NH2 x O (on a mass fraction basis) as a suitable indicator of regions with high chemical activity and correlates well with the heat release for the MILD and non-MILD ammonia flames as well as ammonia-hydrogen flames at both laminar and turbulent conditions.

Hydrogen exhibits special burning characteristics such as fast laminar and, thereby, turbulent flame speed, and a wide flammability limit. Because of these features, the existing numerical models that have been developed for e.g., natural gas or fuels with unity Lewis number assumption could be limited or even unusable. This chapter discusses the models for turbulent flame speed and local displacement speed of pure lean hydrogen premixed flames, particularly for a wide range of turbulence levels. Moreover, the predictive capability of the probability density function (PDF) modeling adopting the widely used laminar flamelet concept for Reynolds averaged Navier–Stokes (RANS) or large-eddy simulation (LES) approaches is assessed a priori using a set of state-of-the-art direct numerical simulation (DNS) data. The general conclusion suggests that the main turbulent parameter dictating the turbulent flame speed is found to be the size of the most energy-containing eddies rather than non-dimensional numbers such as Reynolds (Re) or Karlovitz (Ka) numbers. The local displacement speed models also suggest that a model developed for moderate turbulence level (Ka $$\approx O(10)$$ ) predicts well flames with Ka $$> O$$ (1,000). PDF modeling using the flamelet concept is evaluated up to Ka $$> O(100)$$ , for which the mass fractions of major species are reasonably well predicted.

To obtain fundamental insights into the propagating behaviour of turbulent ammonia-air premixed flames, direct numerical simulations are carried out with complex chemistry for the flame-in-a-box configuration. The study compares the turbulent flame speeds of different mixtures at the same location in the Borghi–Peters diagram, including lean (equivalence ratio, ϕ= 0.81) and rich ( ϕ= 1.2) NH _3 -air, lean ( ϕ= 0.81) NH _3 -H _2 -N _2 -air and lean ( ϕ= 0.41) H _2 -air, as well as their respective equidiffusive counterparts. It is found that the lean NH _3 -H _2 -N _2 -air and H _2 -air mixtures have a higher level of turbulent flame speed enhancement than the lean and rich NH _3 -air flames. While the effect of the diffusive-thermal instability on mean turbulent flame speed is minor for NH _3 -air flames, it is notable for the hydrogen-containing ones. The location of the heat release peak and effective root-mean-square turbulent velocity ( u^' ) at the flame front are also found to influence the different normalized flame speeds for different fuels with similar nominal turbulence parameters. For the NH _3 -air flames the burning rate of the lean one is higher than that of the rich one, mainly because the effective u^' is larger for the lean flame, leading to greater flame front wrinkling. The impact of different turbulence conditions (Karlovitz number 30–557 and turbulent Reynolds number 36–386) on the behaviour of rich ammonia-air flames is also investigated, finding that the level of turbulent flame speed enhancement is closely linked to the size of the most energetic turbulent eddies. Additionally, the flame structure and the effect of the Lewis number are also examined, concluding that the latter is more pronounced in flames subjected to turbulence with a larger integral scale. The size of the integral length scale is a key factor in determining the level of flame wrinkling and distortion of the preheated zone, although the preheated zone is also affected in flames with a high Karlovitz number.

Local flame displacement speed Sd of a turbulent premixed flame is of fundamental and practical interest. For H2-air flames, the interest is further accentuated given the recent drive towards the development of zero-carbon combustors for both power and aircraft engine applications. The present study investigates several three-dimensional Direct Numerical Simulation (3D DNS) cases of premixed H2-air turbulent flames to theoretically model the Sd at negative curvatures, building upon recent works. Two of the four DNS cases presented are simulated at atmospheric pressure and two at elevated pressure. The DNS cases at different turbulence Reynolds numbers (Ret) and Karlovitz numbers (Ka) are generated using detailed chemistry. It has been shown in the previous studies that at atmospheric pressure, the density-weighted flame displacement speed Sd˜ is enhanced significantly over its laminar value (SL) at large negative curvature κ due to flame-flame interactions. The current work justifiably employs an imploding cylindrical laminar flame configuration to represent the local flame surfaces undergoing flame-flame interaction in a 3D turbulent flame. Therefore, to acquire a deep understanding of the interacting flame dynamics at large negative curvatures, one-dimensional (1D) simulations of an inwardly propagating cylindrical H2-air laminar premixed flame, with detailed chemistry at the corresponding atmospheric and elevated pressure conditions are performed. In particular, the 1D simulations emphasized the transient nature of the flame structure during these interactions. Based on the insights from the 1D simulations, we utilize an analytical approach to model the Sd˜ at these regions of extreme negative κ of the 3D DNS. The analytical approach is formulated to include the effect of variable density, convection and the inner reaction zone motion. The joint probability density function (JPDF) of Sd˜ and κ and the corresponding conditional averages obtained from 3D DNS showed clear negative correlation between Sd˜ and κ at all pressures. The obtained model successfully predicts the variation of 〈Sd˜|κ〉 with κ for the regions on the flame surface with large negative curvature (κδL≪−1) at atmospheric as well as at elevated pressure, with good accuracy. This showed that the 1D cylindrical, interacting flame model is a fruitful representation of a local flame-flame interaction that persists in a 3D turbulent flame, and is able to capture the intrinsically transient dynamics of the local flame-flame interaction. The 3D DNS cases further showed that even in the non-interacting state at κ=0, on average Sd˜ can deviate from SL. Sd˜ at κ=0 is a manifestation of the internal flame structure, controlled by turbulence transport in the large Ka regime. Therefore, the correlation of 〈Sd˜〉/SL with the the normalized gradient of the progress variable, 〈|∇c^|c0〉 at κ=0 is explored.

Comprehensive knowledge of local flame displacement speed, Sd, in turbulent premixed flames is crucial towards the design and development of hydrogen fuelled next-generation engines. Premixed hydrogen-air flames are characterized by significantly higher laminar flame speed compared to other conventional fuels. Furthermore, in the presence of turbulence, Sd is enhanced much beyond its corresponding unstretched, planar laminar value SL. In this study, the effect of high Karlovitz number (Ka) turbulence on density-weighted flame displacement speed, Sd˜, in a H2-air flame is investigated. Recently, it has been identified that flame-flame interactions in regions of large negative curvature govern large deviations of Sd˜ from SL, for moderately turbulent flames. An interaction model for the same has also been proposed. In this work, we seek to test the interaction model's applicability to intensely turbulent flames characterized by large Ka. To that end, we investigate the local flame structures: thermal, chemical structure, the effect of curvature, along the direction that is normal to the chosen isothermal surfaces. Furthermore, relative contributions of the transport and chemistry terms to Sd˜ are also analyzed. It is found that, unlike the moderately turbulent premixed flames, where enhanced Sd˜ is driven by interactions among complete flame structures, Sd˜ enhancement in high Ret and high Ka flame is predominantly governed by local interactions of the isotherms. It is found that enhancement in Sd˜ in regions of large negative curvature occurs as a result of these interactions, evincing that the interaction model is useful for high Ka turbulent premixed flames as well.

To understand the turbulence-chemistry interaction of lean hydrogen-air premixed flames at elevated pressures, we conduct a series of high-fidelity direct numerical simulations with detailed chemistry and transport by increasing pressure from 1 up to 7 atm. For a fixed ratio of the root-mean-square turbu-lent velocity fluctuation to the laminar flame speed, two sets of simulations were conducted: one for a fixed ratio of integral length scale to the laminar flame thickness and the other one for a fixed in-tegral length scale. We observe that the turbulent flame speed and volume-integrated heat release rate (HRR) increase with pressure due to the promoted chemical effects. To elucidate the enhanced chemical effects, we conduct a budget analysis of diffusive transport and reaction rate of hydrogen and examine the conditionally-averaged flame structure. As opposed to the laminar flames, the diffusive effects of hy-drogen for turbulent flames are found to increase with pressure in a more non-linear way, which signifi-cantly enhances the radical pool generation, leading to abundant H radical that is required for the rate of the pressure-sensitive reaction H + O2(+M) = HO2(+M) to be accelerated at higher pressures. Due to the strong diffusive transport of hydrogen for pressurized turbulence, not only does another small HRR peak emerge upstream, but also the location of the main HRR peak is moved towards the upstream region. Furthermore, the probability density function (PDF) of the flame curvature, computed for the isosurface associated with maximal diffusive transport of hydrogen, shows a clear trend of the PDF peak shifting to larger positive values as pressure increases. Simulations with the inclusion of thermal diffusion are also carried out at atmospheric and elevated pressure conditions to quantify the Soret effect. It is found that the magnitude of the hydrogen diffusion rate is augmented by thermal diffusion, affecting its mass fraction distribution. The mean turbulent flame speed is enhanced by approximately 8% and 7% for the 1 and 7 atm conditions, respectively.(c) 2022 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Compressible reacting flows may display sharp spatial variation related to shocks, contact discontinuities or reactive zones embedded within relatively smooth regions. The presence of such phenomena emphasizes the relevance of shock-capturing schemes such as the weighted essentially non-oscillatory (WENO) scheme as an essential ingredient of the numerical solver. However, these schemes are complex and have more computational cost than the simple high-order compact or non-compact schemes. In this paper, we present the implementation of a seventh-order, minimally-dissipative mapped WENO (WENO7M) scheme in a newly developed direct numerical simulation (DNS) code called KAUST Adaptive Reactive Flows Solver (KARFS). In order to make efficient use of the computer resources and reduce the solution time, without compromising the resolution requirement, the WENO routines are accelerated via graphics processing unit (GPU) computation. The performance characteristics and scalability of the code are studied using different grid sizes and block decomposition. The performance portability of KARFS is demonstrated on a variety of architectures including NVIDIA Tesla P100 GPUs and NVIDIA Kepler K20X GPUs. In addition, the capability and potential of the newly implemented WENO7M scheme in KARFS to perform DNS of compressible flows is also demonstrated with model problems involving shocks, isotropic turbulence, detonations and flame propagation into a stratified mixture with complex chemical kinetics. (C) 2020 Elsevier Ltd. All rights reserved.

This work was sponsored by King Abdullah University of Science and Technology (KAUST). Computational resources were provided by the KAUST Supercomputing Laboratory (KSL).

A statistical analysis is conducted for turbulent hydrogen-air premixed flames at a range of Karlovitz numbers up to 1,126 by direct numerical simulations (DNS) with detailed chemistry. The local and global burning velocities are evaluated and the deviation from the laminar flame speed is assessed. It is found that the global turbulent flame speed is largely determined by the integral length scale than the turbulent Karlovitz number, due to the flame surface area enhancement. The turbulent flame speed in all examined cases correlates well with the flame surface area, according to Damköhler's first hypothesis; even at Karlovitz number well above 1,000, reaction zones stay intact and only the preheat zone is broadened by the strong turbulence level. The statistical analysis with the probability density function (PDF) for the displacement speed shows that the highest probability of the local flame speed coincides with the one-dimensional unstretched flame speed. Despite some deviations, the mean flame structures and reaction rate of hydrogen of the higher Ka cases are found to resemble those of the laminar flame, and this further confirms that the turbulent flame brush topology is mainly determined by the large scale turbulence behavior. The results also suggest that the engineering modeling based on the flamelet concept may be valid for a wider range of Ka conditions.

Two-dimensional direct numerical simulation (DNS) databases of bluff-body-stabilized lean hydrogen flames representative of complicated reactive–diffusive system are analysed using the combined approach of computational singular perturbation (CSP) and tangential stretching rate (TSR) to investigate chemical characteristics in blow-off dynamics. To assess the diagnostic approaches in flame and blow-off dynamics, Damköhler number and TSR variables are applied and compared. Four cases are considered in this study showing different flame dynamics such as the steadily stable mode, local extinction by asymmetric vortex shedding, convective blow-off and lean blow-out. DNS data points in positive explosive eigenvalue conditions were subdivided into four different combinations in TSR and extended TSR space and categorized in four distinct characteristic regions, such as kinetically explosive or dissipative and transport-enhanced or dissipative dynamics. The TSR analysis clearly captures the local extinction point in the complicated vortex shedding and allows an improved understanding of the distinct chemistry-transport interactions occurring in convective blow-off and lean blow-out events.

To investigate the turbulent flame speed at high Karlovitz number (Ka) conditions, high fidelity direct numerical simulations (DNS) of lean hydrogen/air premixed flames propagating in a channel are performed with forced turbulence. The turbulent flame speed is analyzed with global and local perspectives. The global flame speed is evaluated from the fuel consumption rate while the local flame speed is computed from the displacement speed of the fuel species. It is found that for the global turbulent flame speed, the integral length scale plays a more important role rather than the turbulent intensity in that larger integral scales generate larger flame surface area which leads to the flame speed enhancement. The normalized flame speed is well correlated with the flame surface area, confirming that Damkhöler’s first hypothesis is still valid even at high Ka conditions up to Ka ≈ 700. Moreover, the local displacement speed with the statistic approach shows that the peak of the histogram of the displacement speed is found to nearly match the one computed from the one-dimensional laminar flame, implying that most of the turbulent flame elements burn like the laminar flame.

To provide fundamental insights into the ignition enhancement of methanol (MeOH) by the addition of the more reactive dimethyl ether (DME), computational parametric studies were conducted in a one-dimensional counterflow fuel versus air mixing layer configuration with the incorporation of detailed chemistry and transport. Various computational analysis tools based on the computational singular perturbation (CSP) framework were employed for detailed identifications of complex chemical pathways. CSP tools were also used to develop a 43-species skeletal mechanism for efficient computation of ignition of methanol-DME blends at engine conditions. The overarching practical question was the extent to which the addition of DME improves the ignitability of the methanol. As a baseline analysis, the results of a uniform temperature condition at 850 K showed that the low temperature chemistry associated with the DME fuel was highly effective in promoting autoignition. The increase in the oxidizer side temperature was found to diminish the ignition enhancement by DME blending, as the overall reactivity increases and the dominant chemical pathways become shifted towards the high temperature reactions. Finally, the strain rate effect on the ignition delay time was found to be significant for the pure methanol case, and then the effect diminishes as the amount of DME addition increases. This behavior was explained by examining the spatial locations of the ignition kernels and the Damkohler number history for different strain rate conditions. (C) 2018 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Ignition of fuel-air mixtures is an important issue in modern engine applications, such as homogeneous charge compression ignition (HCCI) or low-temperature combustion (LTC) engines, towards higher efficiency and lower emissions. These engines operate in compression ignition (CI) mode with early fuel injection, such that the overall combustion phasing behavior is mainly dictated by the autoignition of nearly uniform fuel-air mixture. The mixture composition and temperature distribution is not perfectly uniform, however, due to the incomplete fuel-air mixing, wall heat losses, and the presence of residual gases. Therefore, the first occurrence of autoignition is usually localized, and such ignition “kernels” [1] grow into the development of subsequent autoignition or flame propagation in neighboring mixtures. From a Lagrangian standpoint, the evolution of chemical activities within the ignition kernel may be described as a reactive mixture pocket subjected to temporal fluctuations in temperature and composition during the induction period, where the timescales of the fluctuations depend on the level of turbulence and initial stratifications. For this reason, the response of the igniting mixture in the presence of imposed oscillations in temperature and fuel composition has been studied in the past [2,3]. For example, Bansal et al. [2] showed that a harmonic oscillation of temperature can lead to a net advancement or retardation of autoignition of the mixture depending on the frequency and phasing of the oscillation. As an extension of these fundamental studies, the present work employs the computational singular perturbation (CSP) tools in order to provide a more detailed analysis of chemical pathways responsible for the autoignition of homogeneous mixture in the presence of temperature fluctuations.

Large eddy simulations of a turbulent premixed jet flame in a confined chamber were conducted using the flamelet generated manifold technique. The scope of the present research is to investigate the effects of inflow boundary conditions in these simulations. To evaluate the effect of two kinds of inflow boundary conditions, the numerical solutions using those boundary conditions are compared with experimental results. Turbulent flow generated by the self-recycling boundary condition behaves in a more realistic way, while randomly distributed perturbation imposed on the inlet boundary immediately vanished due to the lack of the turbulence structure. Quantitative comparison between experimental and computational results was carried out and analized.

Department of Mechanical Engineering, Pukyong National University, San 100, Yongdangdong, Nam-gu, Busan 608-739, Republic of Korea Environment & Energy Research Division, Korea Institute of Machinery and Materials, 171 Jang-dong, Yuseong-gu, Dajeon 305-343, Republic of Korea Power Generation Research Laboratory, Korea Electric Power Research Institute, 65, Munji-ro, Yuseong, Daejeon 305-760, Republic of Korea

The present study was conducted to investigate the flame instability(evaluated by Markstein length and cellular instability) and laminar burning velocity in a constant volume combustion chamber at room temperature and elevated pressure up to 0.3 MPa to suggest the possibility of utilizing mixtures of syngas added DME-air premixed flames in internal combustion engines. The experimentally measured laminar burning velocities were compared to predictions calculated the PREMIX code with Zhao reaction mechanism. Discussions were made on effects of syngas addition into DME-Air premixed flames through evaluating laminar burning velocity, Markstein length, and cellular instability. Particular concerns are focused on cellular instability caused by hy- drodynamic instability and diffusive-thermal instability.