This study investigates a hydrogen-fuelled burner representative of civil aviation applications, tested on a high-pressure optical facility. The configuration involves coaxial injection of hydrogen and swirled preheated air. Large-Eddy Simulations are performed for three operating conditions up to 13 bar using a moderately refined mesh and a 15-species mechanism. Results are validated against Particle Image Velocimetry, OH Planar Laser Induced Fluorescence, and OH & lowast; chemiluminescence, showing good agreement in flow structure, flame morphology, and NOx trends. A dedicated algorithm based on morphological criteria is employed to systematically identify reaction-zone topologies. This framework is used to analyse the flame response to variations in mass flow rate and pressure. Shear-layer regions are found to dominate fuel consumption despite lower local reactivity. Increasing mass flow rate promotes detached structures, while higher pressure favours stabilisation in hot recirculating regions. The proposed framework provides rare quantitative insight into hydrogen swirl flames at elevated pressure.

Fast and thin premixed hydrogen flames can lead to flashback scenarios which are unusual, especially for swirled configurations. Flashback can occur far from all walls, in the bulk flow, if the flow speed is less than the flame speed: this is a scenario which is usually avoided by increasing flow rates. However, flashback can also occur near walls where the flow speed goes to zero. Injector walls boundary layers always contain a zone where the local flow speed is less than the flame speed, even if the bulk flow velocity is large. The size of this zone is controlled by the velocity gradient at the wall which is the classical parameter used to predict flashbacks in boundary layers. In this study, flashback of lean hydrogen-air flames is computed using DNS (Direct Numerical Simulation) (flame resolved). Without swirl, results are compared and validated against experimental measurements and usual flashback criteria based on wall velocity gradient. DNS are also performed with swirl in a sector of an annular chamber, providing maps of flashback occurrence as function of swirl number and wall velocity gradient. Results show that swirl enhances flashback propensity and that thermodiffusive effects must be accounted to build a flashback criteria, indeed very lean HZ flames flashback for flow speeds higher than expected. Novelty and significance Almost all injection systems designed for hydrogen face a new, key issue in terms of operability: flashback. This study presents for the first time an analysis of the combined effects on flashback of the velocity gradients at the wall and of swirl. DNS of a swirling flow around a central body are performed and flashback maps are produced in a (swirl-velocity gradient) diagram of direct use for systems injecting lean premixed hydrogen-air mixtures.

Hydrogen (H-2) has become a key in the decarbonation process of the aeronautical field and new gaseous injection systems must be designed to suit hydrogen's specific properties (high laminar flame speed, high adiabatic temperature, large flammability limits and fast diffusivity (Le<1)). In this study, Large Eddy Simulation (LES) and Direct Numerical Simulation (DNS) are performed for a co-axial injector including an annular premixed hydrogen-air, swirled injection surrounding a central axial pure hydrogen lance. LES of the injector installed in a full chamber are performed at 12bar and the LES resolution issues associated with these high pressures are presented. LES reveals that the flame is stabilized on the rim separating the premixed and the pure hydrogen streams. The stabilization is produced by a structure called RSE for Rim Stabilized Edge Flame. To avoid resolution issues of the LES, the RSE flame structure is analyzed using CANTERA 1D flames and a 2D DNS of the attachment region. Extinction limits are described and implications for flame stability are discussed.

The use of hydrogen as a fuel is a promising way to reduce the emissions of civil aviation but it requires the development of wholly new injectors for the combustion chamber. Thanks to the increase in available computing power, the application of optimization techniques combined with CFD computations is now possible to develop these injectors. Among the optimization approaches, Bayesian optimization is particularly relevant when the objective functions and constraints of the optimization problem are expensive to evaluate which is the case in CFD-based optimization. Besides, the use of a multifidelity strategy allows to reduce the simulation cost of the Bayesian method. Therefore, this paper investigates the application of a multifidelity and multi-objective Bayesian approach to improve the performances of a laboratory swirl injector using hydrogen and operating in conditions close to industrial targets. This optimization study combines LES simulations as high-fidelity model with 2D RANS simulations as low-fidelity.

The use of hydrogen as a fuel is one of the most promising way to reduce the climate impact of civil aviation but it requires the development of wholly new injectors in the engine’s combustion chamber. Thanks to the increase in available computing power, new methodologies such as surrogate-based optimization, relying heavily on Computational Fluid Dynamics (CFD) along with automated optimization procedures, can be used to design these injectors. One important point is that 3D geometrical models and high-fidelity numerical approaches are often mandatory to capture key features of the multi-species, turbulent reactive flow through the combustion chamber. However, an optimization process only based on these approaches would be much too expensive. The multifidelity surrogate-based approach is a promising alternative and consists in using CFD simulations with various fidelities. However, its effectiveness depends on the correlation between the different levels of fidelity. The final, long run, goal of this work is the multifidelity optimization of H2-air injector for aircraft. In this paper, we focus on the very first steps of this work with the methodology presentation and a comparison of low (2D RANS) and intermediate (3D RANS) fidelities, which shows a fairly well agreement. Then, preliminary results of comparisons between intermediate and high (LES) fidelities, which highlight the need to improve modeling of RANS simulations, are discussed.

Soot prediction has become an important issue due to their impact on health, environment and combustion chamber thermal load.However, soot's evolution mechanisms are still not well understood and stay an open-field of study in computational fluid dynamics.Much soot modeling strategies of various level of complexity exist in literature and most of them having for objective to simulate industrial complexity devices use a very reduce number laminar flames to validate their model.The present work validate a soot modeling strategy, based on a Lagrangian soot tracking associated with a reversible sectional polycyclic aromatic hydrocarbons model, including several variants of gas phase description and soot evolution sub-models, against experimental measurements on a large set of 1D ethylene/air laminar premixed flames.An important variability of results was found and it was shown that some soot modeling strategy could give excellent agreements with experimental data on some flames and vastly overpredict or underpredict soot volume fraction on others.Among all the tested variations, one of them gives a better predictability over all the flames studied even if improvements for its related sub-models should be considered for future work.These results demonstrate the importance of validating one's soot modeling strategy over a sufficiently high number of configurations in order to highlight its main trends and defects.

The contribution of aircraft operations to ambient ultrafine particle (UFP) concentration at and around airports can be significant. This review article considers the volatile and non-volatile elements of particulate matter emissions from aircraft engines, their characteristics and quantification and identifies gaps in knowledge. The current state of the art emission inventory methods and dispersion modelling approaches are reviewed and areas for improvement and research needs are identified. Quantification of engine non-volatile particulate matter (nvPM) is improving as measured certification data for the landing and take-off cycle are becoming available. Further work is needed: to better estimate nvPM emissions during the full-flight; to estimate non-regulated (smaller) engines; and to better estimate the emissions and evolution of volatile particles (vPM) in the aircraft exhaust plume. Dispersion modelling improvements are also needed to better address vPM. As the emissions inventory data for both vPM and nvPM from aircraft sources improve, better estimates of the contribution of aircraft engine emissions to ambient particulate concentrations will be possible.

Soot formation has become an important issue in the design of gas turbine combustors due to its environmental impact and its contribution to radiative heat transfer in the combustion chamber. However, efficient and accurate prediction of soot particles formation, growth, oxidation and interaction in gas turbine combustors is still an open field in computational fluid dynamics. The present approach proposes to combine a reduced gas-phase chemistry, a sectional model for polycyclic aromatic hydrocarbons, and a Lagrangian description of soot particles dynamics. The Lagrangian description has been chosen for its ability to simulate the evolution of the particle size distribution. A numerical procedure is proposed to minimise its CPU cost. This approach was successfully applied to the simulation of steady laminar premixed ethylene-air flames at three fuel equivalence ratios, which constitutes a prerequisite towards its use in an aeronautical combustion chamber.

This work presents Large Eddy Simulations of the unconfined CORIA Rouen Spray Burner, fed with liquid n-heptane and air. Turbulent combustion modeling is based on the Filtered TAbulated Chemistry model for LES (F-TACLES) formalism, designed to capture the propagation speed of turbulent stratified flames. Initially dedicated to gaseous combustion, the filtered flamelet model is challenged for the first time in a turbulent spray flame configuration. Two meshes are employed. The finest grid, where both flame thickness and wrinkling are resolved, aims to challenge the chemistry tabulation procedure. At the opposite the coarse mesh does not allow full resolution of the flame thickness and exhibits significant unresolved contributions of subgrid scale flame wrinkling. Both LES solutions are extensively compared against experimental data. For both nonreacting and reacting conditions, the flow and spray aerodynamical properties are well captured by the two simulations. More interesting, the LES predicts accurately the flame lift-off height for both fine and coarse grid conditions. It confirms that the modeling methodology is able to capture the filtered turbulent flame propagation speed in a two-phase flow environment and within grid conditions representative of practical applications. Differences, observed for the droplet temperature, seem related to the evaporation model assumptions.

Turbulent combustion models for Large Eddy Simulation (LES) aims at predicting the flame dynamics. So far, they have been proven to predict correctly the mean flow and flame properties in a wide range of configurations. A way to challenge these models in unsteady situations is to test their ability to recover turbulent flames submitted to harmonic flow modulations. In this study, the Flame Transfer Function (FTF) of a CH4/H2/air premixed swirled-stabilized flame submitted to harmonic flowrate modulations in a non-adiabatic combustor is compared to the response computed using the Filtered TAbulated Chemistry for LES (F-TACLES) formalism. Phase averaged analysis of the perturbed flow field and flame response reveal that the velocity field determined with Particle Image Velocimetry measurements, the heat release distribution inferred from OH* images and the probability of presence of burnt gases deduced from OH-Planar Laser Induced Fluorescence measurements are qualitatively well reproduced by the simulations. However, noticeable differences between experiments and simulations are also observed in a narrow frequency range. A detailed close-up view of the flow field highlight differences in experimental OH* and numerical volumetric heat release rate distributions which are at the origin of the differences observed between the numerical and experimental FTF. These differences mainly originate from the outer shear layer of the swirling jet where a residual reaction layer takes place in the simulations which is absent in the experiments. Consequences for turbulent combustion modeling are suggested by examining the evolution of the perturbed flame brush envelope along the downstream distance of the perturbed flames. It is shown that changing the grid resolution and the flame subgrid scale wrinkling factor in these regions does not alter the numerical results. It is finally concluded that the combined effects of strain rate and enthalpy defect due to heat losses are the main factors leading to small but sizable differences of the flame response to coherent structures synchronized by the acoustic forcing in the outer shear layer of the swirling flow. These small differences in flame response lead in turn to a misprediction of the FTF at specific forcing frequencies.

The authors adapt recently developed a large eddy simulation / extended partially stirred reactor (LES/EPaSR) model by Sabelnikov and Fureby for simulation of turbulent combustion to Reynolds-averaged Navier–Stokes (RANS) equations. The proposed RANS/EPaSR model is validated against experimental database created at ONERA for an air–methane premixed flame stabilized by a backward-facing step combustor. The RANS/EPaSR model is compared also with the following RANSbased combustion models: (i) quasi-laminar model with reduced chemical mechanism (QL RCM); (ii) premixed flamelet tabulated chemistry (PFTC) without taking into account the turbulence–chemistry interaction (TCI); and (iii) a PFTC with a presumed β probability density function (PDF) for a progress combustion variable.

A strategy to simulate the soot formation in GT combustors is proposed: It is based on LES simulations using a tabulated chemistry and the soot model of Leung. The tabulated chemistry and the relative simplicity of the Leung model give rise to reasonable CPU times. In a first time, this strategy is applied to a burner working with gaseous fuel (ethylene) and experimentally studied by DLR. Good agreement between simulation and experiment is obtained for the temperature and the soot volume fraction. This strategy has been then applied to the TLC burner working with liquid kerosene and experimentally studied by ONERA and DLR. It was possible to recover by the calculation the shape of the soot volume fraction field obtained by the LII measurements. However, because of the temperature overestimate the quantitative level of soot given by the LES is too high compared to the experiment.

For a long time, engine noise has been dominated by fan and jet noise. With their reduction for modern turbojets, for instance, combustion noise is no longer negligible and efforts are concentrated on its understanding. The objective of the European project RECORD is to help to understand the fundamental mechanisms of core noise, in order to reduce it. In this paper, the recent advances achieved within the project for the noise generated through a nozzle are presented. This work includes the definition of a combustion test bench at the EM2C laboratory, equipped with various diagnostic techniques and an outlet nozzle to pressurize the chamber and generate entropy noise. Reactive simulations of this facility are performed by ONERA and CERFACS using LES. Finally, simulated flow fields are post-processed to determine direct (acoustic) and indirect (entropy) contributions to the global noise generated by the nozzle. This work is a first step before the modeling of a full engine, including the turbine stages.

This paper presents a high order finite volume scheme built on a new k-exact reconstruction algorithm for general unstructured grids. The algorithm is capable of computing accurate polynomial approximations using data from adjacent cells only, overcoming a major obstacle to extend classical finite volume schemes beyond 2 nd order spatial accuracy. Moreover, it can easily be integrated in a cell or vertex centered finite volume method that uses the cell averages as the only unknown per grid cell and physical quantity. It is therefore particularly suited to upgrade existing 2 nd order finite volume solvers to higher accuracy without huge efforts in software development. Three numerical test cases demonstrate the viability of the scheme in practical applications. This work was announced in [13, 14].

This paper presents a high order finite volume scheme built on a new k-exact reconstruction algorithm for general unstructured grids. The algorithm is capable of computing accurate polynomial approximations using data from adjacent cells only, overcoming a major obstacle to extend classical finite volume schemes beyond 2 nd order spatial accuracy. Moreover, it can easily be integrated in a cell or vertex centered finite volume method that uses the cell averages as the only unknown per grid cell and physical quantity. It is therefore particularly suited to upgrade existing 2 nd order finite volume solvers to higher accuracy without huge efforts in software development. Three numerical test cases demonstrate the viability of the scheme in practical applications. This work was announced in [13, 14].

A backward Lagrangian Monte Carlo modelling is proposed to calculate by post-processing the PDF of the thermo-chemical parameters of complex turbulent reactive flows simulated with a simple turbulent combustion model. PDF's of minor species such as pollutant species (NOx, soot, unburnt hydrocarbons...) can be easily obtained as long as these species have no significant influence on the main features of the flow. A numerical validation and an example of application of the method to a real burner are presented. If the number of points where information is sought is limited the cost of the method in terms of CPU time is very low and the statistical error can be perfectly controlled. With a first application to a semi-technical scale combustor producing soot the method has been proved very promising for the prediction of pollutant in complex turbulent reactive flows of gas turbine combustors.

Numerical simulation is currently used to help design low NOx devices for aircraft engines. In order to improve the prediction of pollutant species, combustion models based on tabulated chemistry have been commonly used in recent years. However, the short time scales of usual flamelet manifolds can lead to errors concerning the NO prediction in post flame area, as shown in this paper. Thus, a new method is proposed to extend the manifold in order to describe the evolution of species with a long characteristic time. This method is tested in the framework of 0D-1D computations. The use of the extended manifold is shown to be necessary to correctly predict the evolution of NO concentration in burnt gases. As an example, the extended and classical manifolds were compared in the framework of 2D simulations of an industrial-like combustor, showing an evident difference on the NO levels predicted in the outlet section.

The present study aims at contributing to the development of a methodology to predict and improve the ignition performances of aircraft combustors. A model has been developed to investigate the early growth of a spherical ignition kernel in a two-phase flow mixture. It has been combined with a multiphysic code through two different approaches. The ignition kernel model is used to build the ignition probability map of a combustor. The output of the model can also be introduced as an initial condition in an unsteady simulation to test the flame propagation in the combustor. To validate both methods, RANS and LES simulations have been performed on an experimental combustion chamber, reproducing one sector of an industrial combustor.

A numerical investigation of the overexpanded reactive gas flow in the ATAC nozzle of rectangular section, equipped with an H2 injection, is carried out. This study is aiming at reproducing the reacting phenomena occurring when air engulfed in the separated region at the tip of the extension meets the H2 rich overexpanded flow. Both steady Reynolds Averaged Navier Stokes (RANS) and unsteady Delayed Detached Eddy Simulation (DDES) approaches are used to model turbulence and compared to each other. The effect of the chemical nature of the mixture (either frozen or reacting) on the flow dynamics is discussed. The computations are compared to wall pressure measurements and flowfield visualizations by instantaneous Planar Laser Induced Fluorescence of OH (PLIF-OH) and OH spontaneous emission. A self-sustained flow oscillation at a frequency of around 1kHz is found numerically, in good agreement with the unsteady wall pressure measurements. An analysis of the space–time characteristics of the propagating disturbances contributes to a better understanding of this phenomenon.