Search

This paper presents a method for evaluating the risk of autoignition for the canonical problem of an enclosed hydrogen jet in crossflow (JICF), which is highly relevant to the design of mixing ducts. The proposed method is based on the separation of the underlying mixing pattern from the evolution of the chemical reactions, whereas the effect of mixing is maintained on the latter with the purpose of creating a reliable yet computationally efficient design tool for hydrogen gas turbines. Two variants of the incompletely stirred reactor network (ISRN) approach are proposed that provide the evolution of preignition radicals and autoignition kernel location, leveraging a nonreacting computational fluid dynamics solution or an analytical mixing pattern. The ISRN governing equations include all the salient features of hydrogen transport and lead to a conservative estimate of autoignition risk. Application to a few model problems with varied operating conditions suggests that radical buildup in the JICF can lead to autoignition in the vicinity of a most reactive mixture fraction, which is consistent with other laminar or turbulent hydrogen flows. However, the radical formation and autoignition kernel location strongly depend on the prediction of the underlying mixing field and the amount of differential diffusion within the JICF, which here primarily favors lower values of the composite mixture fraction and the transport of hydrogen and radicals away from the jet trajectory.

A lean azimuthal flame (LEAF) combustor fuelled with Jet-A1 is numerically simulated. The LEAF concept, motivated by the ``flameless'' oxidation principle, is based on azimuthally arranged fuel sprays and opposed injections of air in an axisymmetric chamber with a central outlet. This unique mixing configuration results in a highly turbulent toroidal reaction zone and achieves ultra-low emissions, making it attractive for future aero-engines. In this work, Incompletely Stirred Reactor (ISR) theory is employed to assess its capability in predicting the experimentally measured NOx concentration and soot particle size distribution (PSD) at the exhaust. Detailed fuel pyrolysis and oxidation chemistry, NOx formation models, and a detailed physicochemical sectional soot model are employed and solved in a post-processing fashion, leveraging a RANS/flamelet simulation with simple chemistry of a condition with thermal power of 20 kW. It is found that pollutant formation can be well reproduced in this burner, provided that an accurate energy balance is available to quantify the heat loss through the water-cooled aluminium frame and the air-cooled quartz windows present in the experiments. A parametric analysis showed that a departure from the adiabatic condition by even a few hundred W could substantially improve the agreement with measurements. In particular, heat losses were found to be responsible for a significant reduction in NOx emission, affected primarily by NNH and N2O routes, and an increase in soot emission at the burner exit, influenced mainly by finite-rate soot oxidation. Equivalently, it may be concluded that using realistic walls in the experiments, such as ceramic coatings that better reflect aviation gas turbines, would result in lower soot emission but higher NO emission. In this direction, the ISR approach and network approaches based on ISRs may be employed for preliminary evaluations and to identify suitable mixing patterns and residence times that ensure low pollutant formation across a wide operating range.

The effect of fuel droplets on the burning velocity of strained laminar premixed flames is investigated via experimentation and simulation. The twin counterflow configuration was used to obtain reference flame speeds as a function of strain rate for a prevaporized flame and a dilute spray flame simultaneously, both composed of acetone and air. The mixtures were varied with respect to nominal equivalence ratio (0.8–1.4) and strain rate (200–600 s−1). Gas velocities were measured using particle image velocimetry. Droplet size, velocity, and concentration were measured using phase Doppler anemometry: non-reacting flow measurements were taken at a position upstream of the stagnation plane; reacting flow measurements were taken along the stagnation streamline. The droplet Sauter mean diameter ranged between 65–75 µm, and the estimated fuel liquid fraction varied between 6–22%, increasing with nominal equivalence ratio. The results show that the reference flame speed of the spray flame decreases slightly relative to that of the prevaporized flame in the case of lean mixtures, but appears unchanged in the case of stoichiometric and rich mixtures. Gas velocity profiles and droplet measurements along the stagnation streamline suggest that the spray flame vaporization is incomplete, such that the reference flame speed corresponds to that of a lower equivalence ratio. Conversely, the opposing flame is affected by the fraction of droplets that do not vaporize in the spray flame and either penetrate the flame front, causing fuel enrichment, or evaporate near the stagnation plane, reducing the adiabatic flame temperature. The effect appears primarily for rich mixtures with higher liquid fraction, as evidenced by the lower reference flame speed and lower axial velocity rise across the flame. This study provides a systematic framework to examine the influence of fuel droplets on laminar flame propagation using a single-component fuel in the counterflow configuration., UK EPSRC grants EP/T015845/1 and EP/K035282/1. Initial work on the topic was funded by a Royal Society Newton Advanced Fellowship (NA160115) to C. T. Chong.

The study of autoignition propensity in premixers for gas turbines is critical for their safe operation and design. Although premixers can be analysed using reacting Computational Fluid Dynamics (CFD) coupled with detailed autoignition chemical kinetics, it is essential to also develop methods with lower computational cost to be able to explore more geometries and operating conditions during the design process. This paper presents such an approach based on Incompletely Stirred Reactor Network (ISRN) modelling. This method uses a CFD solution of a non-reacting flow and subsequently estimates the spatial evolution of reacting scalars such as autoignition precursors and temperature conditioned on the mixture fraction, which are used to quantify autoignition propensity. The approach is intended as a "post-processing'" step, enabling the use of very complex chemical mechanisms and the study of many operating conditions. For a representative premixer of an aeroderivative gas turbine, results show that autoignition propensity can be reproduced with ISRN at highly reactive operating conditions featuring multi-stage autoignition of a dual fuel mixture. The ISRN computations are consequently analysed to explore the evolution of reacting scalars and propose some autoignition metrics that combine mixing and chemical reaction to assist the design of premixers., Siemens Energy Canada Limited

The design and operation of premixers for gas turbines must deal with the possibility of relatively rare events causing dangerous autoignition. Rare autoignition events may occur in the presence of fluctuations of operational parameters, such as temperature and fuel composition, and must be understood and predicted. This work presents a methodology based on Incompletely Stirred Reactor (ISR) and surrogate modelling to increase efficiency and feasibility in premixer design optimisation for rare events. For a representative premixer, a space-filling design is used to sample the variability of three influential operational parameters. An ISR is then reconstructed and solved in a post-processing fashion for each sample, leveraging a well-resolved CFD solution of the non-reacting flow inside the premixer. Via detailed chemistry and reduced computational costs, the evolution of autoignition precursors and temperature, conditioned on a mixture fraction, is tracked, and accurate surrogate models are trained on all samples. The final quantification of the autoignition probability is achieved by querying the surrogate models via Monte Carlo sampling of the random parameters. The approach is fast and reliable so that user-controllable, independent variables can be optimised to maximise system performance while observing a constraint on the allowable probability of autoignition., Siemens Energy Canada Limited

Accurate predictions of soot emissions from combustion systems are required to implement the design of low-emission aero-engine combustors that can mitigate the effects of particulate matter on human health and the environment. The use of detailed models of soot formation can be unfeasible in terms of computational costs for optimisation procedures involving a large number of numerical simulations of different combustor configurations. A reduced-order formulation for turbulence-chemistry interactions and kinetic post-processing of Computational Fluid Dynamics (CFD) simulations, i.e., the Incompletely Stirred Reactors Network (ISRN) method, has recently provided promising qualitative predictions of soot emissions while allowing the use of complex chemistry at minimal computational costs. However, loss of accuracy and uncertainty in the predictions of relevant quantities, e.g., temperature and pollutant emissions, should be accounted for when reduced-order models like the ISRN method are employed. Hence, the integration of the ISRN method with data-driven approaches included in the framework of Uncertainty Quantification (UQ) has been pursued and is presented in this work. The grid parameters of the ISRN were calibrated via a UQ approach so that the predictions of temperature within an aero-engine model combustor match those obtained by a detailed CFD simulation with accuracy higher than 90\%. The UQ approach results in the determination of a feasible set of grid parameters and information about the correlation between them. Then, the proposed methodology has been applied on soot emissions in the aero-engine combustor to obtain bounded predictions from the ISRN method that are of the same order of magnitude as the corresponding ones provided by the high-order Conditional Moment Closure (CMC) combustion model., European Union���s Horizon 2020 project CoEC, grant agreement No 952181

This effort investigates flame topology, curvature, and flame surface density (FSD) measurements, using CH planar laser-induced fluorescence (PLIF) in prevaporized, premixed liquid fuel jet flames. A McKenna burner with a central jet port, fueled by methane, n-heptane, and n-dodecane is utilized in these experiments, using the flat flame as a methane-air pilot to maintain the stabilization of the jet flame. The measurements were conducted at Re = 10,000 at equivalence ratios of 0.8, 1, and 1.2. CH PLIF experiments were performed using the C-X band with excitation and fluorescence near 314 nm. A novel image processing methodology has been developed to capture flame structure and identify the product and reactant side of observed CH contours. This method allows for the determination of the sign of curvature to calculate curvature statistics and distributions for all cases studied. The effect of non-unity Le number is examined by analyzing fluorescence signal-curvature correlations for heavy hydrocarbon fuels compared to methane cases. The correlation reveals a relationship between the curvature of liquid fuel flames and CH fluorescence signals, particularly in high negative curvature regions. The signal-curvature correlation is stronger for larger Le fuels. In addition, the flame images are binarized to obtain a map of the average progress variable contours and flame surface density. Compared to OH PLIF, where the product and reactant sides are oriented and distinguished, CH PLIF requires more detailed binarization and side detection processing due to flame fragmentation and local extinctions. Observations of simultaneous OH-CH PLIF in these flames often show layers of extinguished CH while localized OH signal remains, which suggests the imaging of OH alone may not capture in-plane extinction and may lead to an over-prediction of flame surface density. Estimates of this over-prediction are determined by calculating flame surface density using artificially connected edges where a local extinction in CH images has occurred to simulate results from pseudo-OH PLIF imaging. Furthermore, the flame surface density, conditioned on local progress variable, is calculated for all cases comparing the effects of fueling, and equivalence ratio. Lastly, to determine potential differences in imaging methodologies between the CH PLIF and OH PLIF, the flame surface density is measured for pseudo-OH PLIF and is compared to the CH PLIF derived measurements.

A new image processing technique with topology tracking is introduced to calculate curvature statistics as extracted by CH radical planar laser-induced fluorescence (PLIF) measurements. The aim is to develop a capability to investigate non-unity Lewis number (Le ≠ 1) effects on curvature statistics and distributions in prevaporized, turbulent liquid fuel jet flames. Such flames typically exhibit significant levels of local extinction, so that the ability to determine the curvature sign from CH layer information is non-trivial if based only on the orientation of the CH layer relative to the reactant stream. This study proposes a simple topology tracking technique that allows for the calculation of curvature statistics on isolated reaction zones, classified based on pre-determined topology arguments. Binarized CH PLIF images are first utilized to distinguish between reaction zones and holes, pockets, or small isolated patches, which are generally not expected to alter curvature statistics. Consequently, a skeletonized model of the remaining reaction zones is employed and combined with the zone boundaries to assign each point of the reaction zone edge to the reactants or products stream. The latter allows for the correct sign and probability density functions of curvature to be evaluated. In this direction, curvature statistics are calculated here for a representative dataset of three prevaporized liquid fuel jet flames with Re = 10,000, equivalence ratio of 1.2, and a preheat temperature of 470 K. This work presents the first statistical curvature study using CH PLIF exclusively for liquid fuels.

Soot evolution in an ethylene lab-scale burner using the Rich-Quench-Lean (RQL) concept is investigated using Large-Eddy Simulation (LES) with the Conditional Moment Closure (CMC) method and a two-equation soot model. Three operating conditions are being analysed by varying the airflow provided in the burner's primary region and the dilution region while keeping the global equivalence ratio constant. The main objective is to study the effects of flame structure and residence time on soot evolution in a burner which reproduces the main features of the RQL concept. Results show the key differences in mixing field and soot tendency between the various operating conditions. A Lagrangian particle tracking analysis and evaluation of the solution in mixture fraction space highlight the significance of history effects and scalar dissipation rate on soot evolution, which should be taken into consideration for the design of next-generation low-emission aero-engines utilising the RQL concept., EPSRC, Rolls-Royce plc

The simulation of soot from swirl flames is a problem of relevance for the development of low-emission aero-engine combustors. Apart from detailed CFD approaches, it is of importance to also develop models with modest computational cost so parametric studies can be performed, especially in view of the need to predict engine-out soot particle size distributions to meet future regulations. In this work, the Incompletely Stirred Reactor (ISR) theory is extended to a reactors network formulation, based on the Conditional Moment Closure combustion model. An ISR is a volume V that is inhomogeneous in terms of mixture fraction but is characterised by homogeneous conditional averages, such as temperature and soot mass fraction conditional on the mixture fraction having a particular value. A network of ISRs is then deployed to separately capture soot production and oxidation regions exhibiting different degrees of micro-mixing rates and residence times, as typically observed in rich-burn swirl flames with additional dilution air. The Incompletely Stirred Reactor Network (ISRN) approach is demonstrated on a calculation of the Cambridge Rich-Quench-Lean combustor, where the residence time in the primary and secondary regions are controlled by the relative flow rates in the main annular swirl flow and the downstream dilution jets. The results show that the ISRN approach is computationally efficient and captures the experimentally observed soot tendency trends as a function of the burner operating conditions., EPSRC, Rolls-Royce plc

This study presents an approach for estimating fuel boil-off behaviour in cryogenic energy carrier ships, such as future liquid hydrogen (LH2) carriers. By relying on thermodynamic modelling and empirical formulas for ship motion and propulsion, the approach can be used to investigate boil-off as a function of tank properties, weather conditions, and operating velocities during a laden voyage. The model is first calibrated against data from a liquefied natural gas (LNG) carrier and is conse-quently used to investigate various design configurations of an LH2 ship. Results indicate that an LH2 ship with the same tank volume and glass wool insulation thickness as a conventional LNG carrier stores 40% of the fuel energy, and is characterised by a boil-off rate that is nine times higher and twice as sensitive to sloshing. Adding a reliquefaction unit can reduce the LH2 fuel depletion rate by at least 38.7% but can increase its variability with velocity and weather conditions. In calm weather, LH2 boil-off rates can only meet LNG carrier standards by utilising at least 6.6 times the insulation thickness. By adopting fuel cell propulsion in an LH2 ship, a 1.1% increase in fuel delivery is expected. An LH2 ship with fuel cells and reliquefaction is required to be at least 1.7 times larger than an existing LNG carrier to deliver the same energy. Further comparison of alternative scenarios indicates that LH2 carriers necessitate significant redesigns if the standards of LNG carriers are desired. The present approach can assist future feasibility studies featuring other vessels and pro-pulsion technologies and can be seen as an extendable framework that can predict boil-off in re-al-time., EPSRC Doctoral Training Grant EP/R513180/1

The simulation of soot evolution is a problem of relevance for the development of low-emission aero-engine combustors. Apart from detailed CFD approaches, it is of importance to develop models with modest computational cost so a large number of geometries can be explored, especially in view of the need to predict engine-out soot particle size distributions to meet future regulations. In this work, the Incompletely Stirred Reactor (ISR) theory is extended to a reactor network formulation, based on the Conditional Moment Closure (CMC) combustion model. An ISR is a volume V that is inhomogeneous in terms of mixture fraction but is characterized by homogeneous conditional averages, such as temperature and soot mass fraction conditioned on the mixture fraction having a particular value. A network of ISRs (ISRN) is then deployed to separately capture soot production and oxidation regions exhibiting different degrees of micro-mixing rates and residence times, as typically observed in practical combustors. The network is positioned and clustered around the mesh of a reference CFD simulation providing the ISRN with the average flow and mixing fields. Here, the ISRN approach is demonstrated on a single sector lean-burn model combustor operating on Jet A fuel and in pilot only mode, which corresponds to conditions of a lean-burn engine at part load. The results of the ISRN are compared with data previously obtained with LES-CMC coupled with detailed chemistry for dodecane and a two-equation model for soot. The comparison provides a satisfactory assessment of the method reliability, demonstrating the ability of the ISRN approach to replicate soot emission both in terms of mean location and magnitude while ensuring high computational efficiency. The ISRN approach can consequently be used to explore soot evolution utilizing complex chemical mechanisms and comprehensive soot models that otherwise would be intractable to couple with current CFD methods. In this direction, real fuel chemistry and a detailed physicochemical sectional soot model are combined to investigate the sensitivity of ISRN predictions.

Soot particle size distribution (PSD) evolution in an ethylene lab-scale swirl Rich-Quench-Lean (RQL) combustor is investigated using a detailed physicochemical sectional soot model coupled with the Conditional Moment Closure turbulent combustion model and Large Eddy Simulation. The aim is to develop predictive capability for the local soot PSD and to explore di erences in soot PSDs with di erent conditions of a burner con guration widely used in practice for emissions control. Two such conditions are studied by varying the air ow provided in the burner primary and dilution regions, which has a drastic e ect on soot emission as shown by previous experiments. The results show a reasonably good agreement with experiments for the mean reaction zone and soot locations and their variations with dilution. The predicted PSDs at the burner exit are fairly well captured for the high-dilution condition but show too few and too small particles for the dilution-free condition, which may be due to an overprediction of the oxidation rates or the unity Lewis number assumption used here that enhances penetration of soot towards the oxidiser stream. The results are analysed to reveal the hierarchy of reaction pathways during soot evolution and indicate how dilution air modi es the soot PSD within the primary zone. In particular, the presence of dilution leads to a broadly sustained uni-modal soot PSD shape and a decrease in particle size. The results demonstrate a framework for capturing soot PSD in realistic combustion devices, which may help meet future regulations based on particle number and size distribution., EPSRC Doctoral Training Award