Insight Into Supersonic Combustion Using High-order Simulations

High-order simulations of supersonic combustion are presented to advance understanding of the complex chemically-reacting flow processes and identify unknown mechanisms of the high-speed combustion process. We have employed 11th-order accurate implicit large-eddy simulations in conjunction with thermochemistry models comprising 20 chemical reactions. We compare the computations with available experiments and discuss the accuracy and uncertainties in both. Jets emanating from above and below the hydrogen plumes influence the combustion process and accuracy of the predictions. The simulations reveal that high temperatures are sustained for a long-distance downstream of the combustion onset. A barycentric map for the Reynolds stresses is employed to analyse the turbulent anisotropy. We correlate the axisymmetric contraction and expansion of turbulence with the interaction of reflected-shock waves with the supersonic combustion hydroxyl production regions. The physics insights presented in this study could potentially lead to more efficient supersonic combustion and scramjet technologies.

Application of Machine Learning in Turbulent Combustion for Aviation Gas Turbine Combustor Design

Numerical simulation of gas turbine combustors requires resolving a broad spectrum of length and time scales for accurate flow field and emission predictions. Reynold’s Averaged Navier Stokes (RANS) approach can generate solutions in few hours; however, it fails to produce accurate predictions for turbulent reacting flow field seen in general combustors. On the other hand, the Large Eddy Simulation (LES) approach can overcome this challenge, but it requires orders of magnitude higher computational cost. This limits designers to use the LES approach in combustor development cycles and prohibits them from using the same in numerical optimization. The current work tries to build an alternate approach using a data-driven method to generate fast and consistent results. In this work, deep learning (DL) dense neural network framework is used to improve the RANS solution accuracy using LES data as truth data. A supervised regression learning multilayer perceptron (MLP) neural network engine is developed. The machine learning (ML) engine developed in the present study can compute data with LES accuracy in 95% lesser computational time than performing LES simulations. The output of the ML engine shows good agreement with the trend of LES, which is entirely different from RANS, and to a reasonable extent, captures magnitudes of actual flow variables. However, it is recommended that the ML engine be trained using broad design space and physical laws along with a purely data-driven approach for better generalization.

Skeletal Kinetic Modeling for the Combustion of Endothermic Hydrocarbon Fuel in Hypersonic Vehicle

Chemical kinetic mechanism plays a vital role in the deep learning of reacting flow in practical combustors, which can help obtain many details of the combustion process. In this paper, a surrogate model and a skeletal mechanism for an endothermic hydrocarbon fuel were developed for further investigations of the combustion performance in hypersonic vehicles: (1) The surrogate model consists of 81.3 mol% decalin and 18.7 mol% n-dodecane, which were determined by both the composition distributions and key properties of the target endothermic hydrocarbon fuel. (2) A skeletal kinetic mechanism only containing 56 species and 283 reactions was developed by the method of “core mechanism​ sub mechanism”. This mechanism can be conveniently applied to the simulation of practical combustors for its affordable scale. (3) Accuracies of the surrogate model and the mechanism were systematically validated by the various properties of the target fuel under pressures of 1-20atm, temperatures of 400-1250K, and equivalence ratios of 0.5-1.5. The overall errors for the ignition and combustion properties are no more than 0.4 and 0.1, respectively. (4) Laminar flame speeds of the target fuel and the surrogate model fuel were also measured for the validations. Results show that both the surrogate model and the mechanism can well predict the properties of the target fuel. The mechanism developed in this work is valuable to the further design and optimization of the propulsion systems.

Reacting Flow Characteristics and Multifuel Capabilities of a Multi-Nozzle Dry Low NOx Combustor: A Numerical Analysis

The fluctuating quality of natural gas (NG) in Peninsular Malaysia (PM) makes it challenging for the gas turbine (GT) combustor to meet the combustion performance requirements from the Original Equipment Manufacturer (OEM). Moreover, the gas quality sensitivity is more apparent in modern dry low NOx (DLN) combustors. Many of the prior combustion investigations were conducted on a modest scale in the laboratory. In actuality, combustion characterizations in complicated DLN combustors are more valuable to the power generation sector. Hence, the current numerical analysis utilized the RANS formulation and a detailed chemistry to examine the impact of ethane (C2H6), carbon dioxide (CO2), and nitrogen (N2) proportions in NG on combustion characteristics in a multi-nozzle DLN (MN-DLN) combustor, with the support of Modified Wobbe Index (MWI) calculations. Validations were performed using the combustor outlet temperature (COT) from the power plant where the actual MN-DLN combustor is operated, which revealed less than 10 % discrepancy. Qualitative validations were carried out by comparing the burn trace from the actual combustor wall to the predicted results, revealing an adequate Structural Similarity Index (SSIM) of 0.43. From numerical results of flame fronts and COTs, the addition of 20 % diluents (CO2 and N2) to NG demonstrated the blowoff risk. When MWIs of Kerteh and the JDA (major NG resources) were used as baselines, MWI ranges of all NG compositions under study surpassed the OEM’s ± 5 % limit. The increase in CO2 proportion results in a wide MWI range, especially when Kerteh is used as the baseline. Therefore, any GTs in PM that have previously been calibrated to use Kerteh's NG are more likely to have combustion instabilities if CO2 levels in NG suddenly increase. The higher MWI range backs up the current numerical results that showed the deleterious effects of a high CO2 composition throughout the combustor firing process. However, increasing the amount of C2H6 by up to 20 % is predicted to have minor effects on combustion characteristics. Overall, the validated numerical model of the MN-DLN combustor provided critical information about combustion characteristics and multifuel capabilities in respect to the NG quality in PM.