Abstract:Accurate low dimension chemical kinetic models for methane are an essential component in the design of efficient gas turbine combustors. Kinetic models coupled to computational fluid dynamics (CFD) provide quick and efficient ways to test the effect of operating conditions, fuel composition and combustor design compared to physical experiments. However, detailed chemical kinetic models are too computationally expensive for use in CFD. We propose a novel data orientated three-step methodology to produce compact models that replicate a target set of detailed model properties to a high fidelity. In the first step, a reduced kinetic model is obtained by removing all non-essential species from the detailed model containing 118 species using path flux analysis (PFA). It is then numerically optimised to replicate the detailed model's prediction in two rounds; First, to selected species (OH,H,CO and CH4) profiles in perfectly stirred reactor (PSR) simulations and then re-optimised to the detailed model's prediction of the laminar flame speed. This is implemented by a purposely developed Machine Learned Optimisation of Chemical Kinetics (MLOCK) algorithm. The MLOCK algorithm systematically perturbs all three Arrhenius parameters for selected reactions and assesses the suitability of the new parameters through an objective error function which quantifies the error in the compact model's calculation of the optimisation target. This strategy is demonstrated through the production of a 19 species and a 15 species compact model for methane/air combustion. Both compact models are validated across a range of 0D and 1D calculations across both lean and rich conditions and shows good agreement to the parent detailed mechanism. The 15 species model is shown to outperform the current state-of-art models in both accuracy and range of conditions the model is valid over.

What problem does this paper attempt to address?

This paper aims to solve the problem of developing a high - efficiency low - dimensional chemical kinetic model in the methane combustion process. Specifically, the paper focuses on how to reduce the number of species in the chemical kinetic model by machine - learning methods under the premise of ensuring the accuracy of the model, thereby reducing the computational cost and enabling it to be more effectively applied in computational fluid dynamics (CFD) simulations. ### Background and Problems of the Paper - **Detailed Chemical Kinetic Model**: Although the detailed model can describe the combustion process with high fidelity, its computational cost is very high and it is not suitable for direct use in CFD simulations. - **Requirement for Compact Models**: In order to improve the efficiency of CFD simulations, it is necessary to develop a compact model with fewer species while maintaining an accurate description of the combustion process. - **Limitations of Existing Methods**: Existing mechanism optimization methods either rely on detailed reaction mechanisms to optimize parameters through uncertainty quantification and minimization; or reduce the number of species by simplifying the reaction network, but these methods often need to modify the numerical solver, which limits the generality of the model. ### Objectives of the Paper - **Develop New Compact Models**: The paper proposes a three - step data - driven method to generate compact models containing 19 and 15 species from a detailed 118 - species model through path - flux analysis (PFA) and a machine - learning optimization algorithm (MLOCK). - **Optimize Model Performance**: By optimizing the effective reaction rate constants, the compact models are made highly consistent with the detailed model in prediction results under a series of operating conditions, especially in terms of laminar flame speed and ignition delay time. ### Method Overview 1. **Path - Flux Analysis (PFA)**: Remove non - critical species from the detailed model to generate initial 19 - species and 15 - species models. 2. **Machine - Learning Optimization Algorithm (MLOCK)**: - **First Stage**: Optimize the performance of the compact model in a perfectly stirred reactor (PSR), mainly optimizing the molar fraction profiles of key species such as OH, H, CO and CH4. - **Second Stage**: Further optimize the performance of the model in terms of laminar flame speed, especially for optimization under fuel - rich conditions. ### Results and Discussion - **0D Simulation Verification**: The accuracy of the compact models in terms of ignition delay time and combustion temperature was verified through PSR simulations. The results show that the 19 - species model (19sOp) performs well under most conditions, while the 15 - species model (15sOp) has certain errors under some low - temperature fuel - rich conditions. - **1D Simulation Verification**: The performance of the compact models was verified through the calculation of laminar flame speed. The 19sOp model shows more than 95% accuracy under various conditions, and the 15sOp model also shows good consistency in terms of laminar flame speed. ### Conclusions The paper has successfully developed two efficient compact models, namely the 19 - species and 15 - species models. These models are not only significantly more computationally efficient than the detailed model, but also maintain high accuracy under a wide range of conditions, especially in the prediction of laminar flame speed and ignition delay time. The 15 - species model performs slightly worse under certain specific conditions, but is still better than existing similar models. Through this method, the paper provides a new idea for the development of high - efficiency low - dimensional chemical kinetic models and is expected to be widely used in future CFD simulations.

Toward Machine Learned Highly Reduce Kinetic Models For Methane/Air Combustion

Toward Development of Machine Learned Techniques for Production of Compact Kinetic Models

Comparative Analysis of Detailed and Reduced Kinetic Models for CH4 + H2 Combustion

Assessment of turbulence-chemistry interaction models in MILD combustion regime

Low-Dimensional High-Fidelity Kinetic Models for NOX Formation by a Compute Intensification Method

Modeling of high-speed, methane-air, turbulent combustion, Part II: Reduced methane oxidation chemistry

Comparison of Kinetic Mechanisms for Numerical Simulation of Methanol Combustion in DICI Heavy-Duty Engine

Evaluation of Chemical Kinetic Mechanisms for Methane Combustion: A Review from a CFD Perspective

Optimally-reduced kinetic models for GRI-Mech 3.0 combustion mechanism based on sensitivity analysis

Towards Predictive Combustion Kinetic Models: Progress in Model Analysis and Informative Experiments

An Optimized Methane Reaction Model for Oxyfuel Combustion

Skeletal and Reduced Kinetic Models for Methane Oxidation under Engine-Relevant Conditions

Optimization of Kinetic Mechanism for Methane Combustion Based on Machine Learning

Evaluation, Reduction, and Validation of a New Skeletal Mechanism for the Cofiring of NH3 and CH4

Model Comparisons of Flow and Chemical Kinetic Mechanisms for Methane–Air Combustion for Engineering Applications

Detailed and Reduced Chemical Kinetic Model of CH4/air Mixtures Combustion Based on High-Precision First-Principles Molecular Dynamics Simulation

Skeletal Reaction Models for Methane Combustion

Modelling Combustion of Ammonia/Hydrogen Fuel Blends under Gas Turbine Conditions

Construction of a Small-Sized Simplified Chemical Kinetics Model for the Simulation of n-Propylcyclohexane Combustion Properties

Study on Reduced Chemical Kinetic Model for DME HCCI Combustion

Chemical kinetic model reduction based on species‐targeted local sensitivity analysis