This open access book introduces and explains machine learning (ML) algorithms and techniques developed for statistical inferences on a complex process or system and their applications to simulations of chemically reacting turbulent flows.
These two fields, ML and turbulent combustion, have large body of work and knowledge on their own, and this book brings them together and explain the complexities and challenges involved in applying ML techniques to simulate and study reacting flows. This is important as to the world’s total primary energy supply (TPES), since more than 90% of this supply is through combustion technologies and the non-negligible effects of combustion on environment. Although alternative technologies based on renewable energies are coming up, their shares for the TPES is are less than 5% currently and one needs a complete paradigm shift to replace combustion sources. Whether this is practical or not is entirely a different question, and an answer to this question depends on the respondent. However, a pragmatic analysis suggests that the combustion share to TPES is likely to be more than 70% even by 2070. Hence, it will be prudent to take advantage of ML techniques to improve combustion sciences and technologies so that efficient and “greener” combustion systems that are friendlier to the environment can be designed.
The book covers the current state of the art in these two topics and outlines the challenges involved, merits and drawbacks of using ML for turbulent combustion simulations including avenues which can be explored to overcome the challenges. The required mathematical equations and backgrounds are discussed with ample references for readers to find further detail if they wish. This book is unique since there is not any book with similar coverage of topics, ranging from big data analysis and machine learning algorithm to their applications for combustion science and system design for energy generation.
Author(s): Nedunchezhian Swaminathan, Alessandro Parente
Series: Lecture Notes in Energy, 44
Publisher: Springer
Year: 2023
Language: English
Pages: 352
City: Cham
Preface
Contents
Contributors
Introduction
1 Combustion Technology Role
2 Governing Equations
3 Equations for LES
3.1 SGS Closures
3.2 LES Challenges and Role of MLA
4 Objectives
References
Machine Learning Techniques in Reactive Atomistic Simulations
1 Introduction and Overview
1.1 Molecular Dynamics, Reactive Force Fields and the Concept of Bond Order
1.2 Accuracy, Complexity, and Transferability
2 Machine Learning and Optimization Techniques
2.1 Continuous Optimization for Convex and Non-convex Optimization
2.2 Discrete Optimization
3 Machine Learning Models
3.1 Unsupervised Learning
3.2 Supervised Learning
3.3 Software Infrastructure for Machine Learning Applications
4 ML Applications in Reactive Atomistic Simulations
4.1 ML Techniques for Training Reactive Atomistic Models
4.2 Accelerating Reactive Simulations
5 Analyzing Results from Atomistic Simulations
5.1 Representation Techniques
5.2 Dimensionality Reduction and Clustering
5.3 Dynamical Models and Analysis
5.4 Reaction Rates and Chemical Properties
6 Concluding Remarks
References
A Novel In Situ Machine Learning Framework for Intelligent Data Capture and Event Detection
1 Introduction
1.1 Overview of Related Work
1.2 Contributions and Organization
2 Approach
3 Results
3.1 Data Capture for Optimal I/O: Mantaflow Experiments
3.2 Detecting Physical Phenomena: Marine Ice Sheet Instability (MISI)
3.3 Reduced Order Modeling: Sample Mesh Generation for Hyper-Reduction
3.4 HPC Experiments
4 Conclusion
References
Machine-Learning for Stress Tensor Modelling in Large Eddy Simulation
1 Introduction
2 Classic Stress Tensor Models
2.1 Smagorinsky
2.2 Scale Similarity
2.3 Gradient Model
2.4 Clark Model
2.5 Wall-Adapting Local Eddy-Viscosity (WALE)
3 Deconvolution-Based Modelling
4 Machine-Learning Based Models
4.1 Type (a)
4.2 Type (b)
4.3 Type (c)
5 A Note: Sub-grid Versus Sub-filter
6 Challenges of Data-Based Models
6.1 Universality
6.2 Choice and Pre-processing of Data
6.3 Training, Validation, Testing
6.4 Network Structure
6.5 LES Mesh Size
6.6 Performance Metrics
7 Summary
References
Machine Learning for Combustion Chemistry
1 Introduction and Motivation
2 Learning Reaction Rates
2.1 Chemistry Regression via ANNs
3 Learning Reaction Mechanisms
3.1 Learning Observables in Complex Reaction Mechanisms
3.2 Chemical Reaction Neural Networks
3.3 PCA-Based Chemistry Reduction and Other PCA Applications
3.4 Hybrid Chemistry Models and Implementation of ML Tools
3.5 Extending Functional Groups for Kinetics Modeling
3.6 Fuel Properties' Prediction Using ML
3.7 Transfer Learning for Reaction Chemistry
4 Chemistry Integration and Acceleration
5 Conclusions
References
Deep Convolutional Neural Networks for Subgrid-Scale Flame Wrinkling Modeling
1 Introduction
2 Wrinkling Models
3 Convolutional Neural Networks
3.1 Artificial Neural Networks
3.2 Convolutional Layers
3.3 From Segmentation to Predicting Physical Fields with CNNs
4 Training CNNs to Model Flame Wrinkling
4.1 Data Preparation
4.2 Building and Analyzing the U-Net
4.3 A Priori Validation
5 Discussion
6 Conclusion
References
Machine Learning Strategy for Subgrid Modeling of Turbulent Combustion Using Linear Eddy Mixing Based Tabulation
1 Introduction
2 ML for Modeling of Turbulent Combustion
2.1 ANN Model for Chemistry
2.2 LES of Turbulent Combustion Using ANN
3 Mathematical Formulation with ANN
3.1 Governing Equations and Subgrid Models
3.2 ANN Based Modeling
4 Example Applications
4.1 Premixed Flame Turbulence
4.2 Non-premixed Temporally Evolving Jet Flame
4.3 SPRF Combustor
4.4 Cavity Strut Flame-Holder for Supersonic Combustion
5 Limitations of Past Studies
6 Summary and Outlook
References
On the Use of Machine Learning for Subgrid Scale Filtered Density Function Modelling in Large Eddy Simulations of Combustion Systems
1 Introduction
2 FDF Modelling
3 DNS Data Extraction and Manipulation
3.1 Low-Swirl Premixed Flame
3.2 MILD Combustion
3.3 Spray Combustion
4 Deep Neural Networks for Subgrid-Scale FDFs
4.1 Low-Swirl Premixed Flame
4.2 MILD Combustion
4.3 Spray Flame
5 Main Results
5.1 FDF Predictions and Generalisation
5.2 Reaction Rate Predictions
6 Conclusions and Prospects
References
Reduced-Order Modeling of Reacting Flows Using Data-Driven Approaches
1 Introduction
2 Governing Equations for Multicomponent Mixtures
3 Obtaining Data Matrices for Data-Driven Approaches
4 Reduced-Order Modeling
4.1 Data Preprocessing
4.2 Reducing the Number of Governing Equations
4.3 Low-Dimensional Manifold Topology
4.4 Nonlinear Regression
5 Applications of the Principal Component Transport in Combustion Simulations
5.1 A Priori Validations in a Zero-Dimensional Reactor
5.2 A Posteriori Validations on Sandia Flame D and F
6 Conclusions
References
AI Super-Resolution: Application to Turbulence and Combustion
1 Introduction
2 PIESRGAN
2.1 Architecture
2.2 Algorithm
2.3 Implementation Details
3 Application to Turbulence
3.1 Case Description
3.2 A Priori Results
3.3 A Posteriori Results
3.4 Discussion
4 Application to Reactive Sprays
4.1 Case Description
4.2 Results
4.3 Discussion
5 Application to Premixed Combustion
5.1 Case Description
5.2 A Priori Results
5.3 A Posteriori Results
5.4 Discussion
6 Application to Non-premixed Combustion
6.1 Case Description
6.2 A Priori Results
6.3 A Posteriori Results
6.4 Discussion
7 Conclusions
References
Machine Learning for Thermoacoustics
1 Introduction
1.1 The Physical Mechanism Driving Thermoacoustic Instability
1.2 The Extreme Sensitivity of Thermoacoustic Systems
1.3 The Opportunity for Data-Driven Methods in Thermoacoustics
2 Physics-Based Bayesian Inference Applied to a Complete System
2.1 Laplace's Method
2.2 Accelerating Laplace's Method with Adjoint Methods
2.3 Applying Laplace's Method to a Complete Thermoacoustic System
3 Physics-Based Statistical Inference Applied to a Flame
3.1 Assimilating Experimental Data with an Ensemble Kalman Filter
3.2 Assimilating with a Bayesian Neural Network Ensemble
4 Identifying Precursors to Thermoacoustic Instability with BayNNEs
4.1 Laboratory Combustor
4.2 Intermediate Pressure Industrial Fuel Spray Nozzle
4.3 Full Scale Aeroplane Engine
5 Conclusion
References
Summary
Index
