Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Combustion Condition Identification using a Decision Tree based Machine Learning Algorithm Applied to a Model Can Combustor with High Shear Swirl Injector

PK Archhith, SK Thirumalaikumaran, Balasundaram Mohan, Saptharshi Basu

Abstract

Combustion is the primary process in gas turbine engines, where there is a need for efficient air-fuel mixing to enhance performance. High-shear swirl injectors are commonly used to improve fuel atomization and mixing, which are key factors in determining combustion efficiency and emissions. However, under certain conditions, combustors can experience thermoacoustic instability. In this study, a decision tree-based machine learning algorithm is used to classify combustion conditions by analyzing acoustic pressure and high-speed flame imaging from a counter-rotating high-shear swirl injector of a single can combustor fueled by methane. With a constant Reynolds number and varying equivalence ratios, the combustor exhibits both stable and unstable states. Characteristic features are extracted from the data using time series analysis, providing insight into combustion dynamics. The trained supervised machine learning model accurately classifies stable and unstable operations, demonstrating effective prediction of combustion conditions within the studied parameter range.

Combustion Condition Identification using a Decision Tree based Machine Learning Algorithm Applied to a Model Can Combustor with High Shear Swirl Injector

Abstract

Combustion is the primary process in gas turbine engines, where there is a need for efficient air-fuel mixing to enhance performance. High-shear swirl injectors are commonly used to improve fuel atomization and mixing, which are key factors in determining combustion efficiency and emissions. However, under certain conditions, combustors can experience thermoacoustic instability. In this study, a decision tree-based machine learning algorithm is used to classify combustion conditions by analyzing acoustic pressure and high-speed flame imaging from a counter-rotating high-shear swirl injector of a single can combustor fueled by methane. With a constant Reynolds number and varying equivalence ratios, the combustor exhibits both stable and unstable states. Characteristic features are extracted from the data using time series analysis, providing insight into combustion dynamics. The trained supervised machine learning model accurately classifies stable and unstable operations, demonstrating effective prediction of combustion conditions within the studied parameter range.
Paper Structure (16 sections, 11 equations, 6 figures, 1 table)

This paper contains 16 sections, 11 equations, 6 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. EXPERIMENTAL SETUP, DATA ACQUISITION AND COMBUSTOR DYNAMICS
  3. METHODOLOGY
  4. Root means square (RMS) values
  5. Signal to noise ratio (SNR)
  6. Phase space Reconstruction
  7. Average Mutual Information (AMI)
  8. False Nearest Neighbors (FNN)
  9. Fractal Dimension (FD)
  10. Hurst Exponent
  11. Recurrence Plot
  12. MACHINE LEARNING (ML) MODEL
  13. Data processing and feature extraction
  14. ML model architecture and training
  15. Classifier testing, validation and accuracy
  16. ...and 1 more sections

Figures (6)

  • Figure 1: Schematic of the experimental setup with flame stabilized at the dump plane using counter-rotating high shear swirl injector. The air and fuel flow lines are shown for clarity along with flame images (insets).
  • Figure 2: (a) Acoustic pressure and associated (b) sound pressure level for stable (black solid line) and unstable (blue solid line) operations, respectively. $f_{dom}$ denotes the dominant peak associated with duct acoustic modes.
  • Figure 3: (a) Root mean square value of the acoustic pressure, (b) dominant instability frequency and associated (c) sound pressure level (SPL) across various $\phi$. The blue and red markers denote stable and unstable operating conditions.
  • Figure 4: Comparison of (a) PDF and (b) Fourier spectrum between full data (blue) and sub-data (red). Left and right columns correspond to stable and unstable conditions.
  • Figure 5: (Variation of (a) $LAM$, (b) $TT$, (c) fractal dimension, and (d) Hurst exponent with time, respectively.
  • ...and 1 more figures