Multifractal analysis of combustion dynamics and blowout precursors in a turbulent combustor

This study investigates the nonlinear dynamical behavior of a combustor during the transition from stable combustion to lean blowout (LBO) by multifractal detrended fluctuation analysis (MFDFA). The transition to blowout was achieved by varying the Reynolds number (Re) from 1717 to 6439, causing the combustor to evolve through intermittency and thermoacoustic instability. Initially, the combustor dynamics were characterized using pressure time-series, wavelet spectrograms, power spectra, and phase-space representations, leading to the identification of distinct dynamical states. These conventional temporal and spectral analyses provide limited insight into the underlying nonlinear interactions; therefore, MFDFA was employed to examine the fractal properties of the pressure signals. For multifractal analysis, the dynamics were categorized into three regimes: aperiodic oscillations, intermittent bursting oscillations, and limit-cycle oscillations. Comparative multifractal spectra revealed significant variations in spectral width across these regimes. The spectrum width was maximum during stable combustion, decreased during intermittency, and collapsed near instability, then increased again beyond instability. Quantitative analysis using multifractal parameters, namely, the Hurst exponent (H), spectrum width (W), dominant singularity exponent (α0), and asymmetry parameter (β), demonstrated distinctive precursor signatures prior to LBO. This methodology enabled the early-warning indicator of flame blowout approximately 15.3% prior to its occurrence. Owing to its computational efficiency and predictive capability, the proposed framework shows strong potential for real-time industrial combustor monitoring and as a precursor indicator of LBO.