Abstract

In this paper, the fuel composition effects of H2/CO/CH4 syngas (0–100% for each component gas) on the self-excited high multi-mode (mainly the 3rd and 4th with their harmonics) combustion instability (CI) characteristics have been studied in a partially premixed swirl-stabilized gas turbine model combustor by investigating phase-resolved high-speed OH∗ planar laser induced fluorescence (OH-PLIF) and OH∗ chemiluminescence at a rate of 12.5kHz and analyzing the proper orthogonal decomposition (POD), spatio-temporal Rayleigh index (RI), and time-lag. Phase-synchronized OH-PLIF images suggested critical clues of CI driving mechanisms, including the periodic alternation of flame attachment/detachment and vortex coupling with everlasting flames at the outer recirculation zone due to high hydrogen fuels’ high reactivity. These PLIF results also demonstrated that the relatively short mixing length and highly fuel-dependent flame length are the root causes of a high instability-mode and sensible mode-shift. Thus, PLIF images were used to calculate the flame length by obtaining the intensity-weighted centroid for the more precise time-lag analysis. The spatio-temporal RI results indicated that the fuel composition affects the location and intensity of CI driving/damping, RI frequency, and instability mode and frequency. POD analysis from high-speed OH∗ images showed that the distinct coherent structures and large roll-up of flames are responsible for generating flame oscillations for each mode. High cross-correlation between the POD modes showed the convection of these coherent structures and axially-alternative swirl-like flame motions. At some particular compositions of H2/CH4/CO, high multi-mode CI was observed (e.g., the 3rd, 4th, and 6th modes appear simultaneously) and the original time-lag model could not be applicable, since the acoustic pressure wave is not simply sinusoidal but largely distorted. Thus, a new time-lag model using skewness time (τskew) was proposed to reflect the distortion from multi-mode CI and the accuracy of this model was verified using the experimental data.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call