Precursor Detection of Combustion Instability

Combustor blowout instability develops under lean operating conditions displaying complex, low frequency thermo-acoustic fluctuations. In thermal pulse combustors, these oscillations can become chaotic. Detecting these changes in combustion dynamics prior to blowout has the potential to yield useful monitoring information and feedback for active control of instability. As the fuel/air ratio at which blowout occurs is uncertain, understanding how these precursors to combustion instability change with fuel composition, ambient conditions, pulsation amplitude, and combustor age is of fundamental performance. The traditional approach taken by combustor designers is to build sufficient margin into the design to prevent blowout under worst case operating conditions. Rather than suppress or avoid possible nonlinear behavior, we will investigate the possibility of exploiting the nonlinear nature of the combustion around blowout to reduce operating margins while maintaining stability or extend the limit of lean operability. Recent advances in nonlinear dynamics have led to the development of analysis techniques that are useful for gaining insight into the behavior of complex physical processes. These techniques will be used to analyze acoustic, chemiluminescence, and temperature oscillations measured in a thermal pulse combustor and identify precursors of combustion instability. Precursors would be identified by searching for qualitative changes in statistical behavior of combustion dynamics reconstructed from experimental data using time delay embedding. Nonlinear signal processing technique suitable for nonlinear and nonstationary processes will also be used to isolate intermittent and transient precursor signals in noise.