Extremum-seeking adaptive controller for swirl-stabilized spray combustion

Abstract

An extremum-seeking controller was demonstrated in a swirl-stabilized spray combustor. Pressure signals measured by a pressure transducer at the dump plane were processed and utilized to operate a high-frequency valve that modulated the fuel stream. The effects of phase delay control were investigated in two modes: a manual off-line determination of the optimum phase delay and an adaptive peak-seeking algorithm. The adaptive scheme was able to achieve similar suppression levels as the manually set optimal phase: 43% and 79% reduction in pressure and heat flux oscillations, respectively. The extremum-seeking algorithm achieved control at the optimal phase within 1.75 s. Stability margins of the extremum-seeking algorithm were determined by varying operational parameters including the perturbation signal frequency and amplitude and the integrator gain. Perturbation frequencies were required to be significantly different from the instability frequency to allow distinction between the perturbation and instability timescales. The amplitude of the perturbation signal had to be an order of magnitude smaller than the pressure fluctuations level to achieve convergence and prevent continuous destabilization of the combustion. Outside the required parameters range, the controller lost its authority and caused destabilization of the system. NOx and CO emissions were recorded at the exhaust of the combustor during controlled operation. The NOx was reduced by 16%, while the CO increased by 33% due to a decrease in temperature.