Rocket nozzle flow control using proper orthogonal decomposition

This paper investigates the use of proper orthogonal decomposition-based reduced order computational fluid dynamics models for model-based controller design. An open-loop optimal control problem is posed and solved concerning the regulation of a rocket engine thrust profile. The control synthesis uses a numerical flow field solver as the plant model. Controllers are synthesized using a reduced order model of the flow field generated via proper orthogonal decomposition. A quasi one-dimensional supersonic convergent-divergent nozzle with varying back pressure is used as a model problem. Treating the nozzle flow as one-dimensional, the reduced order model was used to synthesize a controller that controls thrust along an ascent trajectory by varying nozzle throat area and fuel mass flow rate. The effects of reduced order model accuracy on controller performance are quantified.