Simulating Scramjet Behavior: Unstart Prediction in a Supersonic, Turbulent Inlet-Isolator Duct Flow
thesisposted on 11.06.2019, 16:10 by Ian Avalon Hall
In the pursuit of developing hypersonic cruise vehicles, unstart is a major roadblock to achieving stable flight. Unstart occurs when a sudden instability in the combustor of a vehicle’s propulsion system creates an instantaneous pressure rise that initiates a shock. This shock travels upstream out of the inlet of the vehicle, until it is ejected from the inlet and creates a standing shockwave that chokes the flow entering the vehicle, thereby greatly reducing its propulsive capability. In severe cases, this can lead to the loss of the vehicle. This thesis presents the results of a computational study of the dynamics of unstart near Mach 5 and presents some possible precursor signals that may indicate its presence in flight. Using SU2, an open-source CFD code developed at Stanford University, the Unsteady Reynolds-Averaged Navier-Stokes equations are used to develop a model for flow in a scramjet inlet-isolator geometry, both in the fully started state and during unstart. The results of these calculations were compared against experimental data collected by J. Wagner, at the University of Texas, Austin. In the present computations, unstart was initiated through the use of an artificial body force, which mimicked a moveable flap used in the experiments. Once the results of the code were validated against these experiments, a selection of parametric studies were conducted to determine how the design of the inlet-isolator by Wagner affected the flow, and thus how generalizable the results can be. In addition, precursor signals indicative of unstart were identified for further study and examined in the different parametric studies. It was found that a thick boundary layer is conducive to a stronger precursor signal and a slower unstart. In addition, an aspect ratio closer to 1:1 promotes flow mixing and reduces the unstart speed and strength. Moreover, an aspect ratio in this range reduces the precursor signal strength but, if a thick boundary layer is present, will smear the signal out over a larger area, potentially making it easier to detect.