Experimental and computational analysis of an expansion deflection nozzle in open-wake mode

This paper presents a rapid solution method for flows within an Expansion Deflection nozzle operating within the atmosphere. The technique is based primarily on the Method of Characteristics, with integrated solution of the Rankine Hugoniot shock equations. Such a method is desirable in the case of Expansion Deflection nozzles, as the potential design space is very wide, and optimisation will require the performance evaluation across the full altitude range of operation, from sea level to vacuum. Rapid solution of the flow is therefore necessary in the design phase, ruling out the use of a complete CFD simulation. However, the model presented is based on an inviscid flow model, and hence it must be demonstrated that neglecting the effects of viscosity does not significantly reduce the accuracy of the results produced. This is achieved in this paper through comparison with experimental data produced by cold gas tests. It is shown that the location of the initial shock wave and expansion fan within the flow, and their interaction with the wall, are accurate. Wall pressures in this region are in reasonable agreement, although the effects of viscous boundary layers, which are not modelled, produce a local deviation from experiment under adverse pressure gradients. However, downstream of the first shock wave interaction, more significant differences in the pressure distribution are apparent. These are due to the impact on the flow of separation bubbles caused by the adverse pressure gradient induced by the pressure rise, not modelled in the simplified model. Despite this, thrust coefficients determined by integration of wall pressures differ from the experimental data by only 2.5 to 3.5%, due to cancelling of over and under prediction of wall pressures. A secondary outcome of the work demonstrates that the location of the initial shock wave in the flow is sensitive to the assumed wake pressure, and is most accurately approximated with an ambient wake pressure. This reinforces direct experimental results demonstrating that wake aspiration, generally assumed to occur in this class of nozzle, can be avoided by careful design. The effectiveness of the compensation is between 98 and 100% for the experimental model under the range of pressures considered, again demonstrating that wake aspiration is not occurring in this nozzle.