Computation of hypersonic unsteady viscous flow over a cylinder

Design of the new hypersonic vehicles, such as NASP, will rely heavily on computational fluid dynamics because of the lack of the hypersonic ground test facilities and the difficulty of the reproduction of the high altitude, high speed flight conditions. However, CFD cannot stand alone without hypersonic ground test facilities. Experimental data from the hypersonic ground test facilities are needed to verify the solutions of the CFD codes. A typical characteristic of the hypersonic impulsive test facilities is the extremely short running time, therefore, the flow establishment time around a model is very important. There is always a question as to whether or not the test times are sufficient to allow the establishment of certain types of steady flows over aerodynamic models, especially flows involving strong viscous effects such as boundary layer growth and separated flow. If available test time is not long enough to obtain a quasi-steady state flow over the aerodynamic model, the experimental data from these facilities are not reliable for accurately representing the simulated flow fields. Therefore, the intentions of this study are to assess the time required to obtain the steady state and to study the physical nature of the transients during the unsteady approach to the steady state of the flow over an aerodynamic model in the impulsive hypersonic ground test facilities. Numerical simulation of the hypersonic viscous flow over a two dimensional circular cylinder in shock tunnel has been attempted by using MacCormack's explicit time dependent predictor-corrector finite-difference method. The problem consists of two parts which are a quasi-one-dimensional nozzle flow for a convergent-divergent nozzle section and a flow over a circular cylinder for an aerodynamic model. The time dependent solution of the former has been used as the inflow boundary condition for the latter problem, whose governing equations are two dimensional full Navier-Stokes equations. The flow is assumed as a calorically perfect gas and a laminar flow. For a nozzle flow, the start-up process of the shock tunnel and the time required to the steady state have been studied. For a flow over a cylinder, the transients at early time and the effects of the artificial damping terms and the outer boundary have been studied.