Simulation of turbulent supersonic separated base flows using enhanced turbulence modeling techniques with application to an X-33 aerospike rocket nozzle system

The successful application of CFD and turbulence modeling methods to an aerospike nozzle system first involves the successful simulation of its key flow components. This report addresses the task using the Chien low-Re k-epsilon and the Yakhot et al. high-Re RNG k-epsilon turbulence models. An improved implicit axis of symmetry boundary condition is also developed to increase stability and lower artificial dissipation. Grid adaptation through the SAGE post-processing package is used throughout the study.; The RNG model, after low-Re modifications, and the Chien low-Re k-epsilon model are applied to the supersonic axisymmetric base flow problem. Both models predict a peak recirculation velocity almost twice as large as experiment. The RNG model predicts a flatter base pressure and lower recirculation velocity more consistent with experimental data using less grid points than a comparable Chien model solution. The turbulent quantities predicted by both models are typical of other numerical results and generally under predict peak values obtained in experiment suggesting that too little turbulent eddy viscosity is produced.; After several test cases, the full 3-D aerospike nozzle is simulated using both the Chien and modified RNG low-Re models. The Chien model outperforms the RNG model in all circumstances. The surface pressure predicted by the Chien model along the nozzle center-plane is very near experiment while mid-plane results are not as close but useful for design purposes. The lack of a thick boundary layer along the nozzle surface in RNG simulations is the cause of poor surface pressure comparisons.; Although initial base flow comparisons between the model predictions and experiment are poor, the profiles are relatively flat. To accelerate the progress to a steady-state solution, a process involving the artificial lowering of the base pressure and subsequent iteration to a new steady state is undertaken. After several of these steps, the resulting steady-state base pressure is very near experimental values.; The effect of a slight geometry change on the flow characteristics is also examined through different thruster nozzle faceplate designs. The result of the slight modification is a tremendous reduction in surface pressure and temperature caused by recirculation at the thruster nozzle exit without adverse nozzle performance losses.