Cj. Steffen et al., Computational analysis for rocket-based combined-cycle systems during rocket-only operation, J PROPUL P, 16(6), 2000, pp. 1030-1039
A series of Reynolds-averaged Navier-Stokes calculations were employed to s
tudy the performance of rocket-based combined-cycle systems operating in an
all-rocket mode. This parametric series of calculations were executed with
in a statistical framework, commonly known as design of experiments, The pa
rametric design space included four geometric and two flowfield variables s
et at three levels each, for a total of 729 possible combinations, A D-opti
mal design strategy was selected. It required that only 36 separate computa
tional fluid dynamics (CFD) solutions be performed to develop a full respon
se surface model, which quantified the linear, bilinear, and curvilinear ef
fects of the six experimental variables. The axisymmetric, Reynolds-average
d Navier-Stokes simulations were executed with the NPARC v3.0 code. The res
ponse used in the statistical analysis was created from I-sp efficiency dat
a integrated from the 36 CFD simulations. The influence of turbulence model
ing was analyzed by using both one- and two-equation models. Careful attent
ion was also given to quantify the influence of mesh dependence, iterative
convergence, and artificial viscosity upon the resulting statistical model.
Thirteen statistically significant effects were observed to have an influe
nce on rocket-based combined-cycle nozzle performance. It was apparent that
the free-expansion process, directly downstream of the rocket nozzle, can
influence the I-sp efficiency. Numerical schlieren images and particle trac
es have been used to further understand the physical phenomena behind sever
al of the statistically significant results.