The design of the entry system of a space vehicle requires accurate predictions of atmospheric conditions and vehicle aerothermodynamics. Since the variability of atmospheric conditions has important implications for EDL (Entry, Descent and Landing) mission design and performance, post-flight reconstruction is essential.
In-flight measurements must be fully exploited to gain knowledge on both the vehicle response to the planetary environment and the atmospheric flight conditions, which must be reconstructed accurately to validate or improve ground predictions. The problem of accurately rebuilding atmospheric conditions upstream the shock starting from surface measurements is an inverse problem that requires accurate but sometimes costly methods to be solved.
One technique is Bayesian inference, which is capable of exploiting the results of complex forward Computational Fluid Dynamics (CFD) simulations and Uncertainty Quantification (UQ) to reconstruct the freestream conditions. This approach allows to consider the complex high temperature effects of hypersonic flows, such as thermochemical non equilibrium, radiation, ablation, and catalysis, and to take into account measurement errors, uncertainties on chemical model parameters and also model uncertainties. In the context of this technique the inverse problem is solved with a statistical approach. Results are the posterior conditional probabilities of the quantities that need to be reconstructed given the measurements, and this allows to compute error bounds of the results.
Sketch of aerothermodynamic phenomena during atmospheric entry (left), and simulated temperature field around the nose of EXPERT vehicle at nominal conditions (right)
The object of the research is building a robust, automated and affordable Bayesian reconstruction framework able to rebuild freestream speed and atmospheric conditions starting from wall pressure and heat flux data given by flush mounted sensors. It relies on the coupling of an hypersonic CFD tool with the use of metamodels to accelerate the forward and backward propagation of uncertainties.
