Speaker
Description
As more and more man made space objects orbit around the Earth and crucial orbits are crowded with end-of-life objects. The use of deorbiting maneuvers is a solution to free crucial orbits from near end-of-life space objects by sending back on Earth while controlling the location of the reentry in order to minimize the risk for human populations. Yet, even during a controlled reentry, the reentry is subject to uncertainties, coming from the initial conditions, the material response to atmosphere forces and heatflux, the atmosphere conditions etc. In order toproduce a robust measure of the Gound human risk, the model uncertainties mus be included. In this work, we propose to use a probabilistic predictor to increase the robustness of the risk predictions while maintaining the computational cost at a reasonable level. The approach includes uncertainties coming from unknown model parameters or inputs and the uncertainties coming from our lack of understanding about the breakup. The breakup event is modeled as a random event sampled from an underlying distribution. Our predictor naturally provides the ground risk distribution from which can be derived statistics of interest. In order to keep the computational cost at minimum, the generation of the ground risk distribution is accelerated with a surrogate model.
