A simulation based optimization approach to model and design life support systems for manned space missions.

摘要

This dissertation considers the problem of process synthesis and design of life-support systems for manned space missions. A life-support system is a set of technologies to support human life for short and long-term spaceflights, via providing the basic life-support elements, such as oxygen, potable water, and food. The design of the system needs to meet the crewmember demand for the basic life-support elements (products of the system) and it must process the loads generated by the crewmembers. The system is subject to a myriad of uncertainties because most of the technologies involved are still under development. The result is high levels of uncertainties in the estimates of the model parameters, such as recovery rates or process efficiencies. Moreover, due to the high recycle rates within the system, the uncertainties are amplified and propagated within the system, resulting in a complex problem. In this dissertation, two algorithms have been successfully developed to help making design decisions for life-support systems. The algorithms utilize a simulation-based optimization approach that combines a stochastic discrete-event simulation and a deterministic mathematical programming approach to generate multiple, unique realizations of the controlled evolution of the system. The timelines are analyzed using time series data mining techniques and statistical tools to determine the necessary technologies, their deployment schedules and capacities, and the necessary basic life-support element amounts to support crew life and activities for the mission duration.