Date of Award


Document Type


Degree Name

Master of Science


Department of Operational Sciences

First Advisor

Jeffrey P. Kharoufeh, PhD


This work considers the problem of finding optimal replacement policies that minimize the expected total cost of maintaining a satellite constellation. The problem is modeled using discrete-time Markov decision processes to determine the replacement policy by allowing the satellite constellation to be in one of a finite number of states at each decision epoch. The constellation stochastically transitions at each time step from one state to another as determined by a set of transition probabilities. At each decision epoch, a decision maker chooses an action from a set of allowable actions for the current system state. A cost associated with each possible action is determined by the number of satellites purchased, launched, or held in storage, as well as the operational capability of the constellation. The system is evaluated for a given time horizon using the standard Policy Evaluation Algorithm of Markov decision processes (stochastic dynamic programming ) to determine the optimal replacement policy and the minimum expected total cost. Example problems using notional data are presented to demonstrate the solution procedures. Sensitivity analysis of problem parameters is performed to investigate their impact on the minimum expected total cost of operating the constellation over a specified time horizon.

AFIT Designator


DTIC Accession Number