ABSTRACT
Linear programming is usually used to solve parallel MDPs subject to global resource constraints for each of the time periods (Golabi et al., 1982, Murakami and Turnquist 1985, Smilowitz and Madanat 2000, Madanat et al., 2006). This methodology however has a major limitation; namely, it is restricted to homogenous systems (with respect to costs, activities types and deterioration). i.e. the different components of the system have to be essentially the same.
