ROBUST RESOURCE INVESTMENT PROBLEM WITH TIME-DEPENDENT RESOURCE COST AND TARDINESS PENALTY

Abstract

The Resource Investment Problem (RIP) is a variant of the well-known Resource Constraint Project Scheduling Problem (RCPSP) that requires finding the optimal resource allocation, given a preset completion date, with the objective of minimizing the total cost. The practical relevance of RIP is very obvious; since the decision maker (the project manager for example) wants to know what resources are required to achieve the targeted project completion date. RIP helps to decide the amount of investment in resources that yield the optimal solution, in addition to the optimal tradeoff between completion time and resource investment. In practice, most of the projects are associated with due dates beyond which a tardiness penalty may be applied. To avoid the tardiness penalty, project managers sometimes decide to add more resources, thereby increasing resource investment cost, to the project to finish earlier. In this thesis the (RIP) has been extended to consider time-depended resource cost instead of time-independent resource cost in the classical RIP. The problem was named Resource Investment Problem with Time-Dependent Resource Cost and Tardiness Penalty, abbreviated as (RIP-TDRC). A mathematical model was introduced to simultaneously find the optimal resource assignment and activity staring times. The objective is to minimize the sum of the resources and tardiness cost.

Two versions of this problem are addressed in this thesis: the deterministic version of RIP-TDRC and the stochastic version. For the latter, it is assumed that the activity durations are subject to many uncertainties such as (bad weather conditions, material shortage, employee’s absences …etc.). To solve this problem, a simulation-optimization based algorithm is proposed. This algorithm solves the deterministic problem version iteratively through all possible project completion times and simulates the project considering the uncertainties to find the optimal solution. The performance of the proposed algorithm and the effect of some problem parameters on the solution are assessed through computational experiments. The experiments revealed the usefulness of the algorithm in finding relatively robust solution for small problem sizes.