Stochastic resource leveling optimization method for trading off float consumption and project completion probability

Existing resource leveling (RL) approaches fall short of analyzing the trade-off relation between the total float consumption and project completion probability in a real-life project RL problem. This article presents a stochastic resource leveling optimization (SOLO) method that minimizes the total float consumption along with maximizing the project completion probability. It initializes the earliest start times of noncritical activities, measures the level of resource fluctuations of each candidate solution, computes the probability of completing a project in a target deadline by executing simulation-based scheduling, and identifies optimal solution(s) (i.e., optimal start times of noncritical activities) by implementing genetic algorithm, thereby identifying an optimal resource-leveled baseline. The study is of value to practitioners because SOLO considers both the amount of total float consumption and project completion probability. This study facilitates experimentation with different computation time-saving options given various constraints (i.e., the residual of project completion probabilities, threshold of release and rehire, ratio of criticality index, and number of critical activities). Test cases verify the validity of the computational method.