Multi-objective optimization for resource driven scheduling in construction projects

The main research developments of this study contribute to the advancement of current practice in resource scheduling and planning in construction projects and can lead to: (1) an increase in the resource utilization efficiency in construction projects which can produce significant improvements in construction productivity, cost and duration; (2) an improvement in utilizing the limited availability of resources; (3) a reduction in the duration and cost of multiple shifts operation while circumventing the negative impacts of shift work on productivity, safety, and cost; and (4) an enhancement in analyzing construction project risks in order to improve the reliability of project performance. First, innovative resource leveling metrics are developed to circumvent the limitation of existing metrics and directly measure and minimize undesirable resource fluctuation. A robust resource leveling model is formulated by incorporating the newly developed resource leveling metrics to maximize resource utilization efficiency for construction projects. The optimization model is implemented using genetic algorithms in order to optimize resource utilization efficiency. Second, a resource leveling and allocation model is developed to simultaneously optimize resource leveling and allocation for construction projects. The model is developed as a multi-objective genetic algorithm to provide optimal tradeoffs between maximizing resource utilization efficiency and minimizing project duration while complying with all resource availability constraints. Third, a robust multiple shifts scheduling model is formulated to simultaneously minimize project time and cost while minimizing the negative impacts of shift work on construction productivity, safety, and cost. A multi-objective genetic algorithm is utilized to implement the model in order to support construction planners in generating optimal tradeoffs among project time, cost, and labor utilization in evening and night shifts. The model is also designed to consider labor availability constraints in order to optimally distribute the limited availability of labor on the competing shifts. Fourth, a robust resource fluctuation cost model is developed to provide the most cost effective and efficient resource utilization for construction projects. The model is developed as a novel multi-objective optimization model that is capable of modeling and minimizing overall resource fluctuation costs (i.e. idle costs, release and rehiring costs, and mobilization costs) and analyzing and optimizing the potential tradeoffs between minimizing resource fluctuation costs and minimizing project duration. Fifth, a robust project risk assessment model is developed to overcome the limitations of existing probabilistic scheduling methods including (a) the inaccuracy limitation of the PERT method due to its “merge event bias” by incorporating an accurate multivariate normal integral method; and (b) the impractical computational time of the Monte Carlo simulation method by incorporating a newly developed approximation method. The model is named FARE (Fast and Accurate Risk Evaluation). The development of the FARE model facilitates the optimization of resource-driven scheduling while considering the impact of relevant risks and uncertainties. Sixth, a prototype multi-objective optimization for resource driven scheduling system is developed to seamlessly integrate the aforementioned research developments with commercially available project management software, Microsoft Project 2007, to facilitate their ultimate use and adoption by the construction industry. The system is designed to (1) retrieve project scheduling data from MS Project that can be utilized it in the developed optimization models, and store the generated optimization results in a binary file that can be accessed and processed by MS Project; (2) enable construction planners to benefit from and utilize the practical project scheduling and control features in MS Project during their analysis of the optimal schedules generated by the developed models in this study; and (3) facilitate the input of project parameters and the visualization of the obtained solutions using the newly developed graphical user interface modules.