A two-list genetic algorithm for optimizing work package schemes to minimize project costs

Minimizing project costs is critical in project planning with a work breakdown structure for optimal work package schemes. Furthermore, optimizing work package schemes (i.e., determining work package size and content) becomes more challenging when the project task duration is uncertain. This paper aims to develop a two-list genetic algorithm (TLGA) to optimize work package schemes for minimizing project costs considering deterministic and stochastic task durations. First, this paper defines the deterministic work package scheme problem and the stochastic work package scheme problem. Second, the TLGA, comprising a task list and a work packaging list, is developed. The TLGA can directly generate the work package scheme in the deterministic work package scheme problem and efficiently give work package policies through stochastic distribution simulations in the stochastic work package scheme problem. Moreover, a graphical user interface integrated with the TLGA is developed to enhance the practical application of the TLGA in real construction projects. Finally, the experiments are conducted based on a project case and a project dataset, showing that the TLGA can reduce the total cost by up to 19.57% in the deterministic problem compared to the original work package scheme, and the minimum gap between the TLGA and the state-of-the-art heuristics is only 3.91% under the ideal conditions. However, the TLGA with two parallel computing processes can reduce the running time by about 66% compared to the heuristics. In the stochastic problem, the TLGA provides work package policies under various stochastic distributions. This paper analyses the impact of stochastic distributions on work package policies, paving the way for advancing project planning based on work packages in uncertain environments.