Multiobjective construction schedule optimization with labor resource allocation: Coupling parameter-adaptive genetic algorithm and network flow

Although labor resource allocation is critical in the labor-intensive construction industry, it is typically performed only after the schedule has been determined. This "schedule leads, labor follows"paradigm not only sidelines labor's role in shaping the schedule, but also overlooks how the schedule, in turn, reshapes labor deployment. To bridge this gap, this study proposes a scheduling formulation that allows for the parallel and coordinated optimization of both construction scheduling and labor resource allocation. It captures the continuous and time-varying nonlinear relationship between labor usage and activity duration, and simultaneously considers multiobjectives, including time, cost, labor efficiency, and labor resource fluctuation. To solve this formulation, we develop an innovative integrated model - the parameter-adaptive Non-dominated Sorting Genetic Algorithm III combined with Network Flow (AN3-NF) - which embeds labor allocation directly into the scheduling process. The model uses adaptive parameter tuning to eliminate reliance on fixed values, and the activity node state updating strategy enables dynamic labor allocation. Case studies on two projects with 75 and 1,168 activities demonstrate the performance advantages and scalability of AN3-NF over existing algorithms. Results show significant improvements in solution quality and convergence speed, with reductions of 16.2% in construction time, 3.6% in cost, a 38.3% improvement in labor efficiency, and an 86.1% decrease in resource fluctuations. Discussions on parameter effects, stochastic risk handling, and disruption-based rescheduling further suggest the model's stability and robustness. Overall, the proposed method provides managers with an efficient and precise tool for labor and schedule planning, helping reduce project delays and cost overruns.