Multi-Objective Workforce Allocation for Off-Site Construction Assembly Using Parallel Simulated Annealing

Abstract

This paper proposes a scalable optimisation framework for workforce allocation in off-site con-struction assembly. The method supports daily staffing decisions in a job shop environment where steel-frame products are assigned to parallel stations. Economic performance and delivery reliability are captured by labour cost, total tardiness, and delay severity. The framework combines a production model, parallel Pareto Simulated Annealing, and context-aware neighbourhood moves. These searches adjust the staffing plan by redistributing workers, opening additional stations, or closing underutilised capacity based on tar-diness pressure, slack and utilisation. The method is validated on an off-site construction case study with 1769 structural steel frames, heterogeneous assembly workloads, and seven delivery deadlines. Results show that redistribution moves are ineffective in reducing delays against planned due dates. Configurations with station opening/closing eliminate tardiness and delay severity at low labour cost, converging in about 2 hours.