Agent-based simulation for multi-resource-constrained scheduling of scattered atypical repetitive projects
Résumé
Conventional planning tools struggle to manage the logistical complexity of scattered, short-duration atypical repetitive projects. Although simulation-based planning is well documented, its practical adoption in construction remains limited. This study proposes a simulation-optimization framework implemented in AnyLogic that integrates GIS-enabled agent-based modeling, stochastic simulation, and a GA-based optimization experiment to enable exploration of multi-resource-constrained scheduling decisions. Conceptually informed by multi-mode resource-constrained time-cost trade-off and multiple traveling-salesman formulations, the framework explores makespan-oriented resource-allocation and sequencing configurations under capacity, readiness, and spatial feasibility constraints. Key components include: (i) an adaptive scheduling algorithm for dispersed sites, (ii) a distance-aware rule-based assignment heuristic that supports continuity while limiting avoidable travel, (iii) stochastic disruption modeling, and (iv) spatially explicit visualizations to aid managerial interpretation. Applied to a real-world telecom megaproject, the framework consistently generated substantially shorter schedules across stochastic replications, achieving 46.25% reduction in project duration relative to the practitioner's baseline heuristic plan. Travel time/share, crew idle time, and utilization were recorded as emergent performance metrics and were used to evaluate operational behavior alongside makespan. Limitations include the absence of real-time traffic data, explicit cost modeling, and distance-penalty functions. Future work will incorporate economic and distance-based constraints and broaden comparative evaluation against alternative scheduling approaches.
Citer ce document
Accès au document
Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter
Voir l'article sur le site de la revueStatistiques
Consultations : 1
Téléchargements : 0