Computer Science > Computational Engineering, Finance, and Science
arXiv:2511.15711 (cs)
Abstract:Persistent cost and schedule overruns in U.S. building projects expose limitations of conventional, document-based estimating and deterministic Critical Path Method (CPM) scheduling, which remain inflexible under uncertainty and lag dynamic field conditions. This study presents an integrated 4D/5D digital-twin framework unifying Building Information Modeling (BIM), natural language processing (NLP), reality capture, computer vision, Bayesian risk modeling, and deep reinforcement learning (DRL) for construction cost and schedule control. The system automates project-control functions by: (a) mapping contract documents to…
Computer Science > Computational Engineering, Finance, and Science
arXiv:2511.15711 (cs)
Abstract:Persistent cost and schedule overruns in U.S. building projects expose limitations of conventional, document-based estimating and deterministic Critical Path Method (CPM) scheduling, which remain inflexible under uncertainty and lag dynamic field conditions. This study presents an integrated 4D/5D digital-twin framework unifying Building Information Modeling (BIM), natural language processing (NLP), reality capture, computer vision, Bayesian risk modeling, and deep reinforcement learning (DRL) for construction cost and schedule control. The system automates project-control functions by: (a) mapping contract documents to standardized cost items using transformer-based NLP (0.883 weighted F1 score); (b) aligning photogrammetry and LiDAR data with BIM to compute earned value; (c) deriving real-time activity completion from site imagery (0.891 micro accuracy); (d) updating probabilistic CPM forecasts via Bayesian inference and Monte Carlo simulation; (e) using DRL for adaptive resource allocation (75% adoption rate); and (f) providing 4D/5D decision sandbox for predictive analysis. A Texas mid-rise case study demonstrates localized cost adjustment using RSMeans City Cost Index and Bureau of Labor Statistics wage data. Results show 43% reduction in estimating labor, 6% overtime reduction (91 hours), and project completion matching P50 probabilistic forecast of 128 days, confirming improved estimation accuracy and responsiveness.
| Subjects: | Computational Engineering, Finance, and Science (cs.CE) |
| Cite as: | arXiv:2511.15711 [cs.CE] |
| (or arXiv:2511.15711v1 [cs.CE] for this version) | |
| https://doi.org/10.48550/arXiv.2511.15711 arXiv-issued DOI via DataCite |
Submission history
From: Atena Khoshkonesh [view email] [v1] Tue, 4 Nov 2025 18:49:34 UTC (4,275 KB)