Digital Twin-Driven Cost Estimation for Sustainable Construction Projects

Thesis Abstract

The rapid integration of digital twin (DT) technology into construction management promises improved accuracy and transparency in cost estimation for sustainable projects, yet practical deployment remains fragmented by data interoperability challenges and limited empirical validation. This study investigates how a DT-driven cost estimation framework can enhance early-stage budgeting, lifecycle cost forecasting, and value-for-money optimisation in sustainable construction under uncertainty. The aim is to design, implement, and evaluate a DT-enabled costing model that fuses BIM-based geometry, real-time sensor data, and probabilistic cost drivers to produce dynamic, scenario-aware estimates. Specific objectives are (1) to delineate a data-architecture and workflow architecture for linking BIM, IoT sensors, and cost databases; (2) to develop a DT-enabled cost model incorporating parametric cost relationships, activity-based costing, and Monte Carlo simulation to quantify uncertainty; (3) to validate the model against historical project datasets and live pilot projects; (4) to compare DT-driven estimates with conventional costing approaches in terms of accuracy, reliability, and adaptability to design changes; and (5) to formulate governance and ethical guidelines for data stewardship in DT-based cost management. The methodology adopts a mixed-methods design within a pragmatist paradigm. The population comprises completed and ongoing medium-to-large sustainable construction projects across three metropolitan regions. A stratified sample of 60 projects will be selected, with 40 used for model calibration and 20 reserved for out-of-sample validation. Data collection integrates archival project records (bids, invoices, change orders), BIM models, and IoT sensor streams from site logistics, environmental monitoring, and structural health indicators. Instrumentation includes a structured data extraction protocol, a DT pilot platform capturing live BIM-sensor cost interdependencies, and semi-structured interviews with cost engineers and project managers to capture tacit knowledge. Validity and reliability are ensured through triangulation, inter-rater reliability checks for data coding, and pilot testing of the data integration pipeline. The analytical framework combines quantitative and qualitative techniques (i) regression-based calibration of unit costs and activity durations, (ii) Monte Carlo simulation to propagate uncertainty across design and market scenarios, (iii) Bayesian updating to refine cost estimates as new data arrive, and (iv) thematic analysis of interview transcripts to surface practical barriers and enablers. A comparative analysis will be conducted between the DT-driven model and traditional estimating methods using metrics such as mean absolute percentage error (MAPE), root mean square error (RMSE), and 95% predictive intervals. The model specification includes a hybrid cost function that integrates parametric cost relationships, activity-based costing, and a DT-driven temporal forecasting component. The study is guided by the theory of information asymmetry and the systems theory of cyber-physical platforms, with reference to the Technology-Organisation-Environment (TOE) framework to interpret adoption determinants. It also draws on the Jensen-Lund model for decision-support in construction and the uncertainty theory underlying Monte Carlo methods. The expected findings indicate that the DT-driven framework reduces MAPE by 12–20 percent and narrows predictive intervals by 15–25 percent relative to conventional estimates, particularly in late design development when changes are frequent. It is anticipated that the DT approach will improve sensitivity to sustainability-specific cost drivers (embodied carbon, recycled content, deconstruction value) and enable scenario analysis for life-cycle cost optimization under carbon pricing and material price volatility. The study contributes to knowledge by operationalising a replicable DT-enabled costing architecture for sustainable construction, advancing empirical understanding of data interoperability challenges and the value of real-time information in cost management, and extending theory on information integration in construction cost estimation. Conclusions are expected to advocate for standardized data schemas, governance frameworks for DT data stewardship, and explicit training for cost engineers in probabilistic and dynamic estimation techniques. Recommendations will address scalable implementation in industry practice, including phased rollouts, stakeholder engagement strategies, and policy implications for government-funded sustainable construction programs.