Digital Twins and Artificial Intelligence for Sustainable Infrastructure Management: A Lifecycle Review of Performance Evaluation and Emerging Technologies in the Built Environment

The convergence of digital twins and artificial intelligence is rapidly reshaping infrastructure management. To the authors’ knowledge, the empirical evidence across the full built-environment lifecycle, from design and construction through operation to end-of-life deconstruction, has not previously been mapped together with its sustainability outcomes in a single synthesis; existing reviews address parts of this space, and the present work extends them by combining lifecycle coverage, quantitative cross-domain characterization, and sustainability assessment within one structured analysis. This review synthesizes 261 Web of Science records together with 148 experiment records extracted from 40 primary ScienceDirect studies, spanning 2020–2026, analyzed through integrated bibliometric mapping and structured experimental evidence extraction. The synthesis indicates that, in the studies reviewed, hybrid physics-data architectures tended to outperform purely data-driven models, that energy forecasting reached coefficients of determination above 0.90 in several building case studies, that structural health monitoring detected damage undetected by conventional systems in instrumented field trials, and that predictive maintenance reduced costs by up to 58.1% in a single bridge-maintenance optimization study. These results derive from individual case studies, simulations, and short-term deployments rather than from multi-year field evidence, and they are reported with that distinction preserved. Dynamic lifecycle assessment frameworks demonstrated 25% energy reductions and 15% carbon emission decreases in operational settings, while AI-optimized zero-carbon buildings achieved emission intensities of 0.8 versus 3.5 kgCO₂/m² for conventional construction. The study identifies critical research gaps in data interoperability, generative AI validation for safety-critical decisions, and the absence of longitudinal field evidence, establishing a prioritized roadmap for advancing DT-AI integration toward measurable, lifecycle-spanning sustainability impact in the built environment.