Answers are grounded in this paper and cite their sources.
Develops a data-driven surrogate model to optimize thermal performance in building envelopes, implying efficiency and cost savings.
Thermally activated building envelopes (TABEs) combine hydronic heat exchange with envelope thermal storage for low-grade energy utilisation. Their performance-oriented design requires multi-objective optimization over coupled geometric, operational, and climatic variables. Existing approaches either use one-dimensional equivalent models that cannot represent pipe-induced surface temperature non-uniformity, or rely on detailed solvers that are too costly for large-scale iterative search. This study develops a data-driven surrogate model (DDSM) framework grounded in a validated detailed physical model (DPM) to preserve pipe-level thermal behavior and support efficient multi-objective optimization. The DPM generates 2,000 climate-resolved samples. Three independent XGBoost regressors predict annual cumulative thermal load (ACTL), annual discomfort degree-hours (ADDH), and average surface temperature non-uniformity (ASTN), while life cycle cost (LCC) is computed analytically. SHAP analysis quantifies feature contributions and climate-dependent parameter sensitivities. The DDSM is embedded in NSGA-II to generate four-objective Pareto fronts, and entropy-weighted TOPSIS selects compromise solutions. The DDSM achieves R 2 values of 0.919–0.991 on the independent test set and 0.883–0.989 in an unseen-climate consistency check. Compared with direct DPM-based optimization, it reduces optimization time from 196 h to 2.3 s while reproducing DPM Pareto fronts within 4% error. The comparison also indicates that the conventional one-dimensional equivalent model can select misleading optima, with its selected optimum yielding an ADDH of 9,037.69 °C·h after DPM re-evaluation. In the four-climate case study, selected designs reduce LCC by 16%–22% and maintain ASTN below 0.05 °C. The proposed framework provides a DPM-trained and computationally tractable pathway for climate-specific TABE optimization while retaining pipe-level thermal information.
No takes yet. Share an insight, caveat, or question.
Chang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: