Decision-robust planning of positive energy districts under future climate–demand uncertainty

Positive energy district planning requires high-dimensional multi-vector sizing while accounting for deep uncertainty in climate-demand trajectories that can undermine the persistence of present-day optimal designs. This study proposes an integrated decision-robust planning framework for residential positive energy hubs under future climate-demand uncertainty. Unlike conventional deterministic or stochastic sizing studies, the present approach extends the analysis to post-optimal stress testing and regret-based decision screening under horizon-specific future scenarios. The workflow introduces an end-to-end, surrogate-assisted decision-support planning approach that sequentially combines urban building energy modeling, dynamic multi-energy simulation, multi-objective optimization, and ex-post robustness assessment. A Sobol low-discrepancy design of experiments is used to generate a space-filling training set from coupled simulations, enabling benchmarking of various surrogate regressors for mapping design variables to techno-enviro-economic objectives. Kernel-based surrogates provide the highest predictive fidelity, motivating selection of the radial basis function for surrogate-assisted multi-objective particle swarm optimization and knee-point extraction. Decision robustness is then quantified via scenario stress-testing of knee Pareto-optimal designs under horizon-specific climate-demand futures. While positive energy district feasibility is preserved, the feasibility buffer erodes substantially, with mean positive energy safety-margin loss increasing from 15 % (2030) to 58 % (2090) and an export-driven contraction of normalized net surplus. A TOPSIS-based robustness index and a regret analysis reveal late-century degradation and rising decision sensitivity: mean regret remains moderate (0.20–0.27 through 2070) but increases to 0.32 by 2090, with worst-case regret approaching 0.6. By combining feasibility-margin erosion and regret analysis, the framework provides a decision-support basis for risk-aware positive energy district design selection and long-term investment planning under climate-demand uncertainty.