Data and Code for: Robustness of Near-Optimal Envelope Sets across Climates: Physical Realization, HVAC Weighting, and a 3840-Run EnergyPlus Study

This record contains the reproducibility data, analysis scripts, model-generation code, configuration files, manifests, tests, and publication-figure inputs for the manuscript “Robustness of Near-Optimal Envelope Sets across Climates: Physical Realization, HVAC Weighting, and a 3840-Run EnergyPlus Study.” It is an expanded successor to the previous Zenodo version and should be deposited using Zenodo’s “New version” function. The package contains five complete 768-point EnergyPlus experiment outputs (3840 annual simulation rows in total): the reference-WWR factorial design, low-WWR and high-WWR sensitivity profiles, a 60% WWR stress test, and a physical-assembly realization experiment. It also contains 768 matched reduced-order RC-model cases, candidate-recall and validation results, climate-conditioned factor-effect decompositions, near-optimal-set and tolerance-sensitivity tables, cross-WWR transfer/regret analyses, physical-assembly diagnostics, and an HVAC-weighted screening diagnostic. All retained EnergyPlus runs passed the recorded zero-warning and zero-severe-error quality gates. The package includes SHA-256 manifests for result files, configurations, weather-file references, and promoted outputs. The computational analysis environment was Python 3.11 with NumPy 2.2.6, Pandas 2.3.3, SciPy 1.15.3, and Matplotlib 3.10.3; EnergyPlus reruns require EnergyPlus 26.1.0. Licensed weather files, including the Helsinki IWEC file, are not redistributed. Their filenames and SHA-256 references are recorded in the manifests; users must obtain them from the authorized source before reproducing the weather-file reruns. The compact analysis and figure-generation workflows do not require redistribution of those licensed files. This record supports verification and reuse of the manuscript’s computational results. It does not contain raw EnergyPlus run directories, which were omitted after compact result extraction to keep the archive practical for repository download.