Abstract

Feasibility studies can remain in public-estate investment pipelines after the tariffs, emission factors and cost levels used to produce them have changed. This study develops a controlled current-data rebasing audit that first reconciles a completed feasibility dataset, then freezes its physical energy quantities and retrofit configuration while updating only explicitly identified exogenous accounting variables. The audit is applied to four anonymised 2022 primary-school retrofit cases from one UK local-authority estate. Their baseline OpenStudio/EnergyPlus models were calibrated to 12 monthly utility observations and met the ASHRAE Guideline 14 monthly NMBE and CV(RMSE) criteria, although the retained evidence package does not recover the exact calibration dates or several model-input assumptions. Under the 2022 source-report basis, the combined packages reduce reported operational CO₂ by 56.7–67.8% (case mean 62.6%), leaving a 32.2–43.3 percentage-point residual within the stated location-based accounting boundary. Rebasing the same physical quantities to the 2026 UK Government conversion factors raises the case reductions to 72.5–78.8% (mean 76.3%) and narrows the residual to 21.2–27.5%. Whole-package utility-cost comparisons are tariff-contingent: the three gas-fired cases change from 13.3 to 16.4% cost increases under the 2022 tariffs to 5.1–13.1% savings under Q1 2026 average electricity and gas tariffs, while the oil-fired case saves 33.0% when Q1 2026 electricity is paired with the retained source-case oil price of £0.085/kWh. A small-consumer-band gas test remains mixed (−0.6% to +5.4%). A separate output-level heat-system counterfactual shows why these package results must not be labelled as ASHP-only effects: at η = 0.85 and SCOP = 3.0, the current-price incremental ASHP cost remains positive for the gas cases (£4.3–10.5 thousand/yr) while the oil case saves about £1.6 thousand/yr. Recasting sequencing as cumulative emissions gives 1283 tCO₂ avoided over eight years for the carbon-led sequence, compared with 1106 tCO₂ for fabric-first and 912 tCO₂ for cost-led sequencing. The contribution is therefore a reproducible, bounded method for distinguishing durable physical conclusions from tariff-, factor- and accounting-vintage effects; it is not a statistically representative assessment of the UK school stock.