Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1729
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-17T18:13:47.301100Z
status: published
lens: g_space
sector: climate
headline: AI-driven whole-life costing for decision-ready UK public building retrofits
dek: Abstract Public-sector building retrofits in the United Kingdom require decisions that balance whole-life cost, carbon reduction, service continuity and auditability under limited data and constrained capital budgets. This paper develops a decision-ready whole-life costing (WLC) workflow that integrates environmental, social and governance (ESG) priorities with uncertainty-aware artificial intelligence (AI) for public-building retrofit appraisal. The workflow combines national non-domestic Energy Performance Cer…
gain_title: A reinforcement learning planner paired with an uncertainty-aware surrogate model produced auditable retrofit sequences, scored better on composite ESG performance than conventional baselines, and searched decisions far faster in matched test cases.
problem_title: (none)
trace_subject: (none)
gain_reading: A reinforcement learning planner paired with an uncertainty-aware surrogate model produced auditable retrofit sequences, scored better on composite ESG performance than conventional baselines, and searched decisions far faster in matched test cases.
gain_evidence: improves composite ESG performance relative to conventional WLC-only | supports auditable retrofit sequencing across archetype and budget conditions
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers built a decision-support workflow for retrofitting UK public buildings. It combines national energy certificate data, public-sector decarbonisation delivery evidence and stakeholder input, with a reinforcement learning planner whose objectives are weighted by stakeholder priorities. A Bayesian neural surrogate supplies energy, carbon and cost trajectories with uncertainty bounds.

The approach matters because public bodies must weigh whole-life cost, carbon cuts and service continuity on tight capital budgets and limited data. The results are promising, but they rest on public datasets and scenario-level evidence rather than live projects, so real-world performance remains untested.
limitation: Validation so far uses only public datasets and scenario-level evidence, not project-specific metered data or live deployment, so real-world performance is unproven.
tag: Evidence-backed gain
key_points: The workflow merges whole-life costing with environmental, social and governance priorities for early-stage retrofit appraisal. | Stakeholder interviews, a questionnaire and a co-design workshop were turned into transparent weights using fuzzy analytic hierarchy process. | The trained deep Q-network search was 30.03 times faster than scalar feasible-sequence enumeration across 1000 matched archetype-budget cases. | It outperformed whole-life-cost-only, cost-per-tonne-of-carbon and static multi-criteria decision analysis baselines on composite ESG performance.
rundown: The inputs include national non-domestic Energy Performance Certificate and Display Energy Certificate data and Public Sector Decarbonisation Scheme delivery evidence. These were used to build a scenario library of public-building archetypes, retrofit actions and appraisal constraints.

A Bayesian convolutional neural network and long short-term memory surrogate provides uncertainty-bounded energy, carbon and cost trajectories. This lets the planner evaluate policies repeatedly without expensive detailed simulation.

The authors frame the contribution as an integrated, explainable workflow for early-stage decision support. They propose future work using project-specific metered data, digital twins and live portfolio deployment.
sources:
- peer_reviewed | Complex & Intelligent Systems | https://doi.org/10.1007/s40747-026-02516-9 | 2026-10-16
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

c460bed524e4ac57ba345cd564d9fffee595c4f4238c4cb517c18cec522180d0

previous

0000000000000000000000000000000000000000000000000000000000000000