RecChain: a blockchain platform for renewable energy certificate management with oracle-gated issuance and calibrated anomaly screening

Abstract

Software artefact accompanying the manuscript "RecChain: a blockchain platform for renewable energy certificate management with oracle-gated issuance and calibrated anomaly screening", submitted to Springer Electrical Engineering. RecChain issues renewable energy certificates as ERC-1155 semi-fungible tokens, gated by a decentralised oracle network that screens metered generation readings before any certificate is minted. This archive is the exact version described in the manuscript (git tag v1.0.0-ee-submission). Two issuance paths. The archive contains two oracle-mediated issuance implementations, and the distinction matters when reproducing the results. The legacy path (functions/source.js, contracts/RecDapp.sol) is the Chainlink Functions implementation; every empirical result reported in the manuscript characterises this path, and the platform it ran on was withdrawn in June 2026. The revised path (cre/screening/main.ts, contracts/RecDappCRE.sol) is the migration to the Chainlink Runtime Environment — implemented, unit-tested and exercised under local simulation, but not deployed. No measurement reported in the manuscript derives from it. Contents. contracts/ registry, CRE report receiver, mocks · test/ Hardhat suites, 48 behavioural cases · cre/ workflow, mock meter endpoint, unit and integration suites · functions/ legacy Functions source · evaluation_v2/ the scripts producing every number in Section 6, with input series and committed outputs · server/, client/. Reproduction. npx hardhat test → 63 passing (48 behavioural plus 15 gas-measurement cases); node --test cre/screen.test.js → 9/9; node --test cre/integration.test.js → 8/8; cre workflow simulate screening → 3 issuances, then 2 under an injected anomaly, then none when the metering endpoint is unreachable. Section 6 is reproduced from evaluation_v2/; see the README there for run order. Selected findings. Screening was evaluated on a synthetic benchmark and on measured Greek generation data, with all detectors compared at a matched alarm budget rather than at a fixed σ multiple. The synthetic benchmark overstates F1 by 0.156 relative to measured data, so the synthetic figure should not be read as a deployment expectation. An EWMA control chart attains the best mean rank, but its advantage over the rolling z-score is not significant under Friedman Aligned-Ranks with Finner post hoc (p = 0.128); threshold calibration matters more than the choice of statistic. Gas cost is invariant to batch size. Data and licences. The software is released under MIT. evaluation_v2/data/gr_generation_hourly.csv is the Greek subset of the Open Power System Data time series package, version 2020-10-06 (doi:10.25832/time_series/2020-10-06), released under CC-BY-4.0 and republishing ENTSO-E Transparency Platform records. It covers 2018-01-01 to 2020-09-30 at hourly resolution and is redistributed here under the terms of that licence so the evaluation runs without a network fetch. Development continues at https://github.com/panagiotisvionis/REC_Dapp_Ethereum