hyperGrid: A Provenance-First Visualisation Platform for Synthetic Irish Electricity Demand (SYNTH_GRID_IE)

hyperGrid is a provenance-first web platform for exploring SYNTH_GRID_IE, a weather-driven, hourly synthetic dataset of Irish building and data-centre electricity demand covering 2022–2023. The dataset (build "run8") comprises 437,529,982 hourly rows across 25,006 modelled buildings, a synthetic national reference series, synthetic wind and solar farm registers, a feeder-hour sample, and a synthetic nowcast ensemble spanning the 26 Republic of Ireland counties.

hyperGrid provides the read-only serving layer for this dataset: a dark-themed dashboard and a JSON read API over small, pre-aggregated rollup tables, so the 437-million-row time series is never queried per request. Its distinguishing feature is that the dataset's honesty and labelling rules are first-class user-interface elements — every figure carries its provenance class (REAL, CALIBRATED, SYNTHETIC, or COMPUTED), a scope qualifier (totals cover modelled building and data-centre sectors, approximately 88% of national demand), and the build context (global_seed 20240101), so synthetic or calibrated values can never be mistaken for measured EirGrid data. Views include a system overview, sector breakdown, a national time-series explorer, a county choropleth map, a buildings drill-down, feeder-stress sample, a nowcast fan, and a consolidated data-provenance page.

The platform is built in PHP (server-rendered pages plus a JSON read API), with vanilla JavaScript, Chart.js, and Leaflet on the client, packaged with Docker. Display-shaped rollups are produced offline by a DuckDB bridge over a Backblaze B2 Parquet lake and served from MariaDB.

Developed by the Energy Informatics Group (EIG), International Energy Research Centre, Tyndall National Institute, University College Cork.