Background

This section is the context every other page assumes: NGED's electricity network, what NGED has asked for, how distribution network operators have traditionally forecast, and the two limits — measurement artefacts in the telemetry and a fragmented literature — that shape the rest of the design. Design philosophy states the principles that answer to this problem; architecture describes what is built to solve it.

• NGED's network and its data — the primary substations, bulk supply points, and grid supply points NGED operates, the generation connected to them, and the false zeros, stuck values, missing-data gaps, and months-long commissioning ramps in the trial-area telemetry.
• Weather products surveyed for the weather-product studies — which weather products the project uses, and the forecast models, reanalyses, and satellite retrievals that could join the weather-product studies, with what each carries. A survey, not a list of what the project will ingest.
• Requirements — the phased rollout from 32 time series to roughly 2,500, and the objectives NGED prioritises.
• The manual heuristic forecast — the historical-analogue method that, until recently, was the normal approach among distribution network operators, with no weather model and no machine learning, and the baseline our own forecasts are measured against.
• Switching events — why NGED's meshed network is operated as a radial tree, and why reconfiguring that tree moves load between substations without warning.
• The state of the art in energy forecasting — what the published literature does and does not settle, and where Flexpectation's plan sits against it.