Home › Projects › Peak energy demand prediction

Research

Peak energy demand prediction

A research project predicting daily peak electricity demand on the South West Interconnected System from half-hourly interval and weather data.

The problem

Large electricity users on the South West Interconnected System pay network charges based on their draw during the four highest demand intervals of the hot season. Knowing in advance whether today is likely to be one of those days lets a site shed load at the right moment and avoid a year of higher tariffs.

What we built

  • Extraction and cleaning of half-hourly grid demand alongside temperature and calendar features.
  • Exploratory analysis of the drivers of demand and the resulting cost of energy.
  • Gradient-boosted models at daily and half-hourly resolution that predict the peak for each day, with a gauge for the likelihood that the day lands in the top four.
  • A dashboard concept giving stakeholders a real-time view of demand and the peak-day signal.

Why it matters

This is the pattern behind most of our forecasting work: join operational data to external signals, model the quantity that drives cost, and surface it where someone can act on it.

More Work
View All Projects