Optimise a Power Plant with Machine Learning: Data Science Lifecycle Case-Study

As intermittent sources of electricity, such as wind and solar, become more and more prevalent in the United Kingdom, generators of other fuel types are becoming further exposed to charges levied on them by the National Grid, whose responsibility it is to ensure that the supply of electricity always meets the demand.

One such generator is Drax, who are the largest single producer of electricity in the UK, burning biomass fuel and coal to generate up to 4000MW of electricity, close to 10% of national demand. Drax must pay a balancing charge, called BSUoS, to the National Grid. BSUoS is applied to every MWh which flows from a generator on the Transmission System and arises from the cost incurred by the National Grid in balancing supply and demand. The charge is applied every half hour and over a year averages around £3/MWh, but the range is wide. Therefore, the cost to Drax of generating a MWh is highly variable and the total annual BSUoS charge can exceed £54m.

This talk will focus on how Elastacloud have worked alongside Drax to develop a data science solution to BSUoS forecasting. The combination of Elastacloud’s expertise in machine learning and cloud technology with the subject matter experts at Drax has resulted in a fully automated system that provides accurate BSUoS predictions, for the next 72 hours, directly to Drax’s traders. This means Drax are now better positioned to manage the volatility of production and optimise their generation portfolio.

YOU MAY ALSO LIKE:

• Glean: Data About Code (SkillsCast recorded in December 2022)
• Re-Thinking Time Series Data Solutions (SkillsCast recorded in November 2022)