Production Forecasts for Wind & Solar Farms

The Australia National Electricity Market (NEM) provides a spot market for wholesale electricity distribution.  In order to provide a secure grid, supply and demand are matched every 5 minutes - hence operators need to report production forecasts every 5 minutes. 

For Wind and Solar farm operators, providing accurate forecasts is inherently challenging due to the nature of changing weather patterns. This significantly increased the risk of providing inaccurate production forecasts leading to penalties imposed by the regulator.

In partnership with Worley, Arundo worked with Monash University in Melbourne to develop a machine learning algorithm that produces production forecasts for wind and solar farms using various data sources to improve forecasting accuracy.  Arundo's Edge Agent solution was connected to the farms’ assets, to ingest live production data, and provide containerisation of the model to execute the algorithm. The results were fed through to the market regulator (NEM) via API for verification of the forecast accuracy. 

The combined offering is currently deployed on one wind farm with 43 turbines (180 tags), and one solar farm with 32 inverters.