New method helps renewable energy groups bid better in power markets

A new approach called PL-MBRO improves economic outcomes for renewable-based virtual power plants.

Renewable-based virtual power plants (RVPPs) combine renewable energy, storage, and flexible demand. These groups bid in day-ahead and secondary reserve markets. They must deal with uncertain market prices, demand, and renewable generation.

The new PL-MBRO method uses historical data to create specific deviation bounds. It also automatically finds uncertainty budgets from empirical cluster probabilities. This replaces older methods that used fixed percentile-based bounds and externally specified budget vectors.

The method uses k-medians clustering to divide historical daily profiles into probability layers. A cluster median defines a balanced representative. Favorable and adverse within-cluster percentiles define optimistic and pessimistic representatives.

The uncertainty budgets are mapped to integer full-horizon uncertainty budgets over 96 quarter-hourly periods. The robust inner problem finds periods where deviation layers are most adverse to the schedule.

Tests show PL-MBRO achieves better realized economic outcomes than the percentile-based MBRO benchmark. Improvements reached up to 92.7% in-sample and 57.3% out-of-sample relative to the absolute value of the benchmark realized profit.

Key facts

  • PL-MBRO achieves up to 92.7% in-sample and 57.3% out-of-sample improvements in realized profit.
  • The method uses k-medians clustering to partition historical daily profiles into probability layers.
  • Uncertainty budgets are mapped to 96 quarter-hourly periods.

All sections

Edition