Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Unlocking the Potential of Renewable Energy Through Curtailment Prediction

Bilge Acun, Brent Morgan, Henry Richardson, Nat Steinsultz, Carole-Jean Wu

TL;DR

The paper tackles the challenge of renewable curtailment by arguing that accurate, high-granularity prediction of when and where curtailment occurs can unlock load-shifting-based emissions reductions. It proposes two complementary ML-enabled tasks: detecting curtailment from nodal Locational Marginal Pricing (LMP) data and forecasting curtailment at fine temporal and spatial scales, emphasizing 5-minute granularity and nodal-level precision. A key contribution is the compilation and standardization of historical curtailment and LMP data for North American ISOs, plus a framework for evaluating forecasts in terms of their practical impact on load shifting and emissions. The work highlights a practical call to action: more accessible ground-truth data, data-sharing of nodal LMPs, and carbon-aware scheduling standards to accelerate decarbonization through demand-side flexibility.

Abstract

A significant fraction (5-15%) of renewable energy generated goes into waste in the grids around the world today due to oversupply issues and transmission constraints. Being able to predict when and where renewable curtailment occurs would improve renewable utilization. The core of this work is to enable the machine learning community to help decarbonize electricity grids by unlocking the potential of renewable energy through curtailment prediction.

Unlocking the Potential of Renewable Energy Through Curtailment Prediction

TL;DR

The paper tackles the challenge of renewable curtailment by arguing that accurate, high-granularity prediction of when and where curtailment occurs can unlock load-shifting-based emissions reductions. It proposes two complementary ML-enabled tasks: detecting curtailment from nodal Locational Marginal Pricing (LMP) data and forecasting curtailment at fine temporal and spatial scales, emphasizing 5-minute granularity and nodal-level precision. A key contribution is the compilation and standardization of historical curtailment and LMP data for North American ISOs, plus a framework for evaluating forecasts in terms of their practical impact on load shifting and emissions. The work highlights a practical call to action: more accessible ground-truth data, data-sharing of nodal LMPs, and carbon-aware scheduling standards to accelerate decarbonization through demand-side flexibility.

Abstract

A significant fraction (5-15%) of renewable energy generated goes into waste in the grids around the world today due to oversupply issues and transmission constraints. Being able to predict when and where renewable curtailment occurs would improve renewable utilization. The core of this work is to enable the machine learning community to help decarbonize electricity grids by unlocking the potential of renewable energy through curtailment prediction.
Paper Structure (8 sections, 5 figures, 1 table)

This paper contains 8 sections, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. ML for Curtailment Prediction
  3. Detecting Curtailment Using Nodal Locational Marginal Pricing (LMP) Data
  4. Forecasting Curtailment
  5. Call to Action
  6. Supplementary Figures
  7. Forecasting Metrics
  8. Additional Datasets

Figures (5)

  • Figure 1: Median (solid line) and 25%-75% range (shaded) of curtailment values, aggregated by time of day, for SPP (primarily wind curtailment) and CAISO (primarily solar curtailment).
  • Figure 2: Left: Curtailment frequency as a function of the minimum LMP for different curtailment amounts (MW). Right: Heatmap of the times below LMP threshold at nodal level in SPP.
  • Figure 3: SPP wind curtailment in 5-minute steps on a random day.
  • Figure 4: CAISO curtailment frequency and amount shown as time series from 2019-2023.
  • Figure 5: Illustration of load shifting forecast metrics.