Hydroelectric Generation Revenue: A Case Study
Updated: Nov 29, 2021
Last month’s extreme weather focused attention on Texas, but less unusual weather conditions can also drive price volatility in electricity markets several times per season. In New England, for example, colder-than-expected conditions on 01/29/2021 led to an increase in the price of natural gas and high demand for heating, especially in the morning hours. Here we discuss the reasons behind a one-day January price spike in ISO New England as a case study of how better real time price forecasting can optimize value to responsive resources - in this case hydro - while potentially keeping oil generation offline.
Diagnosing What Happened in the Real Time Market
Electricity prices surged across the ISO New England market on Friday, 01/29/2021 for the HE10 trading window. The morning price surge was a system-wide phenomenon, with prices rising to over $160/MWh at all ISO aggregation zones by 10AM. Between 5 and 6 pm, as another peak was starting to form, additional hydropower was dispatched into the grid. However, there was substantially less demand than the morning cold-driven peak and the HE19 price peak remained $40/MWh lower than the morning’s high.
While dispatch data from individual facilities are not public, our analysis of regional pricing data indicates that the Bangor Hydro region in Eastern Maine was likely responsible for a substantial fraction of the additional generation during the evening price spike. As Figure 1 shows, the Bangor Hydro region had an identical price shape during the morning hours of the 29th relative to the rest of the market. In the afternoon, however, Bangor prices dropped sharply, a reliable indicator of increased supply. The bottom curve in Figure 1 represents the Bangor Hydro region and shows the divergence from other regional prices beginning in the early afternoon.
Figure 1. Real Time Price Shape for 01/29/2021. The Bangor Hydro region diverged significantly from the average price behavior in ISO-NE. The Price Volatility in Context
It is estimated that 2650 MWh of reserved hydro were dispatched in HE 18, with a peak dispatch of around 350MW by extrapolating from ISO-level statistics in Figure 2. While hydro was discharged throughout the day, better price forecasting and local elasticity analysis could have optimized revenue and potentially kept oil generation offline. The elasticity of demand during similar historic days averages 0.028.
It is possible to estimate what an optimal dispatch strategy might have looked like if asset managers had access to improved forecasts. Looking at an average demand curve, pricing data, and dispatch data, it is possible to deduce how responsive the price in the region was to the dispatch: the average sensitivity is set to an upper bound of $0.23 per MW dispatched so that savings are not overstated.
With this in mind, the actual revenue for hydropower produced in the region is thought to have been $185,843. Following the optimal dispatch strategy outlined in Figure 2, the revenues could have been increased by at least 11% to $206,554. This analysis was completed using our quadratic program dispatch optimization tool assuming fixed total discharge of 2650 MWh, a 350 MW maximum discharge, and no demand elasticity yielding a lower bound on revenue increase.
Figure 2: Actual hydropower dispatch vs estimated optimal dispatch in the Bangor Hydro region. Events like this highlight the benefits that could be captured by optimizing dispatch strategies using improved forecasts. Hydropower plants would have optimized their economic value contribution and possibly kept oil generation offline as extreme morning cold led to natural gas scarcity in New England.