The Weather Effect Decomposition tool provides users with the ability to better understand the impact of weather on demand. Users can decompose observed demand based on observed weather, seasonal normal weather and other causal variables. The result is a breakdown of demand into the portion that would have occurred under normal weather circumstances and that which occurred due to actual weather deviations from seasonal norms.
Many factors influence changes in year-on-year power demand such as weather, population, state of the economy, behind-the-meter renewable generation trends, etc. Power demand has been increasing year-over-year (YoY) in ERCOT for the last decade. The 2022 and 2023 summers showed unusually large YoY growth when compared to prior years.
We know that the population and economy of Texas have been growing over the last several years, but can that explain all the growth in metered load? How much did the weather that occurred in the summers of the last two years impact power demand? Using a well-specified statistical power demand model, hourly observed weather data, and hourly normal weather data, we can separate the effect of weather from the observed load data. The result is what we call weather-adjusted load data which represents true base load if the weather conditions had been close to seasonal normal. We can analyze YoY weather-adjusted load to see what growth in ERCOT was year to year controlling for weather conditions.
Weather-adjusted load normalizes the load series for the annual fluctuations in weather so we can separate non-weather-related load growth or contraction from weather effects.
Comparing the figures 1 and 2, YoY weather-adjusted load growth for 2022 and 2023 is half as large as it appears for metered load. The only year that didn’t exhibit weather-adjusted load growth was 2020, which can be explained by COVID policy demand reduction. We can see this clearly in figure 3.
To calculate the weather-adjusted load, we first calculate the “weather effect” of using seasonal normal weather in our model instead of the observed weather. In figure d, we can see the outlier weather effects of 2022 and 2023 compared to the previous four years.
The weather effect is stronger during the peak demand evening hours than in the morning and midday hours.
Figure 5 quantifies how much demand on average per hour was added relative to a normal summer in ERCOT.
While it’s clear by any measure that base load in ERCOT is growing, the unusually warm summers in 2022 and 2023 make the growth rate appear larger than when examining weather-adjusted load. By stripping out weather effects from metered load, we can set expectations for future summers in ERCOT at a more reasonable level.
This analysis is only one example of how using the Weather Effects Decomposition functionality can provide additional insight and value to traders, utilities, aggregators and others needing to study energy demand.
For additional information please contact us directly at firstname.lastname@example.org.