Most power interruptions are relatively short, typically lasting minutes to hours, but they are also unpredictable. It can be difficult to assess the potential of solar photovoltaic (PV) and energy storage systems to mitigate these interruptions without the ability to account for both the unpredictability of these events as well as typical patterns in when and for how long they tend to occur.
To address this challenge, Berkeley Lab has developed the Power Reliability Event Simulator Tool, a publicly available model that can be used to simulate the occurrence of short-duration power interruptions in any county in the continental United States. An accompanying case study intends to show how PRESTO can be used to analyze the performance of a PVESS in providing backup power during short-duration interruptions.
GO DEEPER: Check out the Factor This! energy storage podcast playlist, including episodes on battery storage, long-duration energy storage, gravity storage, and more. Subscribe wherever you get your podcasts.
PRESTO simulates the occurrence of interruption events over a large number (up to 20,000) of simulation years. In each simulation year, PRESTO produces a time series of interruption events with stochastic frequency, duration, and timing. The model relies on a set of probabilistic functions trained on historical hourly power interruption data at the county level, obtained from PowerOutage.US for the years 2017-2021. As a result, the patterns and statistical properties (e.g. mean, standard deviation) of simulated power interruption events match the statistical properties of actual historical power interruptions for the selected county, Berkeley Lab said.
Upon selecting a county for analysis, PRESTO loads default average interruption duration and frequencies based on power reliability data from the Energy Information Administration, which the user can use or override. The user can visualize the statistical properties of the produced dataset – including distributions of duration and frequency as well as seasonal and time-of-day profiles, and download the detailed results in a CSV file. In addition to the web-based interface, an Application Programming Interface is also available for batch processing of a larger number of counties.
PRESTO is primarily designed for use in Monte Carlo or other probabilistic analyses of the impacts of short-duration power interruptions. One intended application is to assist in evaluating the backup power capabilities and customer reliability value of onsite PVESS or other strategies for mitigating short-duration power interruptions. To demonstrate its use in this context, Berkeley Lab developed an accompanying case study analysis that explores several key determinants of PVESS backup power performance, for representative single-family homes in three distinct regions of the U.S.
Among other factors, the case study highlights how PVESS backup performance (defined as the percentage of backup load served), is impacted by how the customer operates the battery under normal day-to-day conditions, which affects the battery’s state of charge whenever the power interruption occurs.
In the example shown in Figure 2, backup performance is especially low for interruptions occurring during evening hours. In this case, the customer is assumed to be taking service under the utility’s residential time-of-use rate, which has a peak period extending from 4-8 pm. This incentivizes the customer to discharge its battery during that timeframe and leaves little charge remaining for any power interruption that happens to strike soon thereafter. The case study also illustrates how PVESS backup performance is impacted by system sizing, backup load configuration, whether or not the customer is able to charge from the grid, and regional variations in heating and cooling demand.
PVESS and grid outages
A report released last year from the Energy Department’s Lawrence Berkeley National Laboratory and National Renewable Energy Laboratory looked at the backup power capabilities of behind-the-meter solar-plus-energy storage systems (or PVESS). The analysis found that backup performance depends, “first and foremost,” on PVESS sizing and the set of critical loads selected for backup.
It said that if heating and cooling loads are excluded from backup, then a PVESS with as little as 10 kWh of storage (the lower end of sizes currently observed in the market) can fully meet basic backup power needs over a three-day outage in virtually all U.S. counties and in any month of the year.
But, if critical loads include heating and cooling, then a PVESS of that size would meet 86% of critical load. A larger PVESS with 30 kWh of storage (the upper end of sizes currently observed in the market) would meet 96% of critical load.
The study said performance tends to be lowest in regions where electric heating is common (the southeast and northwest), and also in regions with large cooling loads (the southwest and parts of the southeast).
Backup performance can also vary within regions, based on differences in the building stock. The report offers performance differences based on heating technology (electric resistance vs. heat pumps vs. fossil heating), building infiltration rates (the leakiness of the building), air-conditioner efficiency and temperature set-points.
