Predictive software could reduce the amount of energy storage needed to transition to a economy that includes growth in renewable energy, according to a new study from the National Renewable Energy Laboratory.
The study, Shifting Demand: Reduction in Necessary Storage Capacity Through Tracking of Renewable Energy Generation, proposes an alternative approach to bridging the mismatch between peak demand and peak generation in an electric system that is increasingly relying on intermittent sources of generation, such as wind and solar power.
One solution to that problem is to store peak generation and dispatch it during times of peak demand using utility scale battery energy storage systems but, as an alternative, the study’s authors analyzed a means of shifting demand by using a forecast-aided predictive control algorithm.
While battery energy storage systems can improve dispatchability, the study’s authors noted that the technology has “several challenges,” including inadequate safety validation, degradation of the batteries and “most crucial,” cost of the systems. “Due to these challenges, it may be beneficial to limit the total BESS capacity required for deployment,” the authors wrote.
Alternatively, the use of forecast-aided predictive control can shift demand “considerably to more closely track” a renewable energy signal, the study found. “This significantly reduces the size of the required utility-scale BESS,” the authors said.
The study analyzed a forecast-aided predictive control algorithm that is used to autonomously control both electric vehicle charging stations and the heating, ventilation, and air conditioning systems in buildings.
Electric vehicles and hybrid electric vehicles, along with heating and cooling systems for buildings, already account for roughly 10 and 40 percent of electric demand, respectively, and both are poised to increase with calls to decarbonize the economy. So, shifting demand for electric vehicles and buildings is “imperative” and provided the framework for the study, the authors said.
“We have an idea of how many people will be in the building, and then from there we can get an estimate of how many electric vehicles will be arriving at the charging station,” Dylan Wald, a graduate intern at the National Renewable Energy Laboratory, a Ph.D. student and lead author of the study, said in a statement. “Everything is intertwined, and we can leverage this interconnectedness.”
The study on forecast-aided predictive control was based on research the National Renewable Energy Laboratory did last year that showed electric vehicle charging and buildings can work together to provide services to the grid. The improved algorithm took that work a step further by including forecasts to improve real-time tracking ability by taking into account how much wind and solar power will be generated, as well as the temperature and time of day and week in order to estimate the energy demand for a building and charging stations, the study’s authors said.
The analysis indicated that under days of more intermittent renewable generation, forecast-aided predictive control performed adequately, however, the performance of the algorithm decreased during weekends when demand is less significant and less flexible. The analysis also found that the forecast-aided predictive control performance is sensitive to the accuracy of the forecasts incorporated in the algorithm.
“This work is showing us you don’t always need such a big battery,” Jennifer King, a research engineer at the National Renewable Energy Laboratory and co-author of the study, said in a statement. “You likely still need a smaller battery."
“That’s a huge implication because we may run into supply chain issues with batteries needed for the grid or for EV charging. We need a different solution,” King said.