It’s about load. In our long-term reliability assessments, we tend to think of peak demand as a fixed quantity. We adjust it up for new loads such as EV growth, or down for demand response and DER growth, and we adjust it for weather. We consider the variability from wind, solar and hydro during that high demand period, perhaps assigning a capacity value to different weather-related resources. And when we consider adding a gas peaker to the mix to ensure we can meet peak demand, we may use a planning reserve margin (eg 15%) to give us a cushion of safety. Utilities planning for high penetrations of wind and solar may even be conducting loss-of-load expectation analyses using data from multiple weather years to examine what happens to resource adequacy when it’s a low wind/high hydro or low hydro/high solar year. In any case, we take this fixed quantity of peak demand and we plan for it and make investment decisions based on it.
We have treated load as price inelastic. This isn’t necessarily wrong – it reflected consumers’ behaviors. When I turn on the light or the stove, I want that service – now. And for services that could be delayed – well, modest price spreads between peak and off-peak periods may save a modest percentage of my electricity bill, but the absolute savings is so small compared to say, my Starbucks habit, that it does not motivate me.
Big data and artificial intelligence are changing all of that. My Google Nest thermostat knows enough about my comfort needs that Google could use my home as part of a larger aggregation and provide regulating reserves without me even knowing it. Or they could provide peak shaving on a hot summer day and reduce peak demand on the grid. I’ll also point out that I paid them $250 so they could have this capability (instead of buying a $50 thermostat from the local hardware store).
Electrification of the transportation and heating sectors is the second big driver for change. Our EV more than doubled our household electricity consumption. These new loads have significant power demands and they may lead to an increase in system peak demand. But importantly, they also have inherent flexibility. The power industry is aware of and eagerly anticipating the grid integration benefits from controlling EV charging, water heating and space conditioning.
What does all this have to do with renewables? Some utilities are interested in future systems with very high penetrations of wind, solar, hydro and storage. Planners already know there is significant difference between a good and bad hydro year. Add the permutations of good and bad wind and solar years to that, plus the climate change-induced extreme weather events, and there will be a lot of uncertainty of output levels of wind, solar and hydro during system peak for any given year. The industry will need to carefully examine the costs and benefits of a new gas peaker or a new storage facility and determine how much over-building of the system is appropriate and acceptable. But it’s not a stretch to imagine a future where on that hot summer afternoon, system peak will be determined not by peak demand, but by however much wind, solar, hydro, and other generation (storage, thermal, interconnections, etc) happens to be available. And it will be the price elasticity of load that determines which loads are served and which are compensated to provide peak shaving. So instead of thinking of demand as a fixed quantity, with various generators meeting that demand, we would think of total generation as a fixed quantity, based significantly on the weather that day, and various loads being served or not served. Or one could think of loads as negative generators.
This is already being introduced. Portland General Electric recently proposed a $1/kWh peak time rebate. Baltimore Gas & Electric has a $1.25/kWh peak time rebate. This level of savings becomes much more motivating – it gets closer to pricing on toll roads (or my Starbucks habit).
Price elasticity of load turns resource adequacy on its head. The loss-of-load-expectation metric of one day in ten years is no longer meaningful because you never have a loss-of-load event if all load has price elasticity. How the resource adequacy concept will evolve remains to be seen. There will always be some loads (hospitals, military facilities, etc) that are price inelastic and need firm power; on the other hand, those may be the same types of loads that already have backup generators. Behind-the-meter storage may also become a more attractive option for consumers.
Load is going to be a key resource in the future. Our models and analysis techniques will have to evolve to reflect that. And it will help us solve a lot of the problems that many are concerned about, as utilities move to very high levels of renewable energy.