A wide swath of people – politicians, news commentators, policy-makers – often decry the growth of renewable energy, saying that it will destabilize the electrical grid. Many of them are motivated to do so out of party loyalty, or in support of traditional fuel sources, or from simple misunderstanding. And they often end up muddying the waters around a rational discussion of renewables and the grid of the future. But, thing is, they make a very good point. Integrating renewables into the grid will be no easy task. For the past century or so, the architecture of the grid has been designed to accommodate certain forms of power generation (namely central power plants transmitting over high-voltage power lines at medium to long distances), to anticipate a fairly steady and predictable growth in demand, and, importantly, to be sized so as to meet peak demand. (See “Acronym Soup”, below, for a quick primer on peak demand.)
The second assumption – that of steady demand growth – no longer holds true. In many areas of the country, even in those with booming economies and increasing populations, demand growth is mostly flat or even declining. This is largely due to the often unheralded success of energy efficiency measures, both those mandated by regulation and those undertaken voluntarily by homes and businesses.
But the other two assumptions are increasingly shaky as well. Two of our major power sources – coal and nuclear – are moving, respectively, toward a rapid death and what looks to be a slow decline, and are largely being replaced by wind farms and solar installations. This leads to huge problems with intermittency: when the wind doesn’t blow, turbines don’t produce electricity; and when clouds cover the sun or night falls, photovoltaic panels sit idle or useless.
And, at least to date, the incentives for installing renewables haven’t aligned with the need to produce at times of peak demand. The goal of keeping power production and consumer electrical needs carefully balanced thereby becomes more and more challenging.
If more renewable energy is to make its way onto the grid, these problems will have to be solved. And given the massive growth in renewables, they’ll have to be solved quickly.
These two problems have found their apotheosis in the famous – well, famous to any energy geek – problem of California’s “duck curve”. The typical demand curve in California looks like the back of a duck, thusly:
Frankly, it doesn’t look much like the back of a duck to, well, to just about everybody, but the name has stuck regardless. Here’s how it works: a good deal of the state’s daytime consumption is met by solar production, both rooftop and utility-scale, so the demand for power from coal, gas, and so on remains lower. But just as demand ramps up in the early evening, the sun is starting to dip below the horizon, and the production from solar panels is declining precipitously. What’s going to meet that demand? For the most part, peaker plants, the old, dirty plants that are incredibly expensive to ramp up so quickly. So the introduction of more solar energy on the grid isn’t necessarily leading to a proportional reduction in prices or carbon emissions. The challenge then, is to find a way to smooth out that demand curve, to shave that duck’s back until it starts to resemble something a little less extreme.
I’d heard about the duck curve quite a bit, but it was a pleasant surprise when my boss at Joule Assets, Mike Gordon, a veteran of the demand response sector, dropped a duck curve problem right in my lap. Pacific Gas and Electric, one of California’s major utilities, had issued a request for NWA (non-wires alternative) solutions for a substation in Huron, a small town in the Central Valley. The substation and the local grid were soon going to be overloaded with the amount of solar electricity produced by local solar farms and rooftop installations, and might even experience “reverse flow”, where power flows the wrong way and, well, something bad happens. I don’t know what that is, honestly: at least in my imagination, transformers explode, cables fly, sparks rain down everywhere? Whatever it is, it must be bad, because PG&E would normally solve the problem by investing upwards of $20 million in the local grid and substation to prevent it. Rather than investing all that capital, and then rate-basing the investment so as to recover it from consumers, PG&E was seeking out NWAs that would (a) potentially be easier to implement, and (b) cost less.
This is a textbook duck curve problem: solar overproduction during the day with rapid ramping in the evening hours. Unfortunately, the RFP had just been offered to Joule, and I had about a week or so to respond. I set about furiously researching everything I could about Huron, and quickly discovered it wouldn’t be an easy target. It’s an economically depressed agricultural town with the lowest median income in the whole state. It contains almost no large- or even mid-scale commercial facilities that could provide big chunks of evening demand response. It has fairly low homeownership, due to a population that doubles during growing and harvesting season due to migrant labor, which also means little easy targeting for DR. And much of its demand, both evening and daytime, derives from agricultural sources, likely largely irrigation pumping, on-site processing, and the like. These might be great opportunities there for shifting usage from one segment of the day to another – known as load shifting — but Joule doesn’t have expertise in agricultural DR.
It felt a bit like trying to squeeze juice out of a dry lemon. I spent hours and hours poring over Google StreetView images of the town, scanning rooftops and storefronts, hoping against hope that there might be a magic bullet or two I could work with. I spent time in the weeds of slightly more arcane solutions like ice storage, a fascinating new technology that involves chilling a large block of ice during nighttime hours, when energy is cheap and clean, and then using it to run daytime AC, when energy costs more. (Keep a close eye on ice storage: my guess is that it’s on the cusp of becoming a feasible and easily-scalable product at both residential and commercial levels.) We considered wading into the waters of vehicle-to-grid (V2G) technology, where the batteries in electric cars can be used both to soak up the excess power from solar overproduction and feed back into the grid during peak times. But V2G is still is in infancy, and many of the technological, regulatory, and financial kinks have yet to be solved. The solution I settled on was a combination of residential DR and small-scale distributed battery storage, both proven technologies that, with a nimble back-end platform, could work together as a combined resource when necessary.
When we think of clean energy, we often think of wind turbines, solar panels, or maybe something new and fancy like tidal power. But at the end of the day, the only way we’ll be able to integrate them at massive scale onto the grid is to find ways of (a) storing some of that energy for use when it’s needed later, and (b) shifting usage so that demand curves smooth out. I felt pretty lucky to be tackling that challenge in a small and discrete way. It felt surprisingly, and delightfully, like developing a case solution for a class at Stern, but with this difference: it mattered, not only to Joule’s bottom line, but to the development of the clean energy grid of the future.
Gideon Banner, MBA ’18