The Duck is Learning to Fly
In 2014, I authored a novel concept, Teaching the Duck to Fly. In that publication, I posited ten strategies to better align resources to loads and loads to resources to enable a renewable-powered grid to provide reliable electricity service.
We are seeing nearly all of those strategies employed today – and some far beyond anything I predicted. The result has been a very different power system than most people imagined possible.
California is the center of attention for many reasons. First, it was an early pioneer in energy efficiency and renewable energy. While Germany was perhaps even more enthusiastic for solar, through generous feed-in tariffs, and Australia and Hawaii have twice the rate of solar penetration, California was and is a big player. With 40 million people, it’s big. As the world’s sixth largest economy, it’s big. And with the nation’s second-highest electricity rates (after Hawaii, which is still petroleum-dependent) it’s a place of opportunity for solar.
California has been a big player in battery storage as well. Today, as much as 15,000 MW of power can flow into batteries – both stationary and those in electric vehicles – during the solar day. And as much as 6,000 MW does flow back to the grid during high-cost hours from the stationary batteries. Perhaps in a few years, as vehicle-to-grid (V2G) technology is implemented widely, that figure will double or triple. More about EV batteries a little later.
Let’s compare where we THOUGHT we would be to where we ARE. Thanks for the folks at gridstatus.io for the inspiration here; read their piece on the role of batteries: https://blog.gridstatus.io/caiso-batteries-apr-2024/
Here’s what the “Duck Curve” looked like a decade ago. The forecast was that peak loads would continue to grow year by year, while mid-day loads would be increasingly served by solar, creating a scary “ramping” requirement for the power system to go from very slack at mid-day to very high loads in the early evening.
Source: California ISO
Power planners were genuinely worried about how they could “ramp” that fast. With combined-cycle gas generation as their only tool, it was indeed challenging; today, with thousands of megawatts of battery power available, it’s much easier.
Note that the mid-day low as predicted to drop from 20,000 MW in 2012 to 12,000 MW by 2020, and the peak loads in the evening were predicted to rise from 24,000 MW to 26,000 MW. Now let’s take a look at what has actually happened.
Source: Gridstatus.io
This graphic, at first glance, looks similar to the 2012 version. But look close: it’s VERY different. The mid-day net load has indeed dropped sharply. But the evening peak load is ALSO lower than before. Peaks did NOT continue to grow. The required "ramping" is half what was forecast a decade earlier. There are a lot of factors at work here: time-varying prices, customers with solar and batteries, more efficient air conditioners and lighting, and overall higher electricity prices are clearly significant drivers.
The California resource mix has changed dramatically since 2014. In addition to 16,000 MW of utility-scale solar, California also has 16,000 MW of behind-the meter solar meeting customer loads, and backfeeding to the grid under legacy net energy metering agreements. 32,000 MW of solar generation during the day that disappears as the sun goes down. More than the total CAISO load depicted in these graphs (and yes, about 20% of this is in the service territories of the non-CAISO utilities like LADWP and SMUD).
One look at the daily power supply for the CAISO makes the utility solar picture clear – but does not reflect the nearly equal amount of customer sited solar at work. See that bright sunspot in the middle: counting the distributed solar at work, it is really TWICE as big.
Source: CAISO
Batteries and Intertie Transmission to the Rescue
The system has adapted to this in two ways. The first is imports and exports. The graphic above shows large imports in the overnight hours. It does not show large exports during the solar day. In essence, California is trading solar power at mid-day for hydropower from the Northwest overnight. Both regions get renewable power. And, when the Northwest gas plants and coal plants are shut down, it really legitimate to argue that it’s a solar for hydro swap. If the thermal plants were running in the PNW when power is flowing south, well, it’s a little harder to argue that California is getting hydro.
First, the import and exports, from a CAISO perspective: up to 9,000 MW of imports overnight, and up to 6,000 MW of exports at mid-day.
Source: California ISO
Second, the battery operations: up to 5,000 MW of charging during the solar day, and up to 6,000 MW flowing back to the grid mostly in the early evening.
Source: Gridstatus.io Data for May 29, 2024
About 30,000 MWh of batteries charged during the day, discharged in the evening. And providing frequency regulation and voltage support services at all hours, reducing the need to keep gas-fired power plants idling (and wasting fuel) to provide reserves.
Non-Utility Solar Is Missing From the CAISO Reporting
What these reports are missing, however, is the role of behind-the-meter storage and solar. The Gridstatus reports show only the grid-controlled solar and storage resources. Here’s a graphic of the growth of non-utility solar in California:
California Non-Utility Solar by Year
https://www.californiadgstats.ca.gov/charts/
This is nearly equal to the installed capacity of utility-scale solar. Distributed solar produces about 10% less energy per installed kW, mostly due to shading issues and the lack of tracking during the solar day. But it also meets loads at the distribution system level, avoiding 5% to 10% in line losses incurred getting power from central solar plants in the desert to the cities and towns where the power is consumed. So that’s about a push in terms of value in meeting load.
An increasing number of these system have batteries installed, including the majority of systems build under the new NEM-3 regulations, where backfeed to the grid gets just pennies per kilowatt-hour. NEM-3 customers receive a fraction of the full retail rate that customers under NEM-1 (pre-2015), or nearly the retail rate under NEM-2 (pre-2023).
Here’s how a solar customer with batteries programs their system to charge when the sun is shining, and displace on-peak purchases from the utility during the high-cost hours. We’ll see a lot more of this as people learn to use the batteries in their EVs as storage systems.
Source: Clean Power Research
The overwhelming majority of existing distributed solar in California is in residential applications. But, with the advent of NEM-3, solar economics for residential consumers (who typically export half of their generation to the grid during the solar day, and receive half from the grid at other hours), that these installations will slow down sharply. But I believe it is likely that there will now be a growth spurt for systems at commercial loads – office and retail – where the systems can be sized to use all generation on-site, thus avoiding the retail rate. For example, a supermarket that users 200 – 400 kW of power, varying through the day, can install 200 kW of solar without worrying about exporting a single kWh. For a discussion of how this worked out in Hawaii, see: https://energycentral.com/c/um/california-will-follow-hawaii-and-see-shift-solar-most-will-now-go-commercial
The Role of the Electric Vehicle Fleet
California has ANOTHER huge battery component that is barely involved here. Yes. There are 1 million battery electric vehicles in California, ranging from plug-in hybrids with 10 kWh batteries up to Ford Lightning pickups with 130 kWh batteries. At an average of 60 kWh per vehicle, that’s another 60,000 MWh of battery capacity, waiting to be tapped for grid service support.
Think about that: if HALF of the EV batteries already in California were available to provide backfeed to the grid during high-cost hours, they would provide as much benefit as all of the existing utility-scale batteries. And this capacity is expected to double in the next two years, and then double again.
Finding ways to harness the V2G capability is an important challenge that provides an opportunity to greatly reduce reliance on polluting generating resources like natural gas and nuclear energy. Fortunately, by 2026, V2G capability will be built into most new EVs, and this opportunity can be realized as soon as utilities put the technology in place to use it. This should be a high priority.
A typical EV owner drives about 40 miles per day, using less than 20% of their car battery. Yes, a few days per year they take a road trip and need the whole battery, but 95% of the time they do not. And people generally know a day in advance if they will be doing a longer trip, so they can allow their battery to serve the grid when they don’t need it all, and reserve their EV capacity when they do. Getting a second revenue stream, by providing storage capacity either to the grid or to their own house with these batteries-on-wheels (really nice wheels, in most cases) is an attractive opportunity.
I’ve thought about how I would personally manage my EV if I were a California customer. First, I would use it for transportation. That would leave me with excess battery capacity about 350 days per year. Then I would use it to shape the power needs of my own house, which typically uses about 20 kWh per day. If I were in a sunny place, I would use my rooftop solar to charge it during the day and using the stored power at night. If I lived in a deep forest, I would buy from the utility at low-cost hours to serve my needs at all hours.
That would use about half of the surplus capacity above what I need for driving most days. The remainder of the capacity I would make available to the grid. Except when big storms are approaching, when I would hold it in reserve for a possible distribution power outage. But that’s only 1 or 2 events per year.
Let’s assume that storm does come, and the distribution outage does happen. That is likely to be localized; my own experience is that while distribution outages can last many days, transmission outages are usually restored in a few hours. That means I’ll be able to drive my battery-on-wheels to a charging station served by underground power lines, and recharge. Just as I drive to the store for milk, eggs, fruit, and toilet paper, I can drive to the store and get more electricity and bring it home. In fact, Walmart makes it easy, as most stores now have high-speed charges (plus milk, eggs, fruit, and toilet paper). If I have a small stationary battery at the house, nothing will be left unpowered while I’m getting more juice. So I can recharge my backup battery pretty easily during a power outage.
Better Battery Usage Means Less Curtailment of Solar
As this type of behavior becomes more commonplace, there will be more and more solar power that is now spilled (curtailed) put to beneficial use. The amount of curtailment is fairly high today, particularly in the Spring, when solar production is prolific and loads are low (not much air conditioning). Here’s solar curtailment for the same period used in the examples above: Smart EV charging and smart battery charging can reduce this wastage.
Summary: Little Things Add Up
This growth of the battery market and battery role is just one aspect of the “Duck” learning to fly. It’s a big aspect. Much bigger than I assumed a decade ago.
But the other pieces are also moving forward. In particular, there are now a dozen companies competing in the ice storage / chilled water storage space for large commercial air conditioning systems. I remain optimistic that this will surge as the engineering community becomes more familiar with this technology.
Responses to time-varying rates also include storage heating, smart water heaters, smart dishwashers, doing laundry at low-cost times, and businesses scheduling their energy-intensive activities into low-cost periods. Plus AI firms shifting their surging “model training” load into low-cost hours. All of these are happening.
It could happen faster. It could happen cheaper. But it’s happening.