by Julian Spector
February 20, 2020

When a group of serious academics takes aim at a promising new energy storage policy, it's our job to pay attention. 

That's what happened recently, as a piece of research concluded that the Clean Peak standard enacted by Massachusetts will do little to reduce carbon emissions from energy storage operations, especially compared to a price on carbon. The findings deal a potential blow to the policy, which has "clean" right there in the name. 

But that blow shouldn't be fatal. For one thing, the researchers agree that storage has a role to play in a cleaner grid, and they offer suggestions for improving the emissions impact of the policy. Furthermore, as I noted in this news story, the critique picks a specific metric — emissions from a battery plant — that is just one of the goals the policy was meant to serve.

This is no idle debate, seeing as the policy is in effect in the major market of Massachusetts, and has been discussed in several other states to boot. It made a pivotal intellectual contribution to clean energy policy by moving the conversation from how to add cheap, bulk renewables to how to deal with the hours that drive the greatest cost and a lot of emissions, too.

This week in Storage Plus, we're digging deep on the theory underpinning the Clean Peak and its critique, and parsing the more and less effective claims against it. If you're interested in how state policies should be fostering energy storage expansion, or just enjoy a good clean energy policy smackdown (it's been a while since the great 100 percent renewables versus 100 percent clean energy debate simmered down), this one's for you.

From proposal to law in two years

First, a little background.

The idea of a clean peak standard burst onto the policy scene in 2016, as Arizona's ratepayer advocate contemplated a future of cheap solar power but expensive peak demand served by natural gas. Renewable portfolio standards had long been popular to spur wind and solar installations, but they require a share of generation for the year; they don't address specific needs of the grid at certain times of day.

Arizona was staring down expensive purchases of new gas peaker plants to serve a few hours of creeping summer peak demand; a clean peak standard was meant to save ratepayers money while shifting peak power to cleaner sources.

“This doesn’t get rid of the [renewable portfolio standard] by any means — it’s more of an add-on, a way to send time-based price signals,” said Lon Huber, an architect of the policy while working at Strategen Consulting on behalf of the Arizona ratepayer advocate.

The idea made it into Arizona regulator Andy Tobin's grid modernization proposal but has not been adopted in that state. Instead, it caught the attention of Massachusetts Gov. Charlie Baker; he proposed it, and lawmakers included it in a clean energy law passed in 2018. It went into effect this year.

The rule, as written in Massachusetts, awards credits to clean energy sources — renewables, energy storage and demand response — that provide power during designated peak periods, based on historical electricity demand. Retail electricity providers must buy credits to match 1.5 percent of their electricity sales in 2020. That target ramps annually after that.

Solar and wind power generate when the sun is shining and the wind blows; Massachusetts' Clean Peak rewards clean resources that serve high-demand hours, typically late afternoon and evening. That's why it's often discussed primarily in terms of energy storage: because that technology can guarantee delivery during the desired times.

Study questions emissions benefit

Published last month, the new analysis of Massachusetts' Clean Peak rule finds that it barely reduces carbon emissions compared to business as usual. The authors, from Columbia University, New York University School of Law and the grid-emissions-tracking nonprofit WattTime, pointed out the conceptual shortcomings of the policy and found that carbon pricing would produce significantly better emissions reductions.

The authors argue that, instead of looking at the average emissions of the grid when a battery plant charges, analysts should focus on emissions from the "marginal" unit that will be dispatched to meet the battery's needs. Viewed through that lens, the emissions profile of the New England grid seldom allows for a clear arbitrage between low- and high-emission power production.

"For energy storage to reduce emissions, there has to be a significant difference between the marginal emissions rate of charge and discharge," said author Burcin Unel, an energy economist with NYU's Institute for Policy Integrity. "When marginal generation is natural gas most of the time, the difference in emission rates is minimal."

The study tests what sort of emissions reductions would ensue if the Clean Peak applied to a battery plant of 20 megawatts/80 megawatt-hours, running at a round-trip efficiency of 85 percent. The analysis does not look at other peak capacity sources that qualify for Clean Peak credits; it's evaluating the implications for energy storage technology specifically.

The authors pulled operational data from ISO New England for March 2018 through February 2019 and simulated the battery plant maximizing revenue in the wholesale market and in the scenario where Clean Peak credits are available. The actual demand peaks, real-time pricing and real-time emissions are known, so the study can determine what would have happened if the battery responded to Clean Peak price signals during that time.

"We analyze this policy and show, contrary to policymaker claims, that it is largely ineffective at reducing emissions, both in absolute terms and in comparison to alternative policies like carbon taxes," the authors write.

The Clean Peak achieves a 5 percent reduction in emissions compared to baseline storage operations with no Clean Peak applied, the study finds. Crucially, that is still progress — the simulated battery released less carbon with the Clean Peak guiding operations.

It is, however, a modest improvement, equivalent to the effects of a $1 carbon tax. A $50 per ton carbon tax, pegged to the social cost of carbon, would cut storage plant emissions 65 percent relative to baseline, the study found. 

The authors see a crucial role for energy storage in the evolution toward a cleaner grid. But they caution policymakers to be careful about how they frame storage incentives, to ensure they don't inadvertently increase emissions or miss out on potential emissions reductions. 

Let's get marginal

The reasons for the simulated disappointment have a lot to do with conceptual differences between how a clean peak policy envisions the grid operating and how power plants actually behave in competitive wholesale markets.

Massachusetts' Clean Peak scheme attempts a form of arbitrage, from off-peak power to on-peak. The off-peak power is meant to stand in for lower-emission power so that releasing it during peak hours displaces higher emission power.

But the researchers question the framework of average carbon emissions to assess the carbon impacts of battery-charging. The economist's approach here is to look instead at the marginal rate of operating emissions: If a battery plant operator decides to charge up the device, what's the next power plant that gets dispatched by the market operator's least-cost algorithm to feed that additional load?

Average marginal operating emissions rates do spike during the peak hours rewarded by Clean Peak credits, but for much of the year, the differential compared to off-peak emissions is relatively flat. (Image credit: Schrader et al.)

If a gas plant fires up to charge the battery, which then loses 15 percent of the power in roundtrip conversion, and the battery later dispatches during peak hours to offset another gas plant, the storage activity actually increases carbon emissions. Based on the marginal dispatch trends in ISO New England in 2018 and 2019, that's often what would happen.

There's also a problem with conflating peak demand and peak marginal emissions, which are actually two separate metrics.

"If you want to reduce emissions, the target metric has to be emissions," said author Christy Lewis, a senior analyst at WattTime, which tracks grid emissions in real time. "That’s what California figured out in its [Self-Generation Incentive Program]."

The SGIP, which supports energy storage installations in California, found that the batteries were increasing grid emissions with their operations; market price signals did not guide them to low-emission charging. That was a problem, seeing as California has built up storage as a linchpin of its decarbonization strategy. After considerable study, regulators eventually introduced an emission-based signal, so that battery operators can ensure they reduce grid emissions through their actions.

Peak demand and peak emissions in Massachusetts happen to line up pretty well. The problem is that the power system in 2018-2019 lacked a strong differential in carbon emission between off-peak periods and on-peak (except in winter, when emissions spiked during peak hours). That made it harder for a battery to displace high emission power with low emission power.

What the critique misses

The marginal emissions lens is crucial to understanding the impact of clean peak standard policies, and states adopting such rules really should think through the issue of whether clear enough differentials exist between high-emission and low-emission periods on the grid.

This study should not lead to the end of clean peak policies as we know them, however.

First off, let's address the argument that it's not a good policy because a carbon price would be theoretically more ideal. That's akin to saying we shouldn't bother encouraging carpooling on the L.A. freeway because dense, walkable urban design would be better for reducing congestion. Yes, but why hasn't that happened yet?

The fact of the matter is, carbon pricing would elegantly internalize carbon emissions into the market price of electricity production, sending a delightfully direct price signal to batteries to charge on cleaner power and avoid dirtier power. But economic elegance does not connote political feasibility, and the Clean Peak holds an advantage that the study glosses over: It became law in Massachusetts, and a robust carbon price has not.

"If we can get momentum behind a carbon tax, then I don’t think there’s any reason not to do that," said Ed Burgess, who holds down the clean peak policy fort at Strategen these days (Huber has since become vice president of rate design at Duke Energy). "But history has shown that a lot of the progress that’s been made is a result of these complementary policies, and I think we still need those to accelerate the market for clean energy technologies."

The other crucial caveat is the timeframe chosen for the study.

Massachusetts already had a decently clean grid in 2018 and 2019, but it may be a less persuasive test case for the emissions impact of the Clean Peak policy. Places such as Arizona and California already have an ample supply of solar power getting discarded on a regular basis because it has nowhere to go. Arbitraging between solar that would be thrown away through curtailment and fossil-fueled peakers would yield a much clearer emissions benefit.

The future will look very different, however, because Massachusetts has embarked on a gigawatt-level expansion of solar and offshore wind. As renewable penetrations increase and gusts of heavy offshore wind blow through, the makeup of marginal power producers will change, likely for the cleaner.

"It’s a question of timing. When and where are we going to hit those tipping points?" Burgess said.

Lastly, it's a mistake to reduce the benefits of a clean peak standard to the single metric of short-term emissions from a particular storage plant. 

"To the extent that you can use things like distributed resources or storage to reduce peak demand, it can be a benefit to lowering overall grid costs," Burgess said.

Massachusetts Gov. Charlie Baker explicitly argued for these system cost benefits when he proposed the state's Clean Peak policy in 2018, at which point the top 10 percent of peak hours drove 40 percent of grid costs for customers. Knocking the policy for its ability to reduce emissions from one of the resource types it applies to fails to account for the full range of policy goals.

For a clean peak standard to merit its name, it does need to reduce emissions in the long run. The goal is to ramp up clean capacity enough to drive the dirtiest peaker plants out of operations. If the battery emits more carbon in its first few years but helps retire the dirtiest peakers from the dispatch queue, it could succeed overall even if it looks shabby at first.

The study raises vital questions about how the Clean Peak policy's energy arbitrage works in practice — questions that Massachusetts should absolutely have an answer for as it evaluates implementation. But the limits of the study period and its narrow framing of policy goals prevent it from dealing a fatal blow to the concept.