We Need to Change Our Assumptions About Gas Plant Reliability

Gas power plants have a problem. And it’s a problem that affects all of us.

In extreme weather, when electricity demand is at its highest and the grid needs gas plants the most, gas plants have been failing at alarming rates. In the worst instance, widespread gas plant failures led to rolling blackouts that impacted millions of people for days on end. So the problem of gas plants buckling under extreme weather has real-world, and in some cases life-threatening, consequences.

Unfortunately, this isn’t a rare problem. Gas plants failed at a scale that jeopardized grid reliability for large regions of the United States during severe winter storms in 2011, 2014, 2018, 2021, and 2022. And earlier this year, gas plants failed at an alarming rate during Winter Storms Heather and Gerri. (Although the country’s federal energy regulator has had a disconcertingly nonchalant attitude towards these events.)

This all begs the question: what should we do about these gas plant failures and the resulting grid reliability problems?

The most promising and comprehensive solution is to meet grid reliability needs with clean resources rather than gas plants. Study after study after study has shown that a geographically diverse mix of clean energy solutions (including solar, wind, energy storage, and transmission) can go a long way towards maintaining grid reliability. However, that’s not what I’m going to focus on here.

Instead, I’m going to dig into one particular solution that could reduce the risk of grid reliability problems: revamping gas plant capacity accreditation. Even with the clean energy transition well underway, gas plants will be around for a while as we phase out fossil fuels. In the meantime, the grid will be counting on gas plants to perform as expected to help keep the lights on. And it’s critical to make sure we’re not overcounting the reliability contributions of gas plants during the transition.

The main point of this blog is to explore the potential consequences of reducing the grid reliability contributions of gas plants to better reflect reality. (Spoiler: A UCS modeling exercise suggested that reducing the reliability contributions of gas plants could modestly increase the need for energy storage and gas capacity. But in the long-term, the jury is still out, and clean energy technologies such as long-duration energy storage could play a major role in meeting grid reliability needs if we accredit gas capacity more accurately.)

Giving gas plants too much credit

There is absolutely no doubt that, when it comes to grid reliability, gas plants have been getting more credit than they deserve.

Most grid operators have programs in place to ensure there’s enough electricity supply to meet demand, often referred to as resource adequacy. As part of these grid reliability programs, grid operators must determine how much “credit” resources get for contributing to resource adequacy. Without going into too many details, it turns out this is a non-trivial task. Grid operators have largely turned towards a methodology called effective load carrying capability (ELCC) to determine the reliability contributions of renewables, and there have been more recent efforts to apply this methodology to energy storage as well.

But it’s a different story for gas plants. Until very recently, grid operators have largely been using outdated, overly-generous methods to assign capacity credit to gas plants. For example, while many grid operators have been assigning capacity credits of more than 90 percent of the nameplate capacity of gas plants (when using ICAP or UCAP), a more rigorous study of the PJM South system found that the real capacity contributions are as low as 76% in the winter and 85% in the summer. And the historical impacts of extreme winter weather illustrate this point even more clearly. For example, during Winter Storm Elliott, forced outages in PJM peaked at 7:00 AM on December 24, with 39 percent of the gas fleet knocked offline during that hour.

Some grid operators are starting to recognize this disconnect and take steps to assign capacity credit to gas plants more accurately. For example, PJM recently began applying ELCC to all resource types for the purposes of capacity accreditation, with gas plants assigned ELCC values less than 80%. ISO-NE has also been reevaluating its capacity accreditation methodologies, and big changes to gas plant accreditation are likely to come, especially since New England has gas system constraints that can limit gas plant electricity generation in extreme cold weather.

Analysis and historical examples all clearly point towards the conclusion that grid operators have been overcounting the grid reliability contributions of gas plants. And a handful of grid operators have been taking steps to fix that. But what will happen on a larger scale when these changes happen across the country?

Gas plants have historically gotten more credit than they deserve for ensuring grid reliability, but some grid operators are starting to fix that. Chehalis Power Plant, a natural gas power plant in Chehalis, Washington. Steven Baltakatei Sandoval/Wikimedia

Modeling the effects of reduced gas plant capacity accreditation

To better understand the consequences of decreasing the capacity value of gas power plants, UCS turned to one of our favorite tools: grid modeling. With the help of our colleagues at Evolved Energy Research, we built off of the modeling conducted for our Accelerating Clean Energy Ambition study. That study found that the US can meet its climate goals of cutting economywide emissions in half by 2030 and achieving net zero emissions by 2050, while delivering significant public health and economic benefits. We used the exact same assumptions as in our Net Zero Pathway, which represents a least-cost mix of technologies and resources for meeting US climate targets.

But for this analysis, there was one key difference: we decreased the capacity credit for coal and gas plants from 95% to 80%. The generic application of 80% across the board certainly isn’t a perfect approach, but without having geographic- and technology-specific reliability contributions, we made this change as a proxy for more realistic capacity contributions from fossil fueled resources. And while our focus was on gas plants, we applied the same 80% capacity credit to coal plants since they have also experienced high outage rates and underperformance during extreme weather events.

I should also mention that one key modeling assumption we did not change is the total reliability requirement; that is, despite the fact we reduced the reliability contributions of coal and gas plants, we did not reduce the total amount of accredited capacity required to meet reliability requirements. That means that this modeling scenario would result in a more reliable grid than our Net Zero Pathway scenario because more resources are required to meet the reliability requirement in lieu of coal and gas. One could make the argument that the planning reserve margin already accounts for coal and gas plant reliability issues; so if one decreases their accredited capacity, then the planning reserve margin, along with the total reliability requirement, should also decrease in tandem (maintaining the same level of grid reliability). While that’s an argument worth exploring, we chose to keep things simple for this modeling exercise.

Before we even did this modeling, I had a hunch what the results would show. I knew intuitively that, if you decrease the grid reliability contributions of coal and gas plants and keep the reliability requirement static, the model is going to need something else to meet the reliability requirement.

So the question was essentially: what additional resources would the model pick to maintain grid reliability?

Filling the grid reliability role of gas plants

When we reduced the capacity credit for coal and gas plants, the model showed an increased need for two types of resources: energy storage and gas plants.

In the 2030 timeframe, energy storage resources (both short-duration lithium-ion batteries and long duration storage) provide most of the additional capacity. By 2050, most of the incremental capacity comes from gas plants, with long-duration storage providing the remainder. However, it’s important to put these results in the context of the larger grid buildout needed to reduce global warming emissions and meet increasing electricity demand. The roughly 80 GW of additional capacity required in this modeling scenario is a relatively modest 15% of the nearly 540 GW of gas capacity that remains on the grid nationwide in 2050, and it’s merely 2% of the nearly 4,000 GW buildout of new resources required by 2050. And although the model selects some additional gas capacity to meet reliability requirements, those gas plants contribute less than 2% of total electricity generation by 2050, so they’re being used very little.

To meet grid reliability requirements with lower capacity credits for coal and gas plants, our grid modeling tools showed an increased need for energy storage and gas plant capacity.

Nonetheless, there’s a way in which these results are counterintuitive. As one of my colleagues put it, “When you make gas plants less reliable, the model decides you need… even more gas?” Indeed, when the grid model needs to pick additional capacity to meet the reliability requirement, it still considers gas plants one of the most cost-effective solutions, even when their reliability contributions aren’t quite as high (and when those plants barely ever generate electricity).

These modeling results point to the larger challenge of transitioning all the way to 100% clean electricity and completely phasing out fossil fuel electricity generation. But I also think these results offer a glimmer of hope in that they demonstrate that long-duration storage, along with high levels of clean energy, can play a role in ensuring grid reliability and reducing the need for gas capacity.

For example, a recent study of long-duration storage in California showed that, with the most optimistic cost projections for long-duration storage, the technology could cost-effectively replace all of the state’s gas capacity by 2045. In addition, clean energy technologies that provide power around the clock, such as geothermal, are poised to make a comeback (after decades of dwindling investments) due to recent advances in the technology. And California in particular is gearing up to make substantial investments in both long-duration storage and geothermal, in addition to kickstarting its offshore wind industry. These supply-side, clean energy solutions will be critical for maintaining grid reliability in lieu of gas capacity.

And it doesn’t end there. Demand-side solutions that reduce or shift electricity demand can also enhance grid reliability and mitigate the need for gas capacity. These solutions include energy efficiency measures, distributed energy resources (e.g., rooftop solar and home energy storage), demand response programs (e.g., those that cycle air conditioners off and on), and more. One of the demand-side solutions that I’m most excited about is vehicle-grid integration with electric vehicles. If we could harness all the batteries in EVs, there’s tremendous potential for EVs to support grid reliability and reduce reliance on gas plants. And in recent news, California’s governor just signed a bill that gives a state agency the power to require that new EVs have bidirectional charging capabilities, which could help ensure EVs are capable of discharging energy back to the grid to ease grid strain during periods of peak electricity demand. (Much more on that topic in the year to come!)

Putting an end to gas plant overcounting

In the short term, putting an end to excessive capacity accreditation for gas plants would likely result in additional energy storage and gas capacity. In the long term, our modeling indicates that additional gas capacity plays the main role, with long-duration storage providing most of the rest.  But modeling decades into the future is an inherently uncertain endeavor, and technological breakthroughs can result in dramatic changes to the least-cost portfolio of resources identified in modeling results. And with so many promising clean energy technologies breaking onto the scene, the jury is still out on whether gas plants will be the most cost-effective resource for filling that reliability role in the long term.

Going forward, utilities and grid operators must reassess the reliability contributions of gas plants to ensure they are not being overvalued. It is possible this reassessment could lead to some unintended consequences in the near term (in the form of increased gas plant retention). But in the long term, continuing to overvalue gas plant reliability contributions means people will continue overpaying for gas plants that are underperforming, and it will only make it more difficult for clean energy technologies to beat out fossil fuels.