The changing composition of electricity supply – and how wholesale energy and capacity markets should adapt to these changes – has been on the minds of those involved in the energy policy arena. Concerns over bulk power system resilience, in particular, are permeating these discussions. Reductions in natural gas prices, as well as increased deployment of wind and solar, are helping reduce wholesale prices of electric energy and capacity. These, in turn, are driving retirements of existing coal and nuclear plants – the major sources of electricity supply for decades in much of the country. Lower natural gas prices are also affecting the financial health of generating companies, as suggested by the recent bankruptcy filing by FirstEnergy Solutions.

In response to these changes, as well as potential threats to bulk electric system reliability (including extreme weather events), FERC initiated a rulemaking on Grid Resilience in Regional Transmission Organizations and Independent System Operators in January. The agency sought input from RTOs/ISOs and other stakeholders on how to define resilience, what the biggest threats to resilience are, and how to address those threats.

The National Energy Technology Laboratory (NETL) recently released a study on those subjects. The study focuses on the “Bomb Cyclone” storm that occurred earlier this winter, which led to a surge in demand for electricity and natural gas for home heating and electricity supply. The study shows that coal-fired generators increased their generation more so than other generator types in six U.S. RTO/ISO territories during this weather event. Based on this increase, the study concludes that coal-fired generators contribute more to resilience than do other types of generators, particularly those fueled by gas.

The study raises some important questions. However, most of the information provided is either incomplete or off-point for a discussion of resilience.

Weaknesses in the NETL analysis of coal

The NETL analysis has two parts. The first part is a short-run analysis of changes in system dispatch by fuel type in response to the Bomb Cyclone with a limited comparison to similar data from the “Polar Vortex” event in 2014. The second part draws on prior studies that show how aging diminishes coal plant performance and raises cost, and uses this information to cast doubt on current EIA projections of coal plant survival and overall coal capacity over the next few decades.

Both parts of the analysis suffer from major weaknesses and a lack of transparency, which makes it difficult to assess key methods and assumptions that form the basis for their conclusions. (For example, the analysis uses different days for different comparisons and does not explain how or why the different days were chosen.)

Nonetheless, we can say with certainty that resilience is not about generation output per se, and thus the primary metric used in this study – change in generation by fuel type during one extreme weather event – is not an appropriate measure. Instead, system reliability and resilience are about the availability of resources to deliver electricity in changing circumstances, such as demand spikes or fuel supply interruptions.

Unlike the assessments of these extreme weather events conducted by the RTOs themselves, the NETL study offers only a cursory discussion of generator availability (the converse of generator outage rates) and does so only for the 2014 Polar Vortex event, not the Bomb Cyclone. Furthermore, it fails to note major market-design changes in ISO-NE and PJM after the Polar Vortex that incented and achieved greater cold-weather fuel assurance by gas-fired generators. It is also important to recognize that availability is not homogeneous within fuel types, as a fuel-focused analysis might suggest, but instead can vary importantly with plant location, condition and maintenance measures, weatherization status, fuel arrangements and other factors.

Rather than dig into indicators of availability, such as forced outage rates and maximum economic offers, the NETL study focuses on generation output. As such, the increases in coal-fired generation that NETL identifies during the storm are not a signal of disproportionate contribution to system resilience, but rather are an outcome of power generation economics and how energy markets favor the lowest-cost resources available to meet demand.

As NETL itself points out, large increases in natural gas demand for a combination of home heating and electricity production caused natural gas prices to spike during the Bomb Cyclone. This spike, in turn, contributed to a price inversion in short-term electricity markets, making coal- and oil-fired generation temporarily more economic than gas-fired generation. At the same time, electricity demand surged, which further contributed to increased dispatch of both coal-fired and oil-fired generation – the latter of which occurred largely at dual-fueled facilities.

After the first part of the NETL report concludes that coal is a cold-weather resilient fuel source, the second part raises questions about the projections of future coal-fired capacity published by the Energy Information Administration (EIA) in its Annual Energy Outlook. The NETL report asserts that age-related thermal efficiency reductions at coal-fired generators will cause more of them to retire than projected in the Annual Energy Outlook. Though this type of analysis is worthwhile, the report does not provide sufficient explanation to assess the methods and assumptions upon which the study relied.

The report’s analysis of the future evolution of aggregate generation capacity, and its comparison with EIA projections, raise the specter of inadequate new capacity to replace retirements. Any projection of future capacity must account for the mechanisms determining resource entry and exit, which neither part of the report acknowledges.

The three regions the report examines most heavily all use capacity markets, where resources only retire if the system no longer needs as much capacity or if more competitive new capacity drives existing plants out of the market. Thus, the basis for the NETL claim of a capacity gap and associated negative reserve margins is inconsistent with market operations.

Recent history provides some insights into the market causes of and responses to coal retirements. Since 2011, over 20 GW of coal capacity has retired in PJM. If one were to go back eight years and accurately predict coal retirements without accounting for new capacity entry, he or she would have falsely concluded that PJM would have a capacity shortfall today. However, such a shortfall has not occurred because new generators have been built. In fact, PJM’s latest capacity auction achieved a reserve margin nearly 7 percent above its target, which has contributed to low capacity prices.

Current markets will likely need to evolve to accommodate expected changes in the composition of the generation fleet. With greater introduction of variable resources, there may be a need for dedicated procurement of additional essential reliability services, including ramping and voltage support. Ensuring markets provide proper incentives for these services will be important in the future and likely influence the types of generators coming on board.

Weaknesses in the NETL analysis of other fuels

Evaluating whether proper incentives for grid services exist requires an examination of how well markets treat differing resource types as substitutes. The NETL study draws incorrect conclusions based on incomplete information and faulty reasoning about the ability of other fuels to substitute for coal retirements as reliable sources of electricity supply.

For natural gas, the study asserts that gas supply to the Northeast is fixed by aggregate pipeline throughput. Yet although the region has faced extensive obstacles to new pipeline development, often a quicker and more economical alternative is to add capacity to existing pipelines. Such options include adding or upgrading compressor stations and “looping” with parallel pipes, all of which could increase peak gas delivery to the region.

The study also asserts that firm contracts for gas pipeline capacity are necessary for gas units to be reliable/resilient sources of electricity supply. For some gas units, however, firm pipeline service is not worth its cost. Generators on interruptible pipeline contracts rarely face fuel curtailments on pipelines without chronic winter congestion. Thus, in many areas outside the constrained Northeast corridors and a subset of eastern PJM, the benefit of firm pipeline service does not outweigh the cost, as the pipeline system has sufficient slack.

Even for pipelines with chronic congestion, firm capacity contracts are often expensive and a poor fit for short duration pipeline constraints. Various physical and financial options are available to firm fuel supplies for natural gas generators without contracting for firm pipeline capacity, and the approach that makes the most economic and practical sense can vary by situation.

For example, dual fuel capability is an option for some generators, while others may elect to invest in natural gas storage or contract for liquefied natural gas injections behind a pipeline constraint. Another alternative is to purchase a physical call option from gas pipeline marketers. A great case study for this option is PJM, where generators commonly used marketers as the most economical route to firm gas supplies following stronger performance incentives enacted in PJM’s capacity market.

The study also mischaracterizes the potential for renewables to replace retiring coal as a reliable energy source. As a byproduct of its focus on generation output alone, the study suggests that wind generation introduces a resilience penalty because its highest hourly generation in the period of Dec. 27 to Jan. 2 was higher than its generation in the highest-load hour of the Bomb Cyclone event.

The study also contends that to substitute for dispatchable generation, wind and solar generators must be backed up by storage or gas with firm pipeline contracts. Yet in fact, many forms of wind and solar generation are dispatchable – such as MISO’s Dispatchable Intermittent Resources – and do not require full back-up generation, which can be provided by resources other than storage or gas plants with firm pipeline contracts.

If there is excess generation capacity, as in most of the United States, no backup for wind and solar is needed until these sources reach a high enough share of generation capacity. Beyond that share, wind and solar can substitute for fuel-burning generators, just not at a 1-to-1 ratio. The relevant consideration is the availability of generators at times of system scarcity.

In domestic capacity markets, resources typically receive credit on the basis of their expected availability during summer peak demand. Thermal resources commonly receive 85-98 percent credit of their nameplate capacity to account for their expected forced outages.

The variable nature of wind and solar greatly reduces their capacity-value relative to conventional plants. For example, in the recent PJM capacity auction for 2020/2021, wind received credit for approximately 13 percent of its installed capacity and solar for approximately 38 percent, with the percentages varying from installation to installation. As a result, PJM counts on wind and solar together for only about 1500 MW of power at times of scarcity, whereas wind supplied 3100 MW of power during the highest-load hour of the Bomb Cyclone.

During some other scarcity events, wind has supplied less than its capacity credit. This variation highlights the need to potentially revise methods that determine capacity credits based on statistical probability of very low region-wide resource availability at times of extremely high power-scarcity.

Both wind and solar availability differ across seasons. The availability of gas plants may also vary between winter and summer based on pipeline conditions and fuel arrangements, but this is very context-specific. Accounting for these differences will be important in improving the calculation of capacity credits, especially as winter scarcity becomes more important. Current annual constructs do not typically recognize these seasonal differences, but this may soon change, as FERC recently ordered a technical conference to evaluate proposed moves to seasonally-differentiated capacity crediting in the context of PJM’s capacity markets.

Implications for markets going forward

An overarching challenge for any evaluation of power system resilience is the lack of agreement on a definition and on appropriate metrics. Helping move the community toward an agreed-upon definition is part of what FERC is trying to do with its current rulemaking on resilience, but these conversations are arguably in their early stages.

This lack of consensus does not mean that we must wait for consensus to emerge to do analysis, but any analysis needs to be transparent about methods, assumptions and data sources and employ appropriate market performance metrics in order to contribute to the evolving discussion of electric supply resilience in a useful and durable way. The NETL study is lacking in several of these respects. Moreover, the study fails to analyze market incentives, which provide the basis for evaluating prospective market conditions and informing market design changes.

Markets can facilitate rapid platform change when supply-and-demand fundamentals shift abruptly. This flexibility encourages innovation and accelerates deployment of low-cost entrants. Yet such turnover in the generation fleet can expose market design flaws. This possibility calls for thorough analysis of market performance, especially by examining the suitability of the incentive structure as the generation fleet evolves.

Determining past and projecting future market performance begins with an analysis of whether market design provides proper incentives for the efficient, reliable and resilient operation of the electric system. Ensuring such incentive compatibility among market participants is necessary for investment and operating decisions that maximize the efficiency of generator investment, operation and retirement.

Some fuel-specific analyses, like the NETL study, have made hypothetical projections of the composition and performance of the future generation mix irrespective of the underlying incentive structure that will drive actual behavior. This approach not only leads to false conclusions, it also fuels consideration of subsidizing uneconomic generators of particular types to keep them from retiring.

Well-designed, fuel-neutral markets provide an efficient method of procuring services for a reliable and resilient grid. Healthy markets provide the right incentives for power suppliers to determine the means of providing these services. In contrast, prescribing the means to provide such services would be less efficient.

A hallmark of electricity market design is to define products around discrete reliability services. If the set of product markets is not sufficient to compensate for the full suite of reliability services, then that set of markets may undervalue any resource that provides the inadequately rewarded reliability services, not just coal-fired generators. NERC has identified potential shortfalls of essential reliability services like ramp and primary frequency response, which may justify dedicated market mechanisms to procure them that FERC and the RTOs have already begun acting upon. Going forward, it is critical that the conversation on grid reliability and resilience evaluates markets with appropriate economic techniques and pursues market design enhancements that comport with incentive compatibility.