Reliability Wars


Power System Planning: Who gets paid (and how much) for backing up the system?

Fortnightly Magazine - June 2005

Ed Krapels—the electric industry consultant from Boston who helped dream up the initial idea of a monster, undersea direct-current cable (the Neptune project) to bring cheap Canadian power south to the Eastern Seaboard of the United States—thinks he knows now why the merchant transmission business is in the toilet.

“Independent transmission projects,” he says, “need to be paid for their services to the capacity market.”

Krapels believes that private-grid developers need the same incentives that regulators in the Northeast have OK’d for power plants. Some plans would pay financial transmission rights (FTRs) in exchange for participant funding by grid investors, but Krapels says that’s not enough.

“Confining transmission projects to FTR payments,” he explains, “is like confining generators to energy-only payments.”

These words—from a press release that Krapels circulated by e-mail in early May through his firm, Energy Security Analysis Inc. (ESAI), of Wakefield, Massachusetts—speak volumes on what’s happening in today’s power industry, and on what the independent system operators (ISOs) and regional transmission organizations (RTOs) are trying to achieve, not only for merchant-grid projects but for merchant generation and system reliability.

Figure 1 - New York’s ICAP Update

But first, some background.

In recent installments of this column, we have reported on moves at the RTOs to create incentives for building new power plants and to ensure adequate long-term supplies of electric generating capacity. In large part, they have chosen to create spot markets for capacity rights. In short, the RTOs decide what utilities ought to be willing to pay to acquire rights to installed capacity (hence the term). Then they force the utilities to buy it. The price, however, is pure fiction. The RTOs set the price by reference to the total cost profile for a typical but hypothetical single-cycle gas combustion turbine power plant—e.g., fixed and operating costs, interest rates and financing, fuel, labor, maintenance, heat rate and efficiency, site preparation, insurance, property taxes, and all the rest. Add in an estimate of revenues (likely sales of energy and ancillary services, based on projected prices), and you have a tentative figure for what utilities ought to be willing to pay for an incremental kilowatt.

Some ICAP plans take account of local transmission constraints and differing degrees of deliverability to individual nodes on the grid, and so are known as locational capacity markets, or LICAP. They employ complicated “demand curves” to plot a market price for capacity that varies over time as a function of regional supply needs (see, “A Kink in the Curve: LICAP and Its Lessons,” January 2005, p. 16). 

These LICAP regimes apply in regions where utilities have divested themselves of generation and have ceded control of transmission lines to the RTOs. Importantly, they allow merchant generators to lay claim to two independent revenue streams: (1) energy sales revenues, dependent on current prices in spot markets or bilateral contracts; and (2) the contributions to electric reliability, as measured by capacity value and paid through ICAP programs at the RTOs.

Figure 2 - California’s RAR Plan

Now let’s return to Krapels and why merchant-grid projects are foundering.

“If an AC transmission project could reduce New York City’s locational capacity requirement by 10 percent,” he notes, “the annual savings could be as high as $25 million.”

But that’s not the way things work today. In the typical case, a merchant transmission line (an “economic” grid addition, not strictly needed for reliability) would not get paid for its capacity value under current RTO regimes for transmission planning and expansion.

In fact, the Krapels theory runs clearly afoul of a basic tenet of rate making at the Federal Energy Regulatory Commission. FERC insists that transmission projects qualify only for “and/or” pricing. In other words, transmission can earn a cost-based return on investment as a compensation for contribution to the public service, or it can earn a competitive profit on the opportunity cost of power, but not both.

Thus, under PJM’s RTEP plan for regional transmission expansion (see Commission Watch, June 15, 2003, p. 13) or the New York ISO’s CRP process for comprehensive reliability planning (approved at FERC late last year) merchant transmission doesn’t receive a public-sector payment. Instead, any grid expansion need that cannot be solved by a purely market-based solution is thrown back into the realm of central planning. The planned solution (as devised by RTO committees, utility planners, or state regulators) then collects a guaranteed return under RTO tariffs or a commission-regulated rate.

That’s why, as Krapels sees it, today’s large-scale AC transmission projects are done only by traditional load-serving utilities and financed by traditional utility revenue streams that compensate utilities for creating system-wide benefits—the very thing that is denied to merchant grid developers.

Maybe Krapels has hit on something. In short, he implies that all system assets, whether generation, transmission, or other—and however financed and owned—should earn some revenues based on economics (the competitive market value of the energy product), and some revenues on contribution to reliability (capacity and system benefits).

FERC and the RTOs largely agree with Krapels on the generation side, where the ICAP plans have gained a grudging acceptance. Yet a big problem remains in converting theory to practice. The projected costs of paying these capacity credits to power producers has climbed to astronomical levels, in part because a key element is missing: consumer rights.

Unlike the traditional rate case or resource planning docket at the state PUC, the RTO ICAP process carries with it no explicit internal requirement that the final number must be “just and reasonable” (J&R), as is required of PUCs in setting retail electric rates. FERC can review the RTO findings, but cross-examination of witnesses and other features are missing from the RTO process, which looks very much like a virtual rate case but without due-process guarantees that are second-nature at the PUC.

In an interesting twist, some opponents of the New York ISO’s CRP proposal had complained about too little of a federal role in the regional planning process. They had faulted the ISO's plan to delegate reliability authority to the state PUC if no market solution could be found to solve a problem with grid operations (congestion, constraints, load pockets, etc.), calling it “unlawful.” Even the PJM RTO had said it was concerned that the state PUC’s expansive role in the planning process could jeopardize the independence of the ISO.

Yet, in its order approving the plan, FERC itself agreed to take a backseat, and saw the plan as an opportunity to “dovetail our regulation” in order to work closely with state officials.

As FERC explained, the footprint of the New York ISO is contiguous with the state of New York, and so the state PUC was “singularly suited” to resolving disputes concerning the ISO’s regional needs assessment and final determinations in the CRP process. (See Docket No. ER04-1144, 109 FERC ¶61,372, Dec. 28, 2004; rehearing granted Feb. 25, 2005).

Moreover, it appears that FERC does not enjoy as much authority over reliability as some might think. Thus, to fall back on the work of the state PUC might actually make sense.

Nearly two years ago, testifying before a U.S. Senate Subcommittee, FERC Chairman Pat Wood III stated: “The explicit authorities granted to the commission in the area of reliability are very limited.”

In fact, according to the New York ISO, a key facet of responsibility for reliability of the bulk power system should reside, and has always resided, with state regulators. FERC, says the ISO, can claim explicit statutory authority over reliability only to: (1) order interstate service upon complaint by a state PUC; (2) ascertain whether certain interstate interconnections are physically reliable; or (3) request reports or studies by reliability councils or other appropriate agencies. (FERC Docket No. ER04-1144, answer filed Sept. 28, 2004.)

Thus it is not everywhere that FERC and the RTOs hold sway. In California, for instance, the ISO (Cal-ISO) has not adopted a market-based ICAP program, nor does it intend to. Instead, it has imposed a must-offer obligation (MOO) requirement to force owners of electric generating capacity to make it available for sale in real time, and has decided on a strategic basis to rely heavily on state PUC programs to achieve regional reliability.

By turning away from an ICAP regime and embracing a planned solution managed by the PUC, the California ISO would return the reliability process back to the state. Once there, regulators would fix reserve requirements and their attendant costs under the standard of just and reasonable rates. 

The jury was still out at this writing on how well the California compromise would work. The California PUC is set to issue a final order by the end of June in its rulemaking on resource adequacy. That timeline, in turn, has prompted Cal-ISO, hard at work in its market redesign, to wait a bit before proposing its next iteration of tariff language on how its new integrated forward market (IFM) will work. But we do know this: Cal-ISO’s comments on the PUC plan, submitted through the winter and early spring, reveal serious concerns over temporal and locational factors. To translate, the PUC might find it a simple matter to set an overall planning reserve margin for each utility. But what mechanisms would ensure deliverability—that reserves would be available at the hours required and at all constrained locations? (See, CPUC Docket No. R.04-04-003, filed April 1, 2004. See also comments of ISO, posted at

Nevertheless, by shifting oversight on reliability from the grid operator to the PUC, the California plan would achieve the one thing that FERC and NERC (the North American Electric Reliability Council) have been seeking for so many years but without success—a reliability standard backed up by force of law.

New York: A Rate Case?

In regions that have adopted ICAP plans, the degree of complexity of analysis of generating costs can take your breath away. In the New York case, the full detail is outlined in a study presented in August of last year by the Boston consulting firm of Levitan & Associates. Consider, for example, only a very small part of the analysis.

The Levitan Study (the ISO’s main case) would assume construction in Long Island or New York City of two, simple-cycle LM6000 Sprint aeroderivative gas turbine units, with a 96-MW capacity rating and heat rate of 9,739 Btu/kWh (at an ambient temperature of 59 degrees F). But for the rest of the state, either the 2xLM6000 or a pair of industrial frame 7FA turbines would be chosen. That’s because, due to economies of scale, the 2xLM6000 plant would be 17 percent more expensive than the 2x7FA plant if operated in the rest-of-state zone. However, the LM6000 unit would be better suited to simple-cycle operation, as it can achieve full load operation in 10 minutes (allowing for higher emissions prior to operation of the selective catalytic reduction system). By contrast, the 7FA turbine would take longer to achieve full load and would be more commonly used in combined-cycle operation. In either case, financing would incorporate assumptions of a 3 percent inflation rate, plus 5 percent interest for construction debt, 7.5 percent over 20 years for the permanent debt term (the plant’s useful life), a debt/equity ratio of 50/50, and an after-tax equity rate of return of 12.5 percent.

This level of detail gives power-plant developers lots of targets to shoot at in taking aim at the final ICAP demand curve and set of prices. And so, in the New York case, the ISO had complained that the developers, by their detailed protests over specific cost and operating characteristics of particular plant models, were “trying to turn this proceeding into a gas turbine rate case.”

The ISO said the plant developers were attempting to steer the ISO ICAP process “in accordance with rate-making principles” to produce rates that would satisfy a traditional “just and reasonable” test. But the ISO bristled at this notion. “ICAP payments,” it argued, “are not statutory entitlements akin to classic just and reasonable rates.”

To defend its position, the ISO had cited the case of Sithe New England Holdings, LLC v. FERC, 308 F.3d 71 (1st Cir. 2002), which puts down the notion that an RTO ICAP hearing is a virtual gen plant rate case:

“[The] ICAP charge is not of this ilk. Rather, it is a payment to suppliers over and above the amount they charge for power. …

“If ICAP charges were abolished by FERC tomorrow, the sellers could object that FERC was behaving unreasonably … but not that they were deprived of a just and reasonable rate.”

As the ISO saw it, the developers’ claim of entitlement to a just and reasonable ICAP rate was nothing but an “abused slogan.”

Interestingly enough, FERC made no comment at all about the “rate case issue,” or any entitlement to just and reasonable rates, in its final order approving nearly all the ISO’s proposed elements for various sets of ICAP demand curves. (See FERC Docket No. ER05-428, 111 FERC ¶61,117, April 21, 2005. For more detail, see Figure 1 and explanatory notes.)

New England: High Costs

The lack of state PUC participation and J&R rate protection stands out clearly in the case to set minimum requirements for installed capacity for utilities and load-serving entities (LSEs) in New England for the 2005-2006 power year, which was still pending before FERC in early May. This requirement, known in New England as OC (Objective Capability), incorporates a 12 percent reserve margin above the nominal load requirement. But while that determination would appear relatively straightforward at first glance, the case in New England has brought a stern admonishment from utilities, state PUCs, and attorneys general from across the New England region. They claim the ISO has ignored stakeholder and consumer interests and has caved to political pressure within the NEPOOL participants’ and reliability committees. And worse than that, they say the ISO has misunderstood its role and has usurped jurisdictional authority to set targets for resource adequacy that resides with the state PUCs. (See FERC Docket No. ER05-175, filed March 21, 2005, comments and protests filed through Apr. 25, 2005.)

To set the capacity requirement, the New England ISO (ISO-NE) analyzes three key elements: (1) load forecasts; (2) unit availability (e.g., how to estimate scope and likelihood of future plant outages); and (3) tie benefits (the projected ability to import power across the grid from other regions, including Canada). It is the second two factors, availability and tie benefits, that have come under fire.

One bone of contention concerns unit availability. The ISO has proposed to abandon the EFOR (Equivalent Forced Outage Rate) model, which calculated random unit failure without regard to whether a plant is running, in favor of EFORD (Equivalent Forced Outage Rate Demand), which focuses on outages that occur when a unit is needed for dispatch.

By all accounts, however, the primary difference between the ISO-NE’s proposed OC target for 2005-06 and that already approved in prior years stems from its move to lower its estimate of tie-benefit capacity from 2,000 MW to 1,800 MW. As lawyer Stephen Teichler explains (representing NSTAR Electric & Gas), that is a very big deal:

“Although a 200-MW difference may sound insubstantial in relation to the over 30,000-MW OC value, [we] estimate that this 200-MW compromise would cost consumers approximately $1.4 billion over the next five years.”

But what was this compromise?

As it turns out, the ISO had put it to a vote. When faced with a range of possible values for tie benefits, the Reliability Committee of NEPOOL (New England Power Pool) had voted 70.65 percent for 1,400 MW, 56.61 percent for 2,000 MW, and 80.12 percent for 1,800 MW. Do the math. Has resource planning become a beauty contest?

To the ISO, all this seems like sour grapes. It counters that a similar vote with similar results was taken for the prior power year of 2004/05, without much controversy. That proves, says the ISO, that current objections center not so much on the ISO’s nominal calculation of the OC target, but more on the much higher dollar impact that will follow if the ISO follows through on its parallel proposal to adopt a still more costly LICAP market.

But the most serious charges come from the Connecticut Consumer Counsel and attorney general, and from PUCs in Vermont, Rhode Island, and New Hampshire (joined by utilities Conn. L&P and Northeast Utilities). They complain, essentially, of a lack of any rate-case-like process to reach a J&R finding:

“Because [the] ISO has ignored the cost of its proposed increase in IC Requirements … it has not provided sufficient information to determine precisely what those added costs will be under any LICAP demand curve being considered.”

Their objections clearly highlight the difference between resource planning at the RTOs versus resource planning at the state PUCs.

California: A Better Way?

PJM, the nation’s most influential RTO market, appears on the verge of joining New England with a locational market for ICAP. The PJM plan, known as the Reliability Pricing Model, originally was scheduled for launch this spring (the proposal, that is) but has since been delayed, at least up until the time of a PJM board meeting that was to have taken place in early May. However, PJM has vetted many details. For example, specific proposed demand curves for three different PJM pricing zones can be found on the PJM Web site. (See, “Whitepaper on Future PJM Capacity Adequacy Construct,” Nov. 2004,

Nevertheless, the real innovation may come from California, where Cal-ISO deliberately has omitted a classic ICAP model from its proposed MRTU (“Market Redesign and Technology Upgrade,” proposed several years ago as the MD02 — Market Design 2002) model. Instead, Cal-ISO has proposed to achieve supply adequacy in two ways.

First, the ISO would offer an availability payment to certain generators under a construct known as RUC (Residual Unit Commitment). RUC is not a true market, however. Instead, after close and settlement of its proposed Day-Ahead auction, the ISO would “reserve” the right to call on certain resources in real-time, should supplies fall short.

Second, Cal-ISO would extend its real-time must-offer obligation (MOO) to the Day-Ahead market as a temporary stopgap, until the California PUC fully implements its proposed rule for establishing a resource adequacy requirement (RAR). In broad terms the RAR can be thought of as a state-mandated equivalent of New England’s OC target. California’s RAR target would impose a planning reserve margin of between 15 to 17 percent above nominal load requirements, to be phased in by utilities over several years.

Space does not permit a full description here of the CPUC plan, let alone the latest version of the Cal-ISO’s MRTU, with its new, bid-based, security-constrained Day-Ahead settlement, and its Hour-Ahead preview of the real-time closing. For an overview of the origins of the Cal-ISO market design and the CPUC initiative, see “Market Design Still Eludes California,” published in Fortnightly’s Spark, March 2004. (Spark is the monthly online newsletter of Public Utilities Fortnightly. Access back issues at, using the user name and password published in each magazine issue, on the table of contents page.)

Nevertheless, for this discussion, at least two key points stand out.

First, the Cal-ISO’s peculiar RUC construct. Though it offers a payment for availability, RUC is not a product market. Rather, RUC is a sort of option—a call on generators that lets Cal-ISO hold the capacity in reserve, at the end of the Day-Ahead closing, as a hedge against a real-time supply deficiency. RUC is not designed as an incentive program to encourage gen plant construction over the long term. It is more like an insurance policy. It is a hedge against the specific risk of a shortage of energy (not capacity) in the short term. As Cal-ISO itself explains, the RUC availability payment “is essentially an up-front reservation payment for an energy service.”

Second, the real capacity element lies with the PUC’s enforceable RAR mandate. With the PUC worrying about the long-term outlook for generating resources, and the huge costs of the attempting to influence long-term investor behavior, the ISO is left free to manage short-term physical operations. (In fact, the Cal-ISO has raised key questions about how the PUC’s RAR standard will not work in practice unless it is designed carefully to account for grid congestion, constraints, and load pockets, which can block delivery of energy or capacity. See Figure 2 and explanatory notes.)

Overall, the California plan represents a bona fide experiment to meld the ISO’s technical expertise with the PUC’s political acumen.

In California at least, someone will be worrying about the cost of ensuring adequate resources, and whether we can afford it. 


New York’s ICAP Update

The Update. In April, with minor modifications, the FERC OK’d the New York ISO’s proposed update to its ICAP plan, adjusting the demand curves that the ISO will use to calculate the monthly market price for electric capacity, for each of the ISO’s three separate locational ICAP zonal markets — (1) New York City, (2) Long Island, and (3) the zone for the rest of the state, known as NYCA (see figure 1).

The ISO had proposed new demand curves for each zone for each of three prospective power years (each year beginning with the summer capability period). This figure shows the old and new demand curves for each of the three zonal markets for the power year 2005-2006. The two succeeding years show demand curves with slightly steeper slopes and slightly higher prices, but those curves are not shown here. (See FERC Docket No. ER05-428, proposal filed Jan. 7, 2005, approved April 21, 2005, 111 FERC ¶61,117.)

Understanding the Figure. The scale on the x-axis assumes a pre-existing margin of generating reserves of 18 percent in excess of load. In other words, the point of 100 on the x-axis denotes a level of capacity equal to 118 percent of the load requirement. That is the equilibrium level that the market is designed to “incentivize,” for all LSEs and for all zones in the ISO. Thus, at the point of x = 100, the curves produce a price for capacity equal to the equilibrium benchmark or reference price. This benchmark price reflects what the ISO believes it ought to cost to acquire new generating capacity, in the form of a single-cycle gas-fired combustion turbine of standard design. Cost is calculated by making certain specific assumptions regarding size, startup times, efficiency, heat rate, emissions restrictions, and all relevant costs for such items as fuel, labor and maintenance, land, site preparation, insurance and financing, property taxes, and so on. The ISO also assumes that the plant will earn revenues from sales of energy and ancillary services in spot markets and bilateral contracts, and credits such revenues against costs to arrive at the benchmark price for the equilibrium point of x = 100.

As shown here, for power year 2005-2006, that benchmark price is $6.78/kilowatt-month for NYCA, $12.52 for Long Island, and $13.70 for New York City.

A key item at issue in the FERC case concerned the point on the graph representing the intercept with the x-axis, known as the “zero crossing point.” That point shows the point at which supplies of electric capacity are considered so plentiful that the price falls to zero. That point, coupled with the reference price at x = 100, defines the slope of the curve and thus the exact price at all other points along the curve.

Some parties in the case argued that the crossing points (such as 112 for NYCA) were set too high and thus produced prices higher than needed at each point along the sloping sections of the curves. Why not 109, for example, which would produce lower prices and a steeper curve for NYCA? Such arguments won a degree of sympathy, but FERC chose to retain the existing crossing points, since it thought that steeper curves would create greater volatility in capacity prices and thus impair incentives for investment. — B.W.R.


California’s RAR Plan

The California PUC’s proposal would imposed a resource adequacy requirement (RAR) that obligates utilities and load-serving entities (LSEs) to maintain enough generating and other resources to maintain a planning reserve margin (PRM) of 15-17 percent in excess of load.

LSEs must acquire a mix of resources to satisfy the number of hours for each month in which its load reaches 90% of its coincident peak load – the time of its maximum contribution to the monthly CAISO system peak. See Cal. PUC Decision 04-10-035, Oct. 28, 2004.

However, this static requirement does necessarily guarantee sufficient delivery capability at particular hours and locations, to overcome local grid congestion and physical and regulatory factors that can limit operations at certain plants (startup ability, emissions restrictions, fuel limitations, etc.).

Thus, lack of a security-constrained, bid-based, locational and temporal market mechanism marks a potential Achilles’ Heel for the California RAR plan in estimating required resources and attendant costs, and differentiates it from New England’s proposed LICAP Plan and PJM’s proposed Reliability Pricing Model.

Nevertheless, the ISO has provided the following rough outline on how RAR units might be allocated to meet real-time obligations.

The Pink Strip. The top (pink) strip (see Figure 2) represents when the system-wide load exceeds 90% of the system-wide peak. The width of this strip represents the number of hours the RAR resources must meet both a capacity and a duration criteria … Given that a supplier with an energy or use-limited resource would normally attempt to maximize its profits by operating the unit only during peak periods when prices are highest, it can be properly assumed that the pink strip will be met through energy and use-limited resources …

The Yellow Strip. The bottom (yellow strip) represents a subset of the system-wide RAR capacity that is needed for local reliability. This subset must be able to produce energy when called upon practically all the time (unless it is on maintenance or forced outage). Depending on the local Resource Adequacy Requirements, this could represent 25% to 35% of the system-wide RAR capacity …

The Blue Strip. The middle (blue) strip represents a subset of the system-wide RAR that is not necessarily needed for local reliability, but is needed to ensure system-wide resource adequacy. The CAISO contemplates that this subset would also be subject to a must-offer obligation to the extent it is not on maintenance or forced outage … [but] … CAISO cannot, and does not intend, to identify the minimum amount of local capacity that would be necessary to respond to all contingencies or possible system conditions …

The arched line (red) shows how the available RA capacity, beyond that procured in the local areas, is effectively reduced because the … resources have legitimate limitations on their availability. … Thus, the available capacity within the middle section of this figure will not be capable of fully covering the defined box. Rather it will diminish with the number of hours beyond those required to meet the top 10% of hours. This principle is reflected by the downward slope of the red line …


In the aggregate, the blue strip should consist of sufficient non-use-limited resources to be capable of meeting system-wide energy needs plus PRM outside the system peak hours. (See, Calif. PUC Rulemaking 04-04-003, Reply comments of California Independent System Operator, Feb. 28, 2005, pp. 6-8,