A new way to measure what matters most: how close a unit comes to meeting its total potential profit.
Approximately 65 percent of capacity additions in the last few years have been gas-fired, combined-cycle units. Recent market conditions have been hard on these new resources, which have suffered from significantly low capacity factors. But such units are popular because of their flexibility. In markets that squeeze generators between high fuel prices and lagging electricity prices, this kind of flexibility is crucial. A capacity factor isn't capable of measuring or valuing that kind of generator attribute for combined-cycle units, or any other type of unit.
A better metric would measure a unit's ability to capture peak prices while minimizing shoulder period and off-peak losses. Furthermore, it would measure the extent to which a unit dispatches according to favorable market conditions. A better metric, from a market-oriented perspective, is what we have described below.
The Economic Dispatch Factor
Operating a unit well means more than keeping it running. Planned outages must be timed to coincide with appropriately low market conditions. Forced outage risk must be reduced and maximum output sustained during periods of high prices. Often this is considered on a calendar basis, whereby shoulder months are used for maintenance so that reliability is highest during the peaking months. These decisions also can be made in real time. In ISO markets, where prices fluctuate hourly, generators that respond correctly to hourly market conditions stand to benefit from the daily peaks and valleys in the price path.
Energy Velocity has formulated a statistic using hourly generation and prices, and monthly marginal operating costs, combining them into a single measure of performance. We call it the Economic Dispatch Factor (EDF). It is similar to the capacity factor in that it is a normalized measure of how near a unit comes to reaching its full potential. However, instead of measuring success in terms of total generating hours, the EDF measures what matters most: profit, or rather, how close a unit came to meeting its total potential profit.
Essentially, the EDF looks at the decisions made for every hour during the year and asks the two-part question shown in Figure 1 below.
Ignoring (very real) complications such as wear and tear, ramp costs, spinning reserve revenues, and the physical limitations of various kinds of generating units, there are clear paths to a perfect operating solution: Operate at full capacity when you can make money and shut down completely when you stand to lose it. Pretty basic stuff. The EDF measures the extent to which a unit's operation matches the perfect operating decision. It also goes one step further and weights the right and wrong decisions by how right and wrong they were in terms of profit and loss.
The basic ingredients for the EDF are hourly generation data,1 the total capacity of unit,2 an estimate of a unit's marginal cost,3 and hourly generator price data.4 These four pieces allow one to define the operating decision tree, quantify the perfect solution, and quantify the extent to which the unit was operated according to the perfect solution. Because the metric ranges between 0 and 100 (measured in percentages), a plant operator can obtain no credit, perfect credit, and everything in between.
The score for each hour is effectively weighted by price. Hours when the market price is very high or very low are weighted higher than average prices. This is an attractive characteristic of this statistic because these are the hours when you make a killing or lose your shirt and therefore probably should be taken more seriously. Another attractive characteristic is that an operator can get as much credit for avoiding losses by turning the unit off during low price periods as it gets for running the unit when the prices are strong.
Examples of the EDF in Action
Figure 2 shows an example of the dispatch of PSEG's Kearny Generating Station 9, an 18.5-MW GT unit in Hudson County, N.J. During the period shown in the chart, the unit posted a 17 percent capacity factor. If the unit were to have generated only when it was the least bit profitable, it would have posted a 35 percent capacity factor. While the unit did not generate every hour that it was profitable to do so, and it did operate during some hours when it lost money, on the whole the dispatch pattern results in an EDF of about 32 percent. Because the EDF is nearly twice the capacity factor, it is clear that the unit did a better job of choosing when to generate than PSEG did if one were to measure total generating hours alone.
Consider all the results for the peaking unit, Kearny Generating Station 9 in 2003 (). During that year, it only ran for 49 hours and had a capacity factor of less than 1 percent, but it picked the right hours to generate, capturing many of the peak price periods. There were 335 hours when the price was higher than the marginal cost but the unit did not run. This did not affect the EDF much, however, because for this particular unit, the best way to obtain a high score is to avoid the losses that would accrue because of relatively low prices. This unit did just that. As a result, this unit scored a 98.61 percent EDF in 2003.
In Table 2 we see the 2003 operating statistics for the Bergen Generating Station CC. Note that for most of the year this combined-cycle unit was generating when the price was favorable. From the capacity factor, it's clear that this unit did not generate at full capacity most of the year. The unit did a good job of ramping its generation to near full capacity when the prices were highest. Note that the unit did generate for 6,146 hours when price was less than marginal cost. That seems like a bad idea, but what these numbers miss is the fact that the unit ramps down to its minimum threshold during most of these low price periods. The EDF score reflects this wise behavior and is therefore much higher than its capacity factor.
EDF Peer Groups
Different types of units will score differently with this type of statistic. In the EDF scoring, base-load units are more often penalized for losing money in the off-peak periods because they are not designed to ramp down. Conversely, they also are more likely to be operating and operating near full capacity during all of the highest prices of the year. A GT unit could do this only if it were to start up in time for a price spike. Thus, just as it would be unfair to compare the capacity factors of a GT and base-load units, it also would be unfair to compare EDF scores outside a well-defined set of peers.
Figure 3 tells two different stories. First, it shows, across all types of units, the relationship between capacity factor and the EDF. When disregarding the type of unit, there is no clear relationship other than the fact that there aren't too many units that operate with capacity factors around 50 percent. Basically, this means that there are base-load units, there are peaking units (or units that act as peaking units), and not much in between.
Clearer trends emerge when one views the chart by type of prime mover. We analyzed the following types: combined-cycle (CC) units, gas turbines (GTs), nuclear facilities, coal steam units, and non-coal steam units. The average EDF scores and capacity factors for each type of unit are presented in Table 3. Note that GTs and CCs have the most variation in their EDF because such units have the flexibility to be operated very poorly or very well. Coal steam units tend to be the best performers because they have high capacity factors and high EDF scores. This is mostly a function of their low costs. They don't lose much money in the off peak, and they're always running near full capacity to capture the high on-peak prices.
Nuclear units have the weakest EDF scores. While nuke units have very high capacity factors, the EDF score penalizes them because they have virtually no ability to ramp up or ramp down in order to reduce the losses associated with off-peak prices.
As a normalized metric of performance, the EDF is particularly useful for ranking the operation of generation facilities. Again, because not all units were created equally, it makes more sense to rank units within peer groups.
1. They have lower fuel costs;
2. They have better heat rates;
3. They are located in a beneficial place on the grid (as reflected in their location marginal prices);
4. They have a higher number of ramp and start periods (i.e., they appear to be more actively managed); and
5. They ramp up and down at the right times.
There isn't too much difference between the performance of the generating units listed in the top 5. However, when you compare units across all of their peers, it becomes quite obvious why some have much better EDF scores than others. The broadest range of scores from our list of top 5 performers occurs in the nuclear category. Entergy's Indian Point 2 received the highest score-nearly 61 percent. Because the unit had a capacity factor of 98 percent, the main variables to consider in the EDF score are the costs and the prices. The prices at Indian Point 2 were higher than costs almost 55 percent of the time, and the unit received a slightly higher EDF simply because when prices were higher than costs, they were significantly higher.
While it was still in the top 5 of the nuclear category, Indian Point 3 (with prices almost equivalent to those of Indian Point 2) received a 54.7 percent EDF score simply because it had experienced a month-long outage. Also, during the outage one of the highest price spikes of the year occurred. Both of these factors combine to result in a lower EDF.
Overall, lower prices in overbuilt markets require generating facilities to perform at their best to achieve even modest success. Capacity factors still serve as a basic metric of success. However, in open markets, where it is possible, we believe a unit's success can be better gauged by considering how well a unit was run from a market perspective instead of just how much it was run. The EDF incorporates many different types of data and boils them all down into a single index of performance that can be used to measure a unit's performance over time and its performance when compared to its peers. The metric is normalized to account for the fact that some units have inherently different input costs (fuel, and operations and management), and normalized to account for the fact that some units receive more attractive prices than others. We believe it is a useful addition to an energy analyst's toolbox.
1. We use EPA CEMS data for hourly generation.
2. We use various public information sources combined with original research.
3. We estimate this using a model that contains FERC Form 1, EIA 423, EIA 906 and other datasets.
4. We match nodal prices from the ISO markets to individual generators. This does imply that our current version of the EDF is only calculated for the ISO markets.
Articles found on this page are available to subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.