Network and Market Management of PJM

*As per the author's request, this transcript is not for quotation.

First Speaker: Mr. Andrew OTT

First, I would like to explain to you how PJM has evolved into the largest wholesale operator in the world. Then I'll talk about how we have split the functions between the various actors in the market and the Regional Transmission Organization (RTO). I will also talk about the some benefits we have seen from combined grid operation within PJM under the market environment.

RTOs are also referred to as Independent System Operators (ISOs) and serve as an information exchange or a coordinator-type function. They do not actually own any assets or directly operate the transmission switching or operation system, but have more of a coordinating function. They are responsible for system operation and operation of all of the markets, or a sort of "air traffic controller" for the power grid. PJM does this through a series of locational price signals, such that each substation on the grid sees a different price. If the price is high in one area, we turn on more generation to attempt to curtail demand there, and in areas where prices are low, we have the opposite effect. If more electricity is needed in an area, PJM raises price signals to tell generators to increase output and, similarly, lowers price signals as an indicator to power output. The most fundamental part of the RTO operation is to balance the supply and demand on the grid, every minute or every four seconds, depending on whether we are dealing with regulations or the energy market.

As an RTO, PJM sets all of the reserve and reliability requirements within the grid operation. Rather than the transmission owners, we are actually legally responsible to the federal government for reliability coordination.

PJM also administers the transmission tariff charges when someone wants to purchase the right to access the transmission grid. Instead of the previous wheeling-type contracts for transmission use, there is now an access fee called a "license plate tariff." As part of the other functions of administering the tariff, we also require transmission owners to coordinate their switching operations through us in order to schedule outages and maintenance on the transmission lines.

PJM does real-time security analysis, day-ahead security analysis, and deployment of voltage control devices on the transmission grid. The transmission owners get the rates approved by the regulators, and they also do distribution system-level operation, in contrast to wholesale marketers or regional grid operators.

The best way to describe PJM's relationship with the other two main market actors--the load serving entity and the generation owner--is that PJM does the reliability part of load balancing and generation scheduling to ensure balance both on the day-ahead and during the operating day. The load-serving entity enters into both long-term and short-term contracts to ensure generation services but not for reliability. While the load-serving entity has a responsibility to follow the reliability instructions, it operates within a profit-motivated market as does the generation owner. The power pool is a voluntary one. Generation owners can decide whether they want to self-schedule or offer into the market and allow us to dispatch it. This decision will depend on their commercial position. But since we have all of the information about system operations, very seldom do they self-schedule because they can make more money by scheduling through the market.

I would now like to compare the PJM locational price model to other ones. The fundamental difference between the three models is the perspective from which the market operations and the systems operation were built. For a locational pricing model, the primary focus was maximizing the operational efficiency of the grid. Regional operations are combined, so all of the transmission is within one market and under one security analysis and one control area, etc. Because the market under this type of model is more complex, there is five-minute pricing as well as a day-ahead forward market, which is an hourly market. Every substation has a price driven by the balancing of the energy component, supply, and demand. There is also a congestion component representing the cost of controlling congestion and a component representing the cost of losses on the transmission system.

PJM has seen significant operational efficiency and a significant 16-gigawatt (GW) investment and demand response. The market is designed to be relatively accessible, less-complex, and focused on the forward-market design. There is much less focus on tying the market operations to the grid operations. The basis of the model is to make the forward market as simple and liquid as possible. The Korean Power Exchange (KPX) model was a multiyear plan, but it stopped at a cost-based pool. Now it has a dominant single buyer. Its model doesn't have much transparency because it hasn't evolved into a forward, or more sophisticated, market.

Combining many separate transmission owners into one market adds a significant amount of market efficiency as shown by the two case studies I will introduce; it makes it possible to deploy all of the generators in the market to meet one forecasted load and to examine the operations of individual transmission operators for the regional benefit, as opposed to the individual area benefit. Having a large regional operation instead of separated locational operations significantly reduces the cost of compliance for reliability within that market; PJM has had 42 GW of merchant investment in generation over the life of the market.

Case Study 1

In October 2004, two large parts of PJM's market today were operating separately. One was an LMP, a locational price-based market. The other was a large bilateral market composed of three transmission owners. Because there was no cost involved for transmission service between them, there was no barrier to trade from a transmission service point of view. A study has found that, before the market integration, the hourly traders could only get about 40% of the utilization of the transmission because the lack of a combined regional operation was inefficient. As soon as these two areas were integrated into one market, 3,000 megawatts (MW) of increased transfer occurred on the transmission grid. The equivalent of three or four 500-kilobyte (KB) lines of additional transfer capability was created without building new lines; the system was being used more efficiently. PJM spent about $40 million in infrastructure costs to operate the grid and saved $180 million a year. The authors of the study said that this was one of the best investments they had seen in this type of industry.

Case Study 2

A similar effect occurred in the state of Virginia, south of the Washington D.C. area in the United States, a very large consumer of power that imports a lot of energy. Prior to its integration in 2005, it had limited transfer capability. When it was brought in under the same market operations and the seam was reduced, transfer capability went up overnight because it was possible to utilize both the transmission and the generation in that area more efficiently, meaning a significant increase of over 1,000 MW on the northern and western paths. It reported to its regulator savings of $750 million over a four-year period and also avoided costs. My colleague, Adrien Ford, will now discuss some of the activities we have conducted in demand response.

Second Speaker: Ms. Adrien FORD

What we mean by "demand response" is a reduction in consumption. End-use customers use less energy, and PJM uses that as a resource in place of generation. The PJM delivery year runs from the beginning of the peak season in June through the following May, and in the current delivery year, demand resources compose 8% of the resource mix. Starting this June, in the 2012-2013 delivery year, another demand resource will come into play on the consumer side of the equation, and that is energy efficiency.

Demand response is a reduction by a consumer in response to some trigger such as a price or an emergency action by PJM. Energy efficiency, on the other hand, is a reduction that occurs due to an ongoing action to reduce energy consumption such as adding insulation to a building. Customers reducing energy consumption can make money in PJM markets during the years it takes for the load forecasts to catch up with the energy efficiency reductions.

One of the two ways customers participate is as an emergency resource--one of the resources that responds to the trigger of an action by the grid operator in an emergency situation where immediate reduction is necessary. They are compensated for their actions during the emergency, but they also receive payments, regardless of whether there is an emergency or not, just for committing to be available. Typically, a capacity market is used to pay generation resources for availability, but PJM's emergency demand response resources also receive capacity payments because we can use them as a potential replacement for a combustion turbine or some other resource.

Economic responders bid into PJM's markets, and, if the price for a particular hour goes above that at which a resource is willing to respond, they respond to the price trigger. Many resources respond to both. We are a wholesale organization so we do not deal directly with the end-use customers; the aggregators in between represent over one million users in our footprint, including about 14,000 commercial and end users.

Nearly 40% of the business segments participating in these demand resource markets come from industrial and manufacturing-related customers, and another 40% have multisource portfolios--we don't necessarily know the segment because the aggregator has pooled them together--to make up one big resource that it brings forward to us at the wholesale level. Resources in other notable segments such as office buildings, schools, and hospitals sometimes respond by turning on a backup generation unit at the site or taking other actions to reduce electricity consumption.

Aggregators that have pulled together at least one-tenth of a MW, 100 KW, can come to the wholesale market. They can participate as an emergency resource directed by the grid operator and receive capacity payments for availability or energy payments for response. By responding to price, they can participate in the energy market as well as in the reserve markets and regulation markets.

There is no direct link between PJM, a wholesale operator, and the end-use customers; multiple organizations perform several different roles in the market which connect with them. As aggregators, curtailment service providers coordinate demand response actions for many different resources and bring them to the wholesale operator as a pooled portfolio. End-use customers can also be linked with load-serving entities and with electric distribution companies, which are the wire companies that make sure the electricity gets through the distribution system.

PJM does not require two-way communication to measure the response provided by demand response resources. The overall system load can be seen on the control room floor, but we need to be able to measure and verify the response after the fact in order to provide an accurate bill. First, we look at the customer baseline load (CBL)--what the anticipated usage was. If there was a demand response action, actual usage will be reduced in comparison. PJM decides what the financial settlement should be based on the resources' response by comparing the hourly measurement to the anticipated usage.

The response when PJM calls on an emergency action is clear from the drop in the load shape. A phenomenon that often occurs at the end of action is the "snap back." Instead of a continued decrease in the load, demand snaps back to the level of the anticipated projection and then continues down. We manage this by bringing the demand resources off of their interruption in a stepped fashion.

Dealing with demand response uses different systems for the control room floor, the markets, and registering demand resources. Alstom's Eric Goutard is here to talk to you in greater detail about the IT side of the systems we use.

Third Speaker: Mr. Eric GOUTARD

I am the head of the worldwide Market Management System Activity within Alstom. Alstom Grid is a transmission company with about 20,000 employees, and it is part of the Alstom Group, which works in power transportation and transmission and has about 100,000 employees. I work on the product line in charge of Network Management Solutions, which comprises all of the IT systems used in the control room for transmission and grid management. We do generation and distribution, market management system activity, which is my core responsibility, and demand response.

Alstom has experience with many different market mechanisms. In addition to delivering the IT tools for market operation and demand response at PJM and many other U.S. ISOs and RTOs (MISO, ISO-NE, SPP), we also have a presence in Europe and in the Asia-Pacific region, where the KPX and the Transpower in New Zealand are some of our solutions.

As an example that illustrates the kinds of IT systems that support markets and grid operation, up to 2008-2009, PJM was operating two market systems: day-ahead and real-time. The real-time market operation was rather sequential, solving the real-time dispatch every five minutes. But the complexity of the grid was growing, with increasingly more intermittent generation and constraints on the grid. The analysis needed to evolve into a real-time look-ahead type and thus become more oriented on multi-time points optimization. That meant moving from sequential optimization to a kind of multi-time point trajectory in order to reach the optimum over a certain period of time in the future, improving the minimization of system and market operation costs. Typically, the PJM generation control application (GCA) is looking at a two-hour rolling time window.

Mr. Andrew OTT
When PJM had what I will call "traditional resources"--large central station generation that responded to dispatch instructions--it was enough to send out dispatch instructions that were isolated every five to 15 minutes depending on the time step. However, new technologies, demand response, and new alternative resources, such as grid size batteries, were coming into play, and we needed to have more predictability. To do this, PJM worked with Alstom, which tends to be the most innovative market system provider. We found that time coupling of dispatch--looking at how dispatch cycles are connected in time--gives a much more accurate "dispatch envelope." This trajectory of dispatch over time allows PJM to predict prices in the future.

About one million individual customers use electrical appliances in their homes. Price-responsive demand allows them to cut that load off if the price goes above a certain amount every five minutes. But each cannot individually offer that into our market. So now, PJM actually creates a load forecast that is both weather and price dependent and uses the forecast as an operational tool. This gives us what we call a "price responsive load," where at every substation, we try to predict the load response to the price based on the automation that is installed. This requires a more sophisticated dispatch. On the cost side, we are able to displace generations with intelligent demand response, so the payback is less than two years.

Mr. Eric GOUTARD
The key complexity here is managing a single optimization problem. This is about maximizing the social welfare and minimizing the cost, which means we are not focusing on just the price or the cost aspects. It means taking all of the technical constraints into account in a single optimization, not only those on the generation side or those on the demand response side. It also means taking into account the complexity of the grid. The integration into a single optimization--great reliability aspects combined into a single approach with the market operation--is what makes this solution, and the underlying market design, different from what we find in Europe.

In the United States, PJM is a member of the Association of Power Exchanges (APEx) and also has responsibilities with regard to grid reliability. In Europe, the power exchange and the grid system operation are two separated roles, while in the United States, you will find both roles under the same umbrella. Norway, Sweden, Finland, and Denmark are all operated on the market point of view under a single umbrella by the Nord Pool, which operates two markets--the Elspot day-ahead market and the Elbas intra-day market. The Nord Pool market is quite liquid in Europe. About 75% of the electricity is traded on the market. In France, it is more like 15% that is traded on the market and 85% is bilateral. The key point is that each transmission system operator (TSO) is independent and operates its own grid, in coordination with neighboring ones. That's why special IT systems to help with the coordination of system operations are required. Alstom delivers the Nordic Operation Information System (NOIS), which is in charge of summarizing and sharing information related to reserve management, transmission capacity management, and balance management, as well as outage management coordination among the four Nordic TSOs.

Even in Europe, there is a trend nowadays for regional cooperation and for more "flow-based" market coupling. In Europe, in order to improve the "copper plate model," flow-based methods are starting to be introduced which now take into account not only how one MW injected into one zone affects the intertie between two countries, but also how it affects the interconnection physical line. And if you open the "box," it is really close to the calculations which are made in the United States, such as at PJM, for example, because it is required to have an explicit model of the grid in the calculations so that sensitivity factors (called shift factors) can be computed to determine the so-called Power Transfer Distribution Factors (PTDF) which are key elements for "flow-based" market coupling. In Europe, the shift factors are computed at a zone/area level (zonal sensitivities), while in the United States, they are computed at the electrical node level (nodal sensitivities).

Let's move now to demand response. Demand response systems are necessary to make possible what Adrien explained before. Nowadays, there are demand response operation centers. Curtailment Service Providers (CSPs) are aggregators of demand response capabilities which notify their customers of some event they have received from an upper level operator like PJM. Curtailment operations are then carried out either via direct load control or by some other method. An important step is verifying how the actual response matches what was expected. Finally, the system needs to be restored.

To do all of this, an operator needs IT systems with functions such as allowing different participants and their demand response contracts to be registered, which mostly entail the curtailment service provider entering the flexibility they contracted with their portfolio of demand response. The operator needs to keep track of and compute the customer baseline load (CBL)--the expected answer from the demand response. It also needs to be able to measure, get metering data via the infrastructure, and do the corresponding settlements. In addition, an operator needs many connections with the control room, the market management system, and for scheduling, as well as more and more functions to support the decision making. For example, 14 different steps are required in order to avoid the "snap back" at the end of an event. The operator may also like to know, at a certain stage, what could be made available for the system. All of these functions are available in the Demand Response Business Network (DRBizNet) platform that Utility Integration Solutions (UISOL, an Alstom company) uses at many customer locations, including at PJM.

YAMADA Hikaru (President, Sprint Capital Japan Ltd.)

My comments deal primarily with the question: "What kind of market should Japan target?" What is the best design for a network with constraints such as a shortage of power resources? Rather than simply looking at problems and solutions, the attendees of today's seminar were able to hear about the importance of market design, which is something that will require more discussion in terms of what Japan should do.

We also learned the importance of information disclosure. As information is not always available in Japan, this is something that we may have to solve. The speakers also emphasized capacity. In terms of generation capacity, we have to look at the capacity markets and the capacity models. The PJM model, as I understand it, is a three-year forward capacity market starting from the third year, from June to May of next year. Thus, the capacity market is available for the generator and demand response service provider. Both are of equal footing, even though the generation side is much larger. The other type of capacity that we heard about was transmission capacity, and this is another important point; if capacity is available but remains unused, capital expenditures are not being wisely utilized. In order to improve the efficiency of the current infrastructure, it is important to always think about how much of each of these generation capacity and transmission capacity can be utilized. Budget or site constraints should make us think about how to make more effective utilization of the existing infrastructure in terms of transmission and generation, rather than thinking about building more.

Lastly, the speakers talked about using metering data to verify how much of a reduction was made. To verify the data, you have to know it, and the consumer also needs to know it. There has been considerable discussion regarding who owns the data generated by utilities in Europe and the United States. In the United States, it depends on the state, but in Europe, I believe there is an European Commission (EC) statement that the data should belong to the consumers. Japan needs a definition saying that the information belongs to the consumers, and that they will allow an RTO or a public entity to use the data, not only so that it can be used efficiently, but also in order to make efficient use of the system itself. If we do not act with efficiency, it is necessary to build new assets continuously every year or every month. It is best to utilize maximum capacity within the existing infrastructure.

Mr. Andrew OTT
PJM learned very quickly that commercial deployment of demand response and infrastructure that is geared toward getting demand response to develop requires lowering barriers to entry. We saw the barriers reduced by about 90% in our footprint based on two fundamental developments. One was standardization of data, without which the costs become prohibitive to commercial investment. The second big item was the states' decision to allow easy, low transaction-fee access to the wholesale market and access to the revenues; to achieve commercial demand response, it is necessary to remove the barriers to entry.

Q: You said that demand response is equal to generation. However, since demand response is voluntary, how do you handle the gap between what is expected and the actual response?

Adrien FORD
When it comes to demand response that serves as a capacity resource, response is no longer voluntary. A customer that bids into the capacity market and offers itself as an emergency resource is responsible for compliance when called upon, and if it does not do so, there are financial penalties. It is a little different for the economic responder as economic demand response is voluntary. However, in the years that we don't have an emergency event, we have conducted response tests. In the first year of testing, we had 110% compliance. Last year, our compliance dropped somewhat to 90%. Right now, we are working on better ways to ensure that we have 100% availability.

Andrew OTT
PJM carries three types of reserve: synchronized response, which is a 10-minute reserve; frequency regulation, which is a four-second response; and a 30-minute reserve. These are to cover for unforeseen events. Operationally, we have found that our reserves cover forecast errors in economic demand response; even though economic demand response is voluntary, PJM has gotten good at predicting the price responsiveness. Surprisingly, though, demand response is not nearly as unpredictable as might be imagined.

Q: Can demand response be used to control a transmission constraint?

Andrew OTT
Yes, definitely.

Q: If transmission capacity is overloaded, is the system designed so that demand response is consigned to specific customers?

Eric GOUTARD
When a curtailment service provider that can provide demand response services registers, it specifies many pieces of technical information, among them, the location. The upper level operator can use this locational information to help solve the problem.

Andrew OTT
In order to use demand response as an operational resource, PJM needs to know some very fundamental things, one of which is the accurate location of the demand response. In the case of economic response, we need a reliable prediction of the response. Decisions about moving a generator's demand response up or down are made based on what is the most economically viable in real time, regardless of commercial relationships that exist outside of operations.

Q: There was an emphasis today on efficiency of the markets. However, in the wake of the recent earthquake and in light of the damage to the Japanese auto industry, how does efficiency in normal times balance against possible human error or natural disaster?

Andrew OTT
Operating more efficiently means combining regional operations to make it into a larger one. PJM still operates with what we call a "single contingency criterion." A certain amount of reserves and extra capability are held back within the system to enable us to withstand equipment loss. Combining the regional operation to be more efficient still respects this. Thus, it is more of a systems type of efficiency than the supply chain efficiency that you describe.

Adrien FORD
Combining transmission operations means increased redundancy, which in turn means having more reserves to call upon if one unit goes down. The increase of coordination between these operations is what we mean by efficiency, and it gives us more reserves upon which to draw.

Q: In the United States, is the system operator legally responsible for an outage even if the transmission owner is negligent? Would the customers demand compensation from the system operator and the system operator demand it from the transmission owner?

Andrew OTT
If the federal government were to impose a fine for not meeting reliability standards, it would be on the RTO, and then we would distribute the fine to our transmission owners. The entity responsible for ensuring compliance with the standards is the RTO. However, the relationship between the transmission owner and the RTO is still one of mutual cooperation. PJM pays for the transmission operation center for each of our transmission owners. We have video communication with their control centers, and we both work to maintain reliability.

Q: PJM is often cited as a very successful case of the separation between transmission and generation. However, I have heard that there are many failed cases, even in the United States. What is the main difference between PJM and the failed cases?

Andrew OTT
You mean like the California ISO, for example? If you look across the United States, every RTO basically operates on the same model. The California ISO has put in a regional system operator type model, very similar to what PJM is now, as has Texas, with the Electric Reliability Council of Texas (ERCOT).

PJM actually had an operational failure in 1997. It was not well known because we quickly corrected it. At the time, we had a single clearing price system across the entire market. On a hot summer day in August, we couldn't manage the transmission congestion, so everyone pulled out dispatch, and the price dropped to zero. Everybody pulled out of the market, and it collapsed. Within four months, we put in an LNP-based market, and it became successful. Something very similar also happened in New England.

In California, the biggest problem was the fundamental separation between the power exchange and the system operator. The state also required its utilities to divest all of their generation and not buy in long term contracts. The combination of these things is what took down California. Most of the failures in the United States have been related either to the way the system was set up or regulatory decisions, or a combination of the two. We haven't had any such failures recently.

Q: I was amazed to see that the electric power price is very different between the U.S. utilities that have introduced demand response and those that haven't. What is the difference in the electric power price due to introducing demand response, and what is the impact of introducing this system?

Adrien FORD
Some of that depends on what the alternative price of generation would have been and the prices at which the demand response bid in. But generally speaking, it is a more affordable resource than new generation, so that means a reduction.

Andrew OTT
Generally, demand response does not have much impact on wholesale price until it is above $75. When the price gets to $150 or $300, then you start to see a tremendous influence. We actually did a study in 2008 where we had three consecutive hot days, and we got 3,000 MW of response each day. We estimated that it saved customers about $280 million. Demand response providers' revenue out of our markets was around $450 million in 2010, and it is my understanding that it is just under $1 billion now.

Q: Do you have many competitors in the area?

Andrew OTT
No. PJM is the grid operator. We have a territory, and we are basically a monopoly.

Q: What makes your management efficient: pressure from the capital market or government regulation?

Andrew OTT
A transmission owner's participation in an RTO is voluntary. PJM just grew by 20% last year because two transmission owners left the RTO to our west and joined us. There is some competition between providers at the borders. Probably even more important is that PJM is managed by an independent board elected by the members. This gives a fairly significant performance incentive to make sure management operates efficiently. In our case, we happen to be the lowest-cost RTO of any in the United States. We like to think that this is not only because of our size, but also due to our efficiency.

Adrien FORD
We have over 750 members from various industries, and they are very actively engaged. We have committees that help decide how we are going to change the rules under which we operate, and the members are very actively involved in determining our path forward.