But not for long (em as power producers and

customers get more creative in matching plants with loads Dynamic scheduling is a "sleeper" issue in the move toward electric competition. Industry players are debating independent system operators. They are focusing on issues of governance and the form of transmission pricing. Consequently, they are ignoring critical issues concerning ancillary services. These services are not receiving the attention they deserve.

Although electric utilities have used dynamic scheduling for at least two decades, it has taken this long to start growing in popularity and importance. This growth is a consequence of major changes under way in U.S. bulk-power markets, and, in particular, efforts to unbundle generation from transmission and increase competition among generation providers. It is an issue that will grow in importance over the next few years, primarily because of its potential to expand the geographic scope of competition in bulk-power markets.

Because of the growing importance of dynamic scheduling in bulk-power markets, we collected and analyzed data on utility experiences with this practice. We spoke with individuals from 32 investor-owned utilities, federal and state utilities, public power utilities and other regional and national entities about the following issues:

• Alternative definitions and applications of dynamic scheduling;

• Size of loads and generation that are dynamically scheduled;

• Any differences between dynamically scheduling loads vs. generation;

• Possible limits on the distance (or the number of intervening control areas) between the control area in which the load is located and the control area providing the generation services for that load (or by dynamically scheduled generation);

• Possible limits on the number of schedules that can reasonably be accommodated;

• Speed of response, reliability, and cost of sensors, telecommunications, computing and other equipment required to accomplish dynamic scheduling;

• Indicators of the success of dynamic scheduling (and the metrics used to measure success);

• Contingency actions taken when metering, communications, or computer equipment fails or when transmission outages occur; and

• Cost-causation factors and the costs to dynamically schedule loads and generators.

Reasons for Implementation

Perhaps the most powerful reason to offer dynamic scheduling is that it promotes competition and increases choice. It allows electricity consumers to purchase certain services from entities outside the control area of their native distribution utility and allows generators to sell certain services to entities other than the host distribuor. Increasing the number of possible suppliers and consumers increases competition, which should encourage innovation and reduce costs and prices. If such choice becomes widely available and used throughout the country, the Federal Energy Regulatory Commission may decide to relax the requirements it now imposes on transmission providers to offer six ancillary services to transmission customers at regulated prices. Instead, companies could provide some of these services voluntarily at market-determined prices.

Various factors might motivate the use of dynamic scheduling. The most frequent example is for jointly owned generating units. Dynamic scheduling of a load (e.g., a municipality) from one control area to another also is common.

An electric customer that manages loads in more than one control area (e.g., a manufacturer with several facilities) might want to aggregate its load and purchase all its electrical needs from one provider. Such aggregation can reduce costs for some services, especially regulation (see Table 1).

Similarly, the manager of several generating units in several control areas might want to aggregate their outputs and sell all the basic energy and capacity, as well as ancillary services, to a single customer or another control area. In addition, cost reductions might motivate dynamic scheduling. For example, if control area C pays higher prices for ancillary services (such as regulation and spinning reserve) than control area A, a generator located in control area A might want to dynamically schedule its output to C.

Examples in the U.S.

In some generation examples we have uncovered, the units are used for regulation, and in other cases they are not. The Western Area Power Administration (WAPA) operates the 17 hydroelectric units at Hoover Dam. Four other control areas in the Southwest send signals to WAPA every four seconds, requesting time-varying amounts of generation (up to their contractual allocations); because hydro units respond very quickly compared with fossil units, the four control areas use their Hoover rights for regulation. WAPA aggregates the four requests and sends the total to Hoover Dam.

Similarly, Otter Tail Power purchases supplemental regulation and load following services from Manitoba Hydro because Manitoba Hydro's hydroelectric units can provide faster regulation service at lower cost than can OTP. OTP began receiving up to 50 MW of regulation service from

Manitoba Hydro in 1990. OTP's compliance with the A1 and A2 criteria has improved substantially, from less than 80 percent before obtaining regulation service from Manitoba to more than 98 percent.

Montana Power operates the four Colstrip units in Montana, for itself and the four other owners. Although these units are operated in baseload mode, rather than as regulating units, dynamic scheduling still made sense. Absent dynamic scheduling, all the errors in plant output (i.e., differences between actual and scheduled output) would contribute to Montana Power's inadvertent interchange but not to that of the other owners. Dynamically scheduling the units equitably shares any errors in plant output and reduces the need for control-area and plant operators to manually change schedules.

PacifiCorp operates generating stations and serves loads in seven Northwestern states (Washington, Oregon, California, Montana, Idaho, Utah and Wyoming). Most of its generation is in its eastern division (Utah and Wyoming), while most of its load is in its western division. The eastern generators are primarily large coal units. The western generators are primarily hydro units. Before the merger that created PacifiCorp, the two original utilities operated separate control centers in Salt Lake City and Portland, Ore. Now, PacifiCorp operates its entire system from its Portland facility. This consolidation allows PacifiCorp to use the eastern units for baseload capacity in the West and to use the western hydro units for regulation in the East. The Idaho Power

system lies between the two PacifiCorp divisions. PacifiCorp purchased firm transmission rights for 1,600 MW of capacity flowing from east to west across the Idaho Power system. Of this total, 1,500 MW is used for base capacity, and the additional 100 MW is assigned to regulation. PacifiCorp also compensates Idaho Power for losses on the Idaho transmission system based on contractually agreed upon assumptions, not based on metered flow or detailed calculations.

Several utilities in the Southwest co-own several generating units (see Figure 1). The outputs from these units are dynamically scheduled from the plant's operator to the other owners. The coal-fired Navajo station is operated by the Salt River Project (SRP) but is physically located within the control area of Arizona Public Service (APS). The reverse is true for the Palo Verde nuclear plant, which APS operates and located within the SRP control area.

Using it to Schedule Loads

The examples of dynamically scheduled loads show that the amounts of power transferred are generally much smaller per delivery point than for generation. For example, the Geneva, Ill., load of about 45 MW is the aggregate of loads collected at eight points on the Commonwealth Edison transmission system. The city of Geneva aggregates the load and then sends the aggregate signal to both Commonwealth Edison (the physical host) and Wisconsin Electric (the electronic host).

The Cajun Electric Power Cooperative load, which is scattered throughout the service territories of four investor-owned utilities, involves the separate metering and telemetering of loads from about 125 points to the Cajun

control center as well as to the physical-host control centers. On average, each load point is about 10 MW. (Cajun also uses dynamic scheduling for five generating units that it operates, which are located outside its control area.) Cajun has many transmission contracts with the surrounding utilities to cover these transactions. Losses are handled on a contractual basis, depending on the utility and the voltage levels at which loads are measured.

The Central Arizona Project is perhaps the most interesting example uncovered in this study because it involves the simultaneous scheduling of loads and generation. The project brings Colorado River water to the areas around Phoenix and Tucson in Arizona. It is the largest irrigation project in the world. It includes 15 major pumping stations with a load of about 550 MW, all in WAPA's control area. WAPA dynamically aggregates these loads and sends the aggregate signal to the Salt River Project control center in Phoenix. SRP meets the project's load with output from the coal-fired Navajo plant in Page, Ariz. SRP meets any regulation requirements for the project from its overall resource pool. The Navajo plant is located in the APS control area, so it schedules the plant's output across its transmission system. The three, 741-MW Navajo units total 2,223 MW and are co-owned by six utilities: Los Angeles Department of Water and Power, Nevada Power, Tucson Electric, SRP, WAPA and APS. APS is the physical host for the power plant, and SRP is the electronic host. WAPA is the physical host for the project load, and SRP is the electronic host.

The Lower Colorado River Authority (LCRA) traditionally had difficulty meeting its performance criteria, primarily because of a volatile steel-mill load in its service area with a load that can change rapidly by as much as 100 MW. Beginning in January 1995, LCRA purchased supplemental regulation service from Houston Lighting & Power. Since then, LCRA's compliance with the North American Electric Reliability Council A2 criterion has improved from 74 percent to more than 92 percent. LCRA uses its own generating units to provide regulation service within a ±12- or ±25-MW bandwidth. When the LCRA area control error, or ACE, moves outside this bandwidth, usually because of steel-mill operations, HL&P provides supplemental regulation to move the ACE back inside the bandwidth (see Figure 2).

This transfer is profitable for both utilities, in part because HL&P's generating capacity is about six times as much as LCRA's. This supplemental regulation can be considered dynamic scheduling either of load or of generation. In one sense, this example is the inverse of what Manitoba Hydro provides to OTP. In another sense, LCRA is dynamically scheduling the volatility associated with its steel-mill load to HL&P. Because the two control areas are generally not contiguous, LCRA pays wheeling plus loss charges to the intervening utilities.

What Is the Cost?

Dynamic scheduling involves both initial and ongoing costs. Initial costs include those related to the purchase and installation of additional metering, telemetering, communications and computing equipment. For example, utilities may need to obtain or upgrade communications systems, such as leased telephone lines, microwave systems, or fiber-optic-cable systems. Energy-management systems in control centers may need expansion, involving both hardware and software. In other cases where data are already being collected and telemetered to the control center, initial investments might be quite modest. The primary ongoing costs relate to communication systems (especially to lease telephone lines) and to periodic inspections, maintenance and repair of field equipment.

Because these costs are so dependent on the specific circumstances, we obtained only a few estimates of the initial and ongoing costs of dynamic scheduling. Based on these estimates, the incremental initial cost is $10,000 per metering point and the ongoing costs are $2,000/month.

For a single metering point (e.g., the output of a single generator), the monthly payments are $2,130. For a 100-MW generator with a load factor of 65 percent, this $2130/month is equivalent to 0.045 mills/kWh. Compared to spot prices (roughly 25 mills/kWh), dynamic scheduling is very cheap (em well less than 1 percent of the cost of power.

If dynamic scheduling involves the use of another utility's transmission system (as it often does) and if that utility charges for transmission service and losses, then the cost increases dramatically. %n*%n Assume that this dynamic scheduling of a 100-MW generator involves a transmission charge of $1.5/kW-month and losses of 3 percent charged at an average price of 25 mills/kWh. Now the cost increases from 0.045 to 3.96 mills/ kWh (0.045 for dynamic scheduling + 3.16 for transmission + 0.75 for losses). These results suggest that the costs of dynamic scheduling itself might be a very small percentage (about 1 percent in this example) of the costs of transmission and losses (see Figure 3).

Some Issues to Consider

Dynamic scheduling is a reality. Many utilities have used it for at least two decades, and it works. One system operator at the receiving end of a dynamically scheduled generator said that it operates "as if it was in our backyard."

We uncovered a range in the size of loads and generation that is dynamically scheduled. Generally speaking, the generation applications are for larger sizes than are the load applications. Several generation applications involved more than 1,000 MW, whereas many load applications were less than 100 MW. The number of points from which data need to be collected and telemetered may emerge as more important, especially from a cost standpoint, than the size of the load or generation. The loads typically involved several metering points, whereas the generation examples generally involved only the generating station itself.

None of our survey respondents offered clear rules concerning the maximum distance or maximum number of control areas over which or through which dynamic scheduling could be conducted. Our cost analysis suggests that the costs of transmission service and of loss replacement dominate the direct costs of dynamic scheduling. The need to compensate multiple control areas for transmission and losses would limit such applications of either dynamic or static scheduling.

Typically, data are telemetered at the same rate as that used in the control area's automatic generation control, or AGC, system once every two or four seconds. We learned of no special problems associated with the accuracy and reliability of the meters and

telemetry equipment used for dynamic scheduling. Indeed, most respondents indicated that problems associated with accuracy and reliability were no different from those associated with other systems for data collection and transmission, such as those used to record and transmit tieline flows and voltages to the control center. In a similar fashion, the control centers have prepared contingency plans that specify the actions to take when equipment fails.

The costs of dynamic scheduling appear quite low compared with both the cost of power and the costs of transmission and transmission losses. In addition, the many applications of dynamic scheduling we found suggest it is cost effective.

Although our findings are generally quite positive about dynamic scheduling, we need to be cautious about its widespread use, given its likely increase in the complexity of control-area operations and its possible adverse effects on reliability. We should be mindful of concerns about metering complexity, metering and telemetry failures, metering and computing errors, and possible confusion during emergency conditions over which entities are responsible for meeting which loads. These concerns, however, need not limit the cooperative and coordinated expansion of dynamic scheduling.

The number and complexity of dynamic-scheduling transactions will likely increase in the future as generation becomes more competitive, open access of transmission networks becomes a reality, pancaking of transmission rates is reduced, and market participants (customers, suppliers, marketers, brokers and others) seek more choices. t

Eric Hirst and Brendan Kirby are senior researchers in the Energy Division, Oak Ridge National Laboratory. Their research focuses on electric-industry restructuring. The work described here was sponsored by the Detroit Edison Co. and the U.S. Department of Energy, Office of Utility Technologies.

What Is It?: The Details of Dynamic-Scheduling

The control area is a critical component

Dynamic scheduling. The electronic transfer from one control area to another of the time-varying electricity consumption associated with a load or the time-varying electricity production associated with a generator.

The FERC, in its Order 888 on open-access transmission, defined dynamic scheduling as:

"[T]he metering, telemetering, computer software, hardware, communications, engineering, and administration required to allow remote generators to follow closely the moment-to-moment variations of a local load. In effect, dynamic scheduling electronically moves load out of the control area in which it is physically located and into another control area."

The FERC decided not to require transmission providers to offer this service. It also did not address the dynamic scheduling of generation.

Control area. The NERC defines a control area as: "An electrical system bounded by interconnection (tieline) metering and telemetry. [Each control area] controls its generation directly to maintain its interchange schedule with other control areas and contributes to frequency regulation of the Interconnection."

Four bulk electric networks (interconnections) exist in North America. Within each interconnection, all the generators are synchronized and therefore operate at the same frequency, and electricity flows freely on AC transmission lines. Electricity flows between the four interconnections are limited and occur only on DC links.

Control areas seek to minimize any adverse effect they might have on other control areas within the interconnection by minimizing their area-control error (ACE). ACE is the instantaneous difference between actual and scheduled interchange, adjusted to take account of any difference between actual and scheduled frequency in the interconnection.

Each control area maintains some generating units on automatic generation control (AGC) "to continuously balance its generation and interchange schedules to its load." AGC refers to equipment in the control center that automatically computes ACE every few seconds and, based on the calculated errors, sends signals to generating units to increase or decrease output to reduce ACE to zero. AGC also refers to equipment at individual generators that respond to the control-center AGC signals.

NERC defines two key control-area performance criteria. The first (A1) requires that, on an instantaneous power basis, the control area be in balance with the rest of the interconnection (i.e., ACE must equal zero) at least once every 10 minutes. The second criterion (A2) requires that the control area's energy imbalance (average ACE) be within a certain limit (roughly 0.2 to 0.5 percent of peak demand) every 10 minutes.

The electronic transfer of a load (or of generation) requires adjustment to either the actual or the scheduled interchange terms in the ACE equations of both the physical-host control area and the electronic-host control area.

Table 1. The Effects of Aggregation

on Regulating Requirements for an Industrial Customer

Load Load Load Sum of Metered

#1 #2 #3 loads total

Mean (MW) 84 98 88 270 270

Standard deviation (MW) 2.4 11 0.9 4.4 2.4

Range (MW) 12.7 5.5 4.8 23.0 13.6

The table shows hourly statistics based on 10-second data for three separately metered components of the load at an aluminum plant as well as the plant's total load. Clearly, aggregation reduces the volatility of the load that generators must serve. The sum of the ranges of the three components is 23 MW, but the range for the total load is less than 14 MW, a 40-percent reduction.

*Inclusion of transmission costs and losses assumes that dynamic scheduling is compared to a reference in which electric service is provided from within the physical-host control area. If, however, the reference is a static schedule, then there may be little or no incremental transmission costs or losses associated with dynamic scheduling.

Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.