1 Introduction

Energy storage could offer great flexibility to the operation of the grid, and thus provides values by improving stability, hedging against high-energy price or peak demand, or contributing to ancillary services [1,6–10,]. One of the services that energy storage has great potential to offer is regulation service [11, 12]. In the past, regulation service was mainly provided by the synchronous units responsive to automatic generation control (AGC), which match the supply of electricity with the demand in sub-minute intervals. This service is critical to ensure that the grid frequency is maintained in compliance with reliability standards. As a fast-acting resource, energy storage is emerging as a promising alternative to providing regulation service since they have different capabilities and operating characteristics compared to how current generator technologies offer regulation today. The key attributers of energy-storage technologies include:

• Fast Response – energy storages have the ability to quickly increase or decrease their provision of frequency regulation service, which is known as ramping capability.

• Accuracy in responding to the dispatch signal-energy storage can precisely follow the AGC signal if its energy is neither depleted nor charged at full capacity.

• Operates around a zero or slightly negative base point – energy storage can inject and withdraw power from the grid. Energy storages with a well-designed charge/discharge strategy will result in a close-to-zero energy charge.

• Scalability – Batteries/flywheels can be connected either in parallel or in series to meet the capacity requirements due to their modular design.

Today it is attractive to use energy storage for regulation service since it responds nearly instantaneously to a control signal and accurately follow instructions in maintaining frequency. This will reduce costs by introducing new competition to the market and decreasing the amount of overall regulation procurement. Moreover, existing fossil fuel-powered plants displaced by storage-based frequency regulation would be available for energy dispatch, potentially allowing the plants to run at a higher capacity, improving their energy efficiency and reducing emissions. The advantages of dispatching storage resources to utilize their fast response capability are summarized in Table 4.1.

Since there are significant benefits and advantages for energy storages to participate in regulation service, pilot projects have been tested or new market designs have been implemented to remove the barriers for energy storages to participate in the regulation service market. For example, to just name a few, Beacon Power has been operating up to 3 MW of flywheels to provide fast-response regulation service in ISO New England, while ERCOT (Electric Reliability Council of Texas) has successfully tested a 36 MW battery in a pilot project to examine its capability and performance in providing fast regulation service.

One significant regulatory change is the issue of FERC (Federal Energy Regulatory Commission), order 755, which requires undue discrimination in the procurement of frequency regulation in organized wholesale electric markets [1]. Moreover, the order directs that providers of frequency regulation receive just and reasonable and not unduly discriminatory or preferential rates.

This chapter will discuss recent efforts in North America using energy storages for regulation services and its associated challenges, solutions, and issues.

2 Overview of Secondary Regulation Control

Maintaining frequency at its target value is critical to operating a power system. The overall task of controlling frequency is organized in three levels in North America; namely, primary, secondary, and tertiary frequency control. Among them, the secondary frequency control is used to maintain a balance between system load, interchange, and generation and to provide frequency support with certain precision dictated by its control performance standards (CPS). The balance is achieved by varying outputs of regulation generation through automatic generation control (AGC)¹ every several seconds. In addition to machines, other emerging technologies may also provide such ancillary service, e.g., load and energy storage.

To successfully maintain balance between conventional generation, load, wind and solar generation, and interchange within their control areas, balancing authorities (BAs) in the United States need to maintain sufficient balancing reserves and execute proper procedures. Reserves and the balancing effort must be sufficient to compensate for random minute-by-minute load and wind/solar generation changes. A deficient characteristic of these reserves (such as insufficient ramping capability, system inertia, or frequency response) could result in interconnection frequency deviations, transmission system violations, stability problems, and other reliability and control performance problems. The balancing services and balancing reserves are expensive, and each BA tries to minimize these requirements without compromising system reliability and control performance. BAs that do not meet the requirements jeopardize the system reliability, and are thus penalized financially.

Various metrics (e.g., CPS1 and CPS2, balancing authority ACE (area control error) limit (BAAL), frequency deviation limits) are introduced by North American Electric Reliability Corporation as mandatory standards to check how well a BA balances its generation against load, wind and solar generation, and interchange.

2.2 Benefits of Fast-responding Energy Storage

The idea that dispatching faster responding resources will reduce the costs to procure regulation was demonstrated in a study performance by the Pacific Northwest National Laboratory.² In this study, it was concluded that batteries or flywheels could be as much as 17 times more effective than conventional ramp-limited regulation resources because of how quickly and accurately it responds to a system imbalance. If California ISO dispatched fast-responding regulation resources, it could reduce its regulation procurement by as much as 40%.

An example of actual regulation responses from a synchronous unit is given in Figure 4.1. The actual response from a conventional unit was less than 50% of the expected response. It clearly shows that this conventional unit cannot follow the control signal precisely because of its ramp-rate constraint. Since this slow-ramping resource also cannot switch directions quickly, it still provides negative regulation in a direction that is counterproductive to the needs of the grid. In contrast, energy storages, which are more flexible and can ramp more quickly, will reach their dispatch target faster and more closely.

Another study was performed by PJM Interconnection to examine the regulation requirements assuming different-sized energy storage, in which 12 study days were chosen throughout a year to capture differing load and generation patterns. The model in this analysis used existing PJM system resources, dispatch patterns, regulation capability, loads, and interchange. The study covers an entire 24-hour period to identify the inter-dependency between fast resources and traditional regulation mixture. In this model, the fast filter of the raw ACE, which is called dynamic regulation signal (REGD), is sent to energy storages to fully utilize their fast-ramping capability. It was found that the energy storage could help to reduce the regulation requirements without compromising the reliability performance, as shown in Figure 4.2. The points along the contour have similar reliability scores when the ratio between energy-storage and regulation requirement varies (the higher the CPS1 score, the better the secondary frequency regulation performance). The general observation is that the reliability improves as the energy storage’s size increases. However, oversized energy storage could be harmful to the system reliability and thus increases the regulation requirement. Based on the current business practice, the on-peak regulation requirement for PJM was 778 MW, which corresponds to the 1% regulation requirement in Figure 4.2.

The effectiveness of energy storage can also be measured in terms of the ratio between the capacity provided by the fast-moving resource and the equivalent capacity of a traditional resource that could achieve the same reliability performance. This ratio is also referred to as the benefits factor. Using the data from October 2012 through September 2013, the benefits factor can be calculated, as shown in Figure 4.3, which is high at the low penetration of energy storage and becomes diminished as the size of the energy storage increases. This clearly shows that oversized energy storage would not add value to the grid secondary frequency regulation. The benefit factor in Figure 4.3 is used in the market 1) to prevent over-procurement of regulation MWs without consideration of their speed or accuracy, and 2) to prevent over-procuring too many fast-moving resources and facing system instability because of lack of duration or over-shooting of the control signal.

3 Procurement of Regulation Services and Compensation

The ISO uses regulation for system balancing to manage the differences between generating units’ responses to dispatch instructions and actual load within two consecutive time intervals of security-constrained economic dispatch (SCED) [13]. In general, the ISO procures regulation up- and regulation-down capacity separately. In the day-ahead market, the ISO procures 100% of regulation needs in hourly intervals. If additional regulation requirements arise in real-time, the ISO will procure incremental regulation up and regulation down during real-time unit commitment or through supplemental ancillary service markets. Because net-load forecast uncertainty and the minute-by-minute variations largely drive the incremental need for regulation, the same regulation requirement is not actually needed in each hour of the day. To further improve the efficiency of regulation procurement, the ISO procures regulation on a variable basis by hour. If energy-storage resources are present, they can submit their bid into the ancillary service market as traditional generators do. However, depending on the different market structure, the capacity of energy storage that can be awarded may be limited by a cap in MWs (in ERCOT) or by a maximum percentage (in PJM) due to their short-term responses.

Traditionally, the ISO currently co-optimizes energy and ancillary services when determining regulation-capacity awards and market-clearing prices. Prior to FERC order 755, most ISO markets paid regulation providers a “capacity payment” for their available regulation capacity provided on an hourly base. Regulation capacity price included capacity offer cost plus estimated lost opportunity cost (“LOC”). Under these market operations, payments are not tied to resource performance, i.e., resources are paid the same regardless of whether they are used, and how accurately they responded. Fast (energy storage) and slow (traditional generators) resources are paid the same despite the fact that fast resources are more effective in secondary frequency regulation.

On October 20, 2011, the FERC issued new rules for compensating frequency regulation resources, i.e., “pay-for-performance,” since current compensation for frequency regulation is considered to be unjust, unreasonable, and unduly discriminatory. This order acknowledges the inherently greater amount of frequency regulation service being provided by faster-ramping resources, and requires Independent System Operators (ISOs) or Regional Transmission Organizations (RTOs) to pay regulation resources based on the actual amount of regulation service provided (i.e., speed and accuracy). FERC order 755 directs two-part payments, i.e., the resources must be paid based on 1) the amount of MW (capacity) reserved to provide regulation and 2) the actual amount of service provided during the hour (performance) [1–3].

4 Market-Clearance Processing: PJM as an Example

Under new market design, the regulation requirement includes a two-part clearing: capacity and mileage. Mileage is defined as the absolute change in the AGC instructions. The capacity of the regulation requirement can be determined prior to the operation hours either as a fixed percentage of the projected load or the function of minute-by-minute variation of net load [4], historical CPS1 score, and installed capacity of renewable resources. The mileage requirement could be the capacity requirement multiplied by the system mileage multiplier. The market will determine regulation awards for capacity and mileage, but the mileage award is not financially binding.

In PJM, resources will offer into the regulation market with a Capacity Offer MW, Capacity Offer Price, and a Mileage Offer Price. For each resource, PJM will calculate a performance-adjusted capacity cost by dividing the capacity offer MW by the performance score. Thus, a poorer performing resource will appear more costly to the clearing process, and thus less desirable. Also, the mileage correlation factor, which is a unit-less scalar, is used to equalize the increased mileage service provided by the traditional resource. PJM will calculate the mileage correlation factor as:

$\mathrm{Mileage}\mathrm{Correlation}\mathrm{Factor}=\mathrm{Mileage}\mathrm{of}\mathrm{Offered}\mathrm{Resource}\left(\mathrm{\Delta }\mathrm{MW}/\mathrm{MW}\right)/\mathrm{Mileage}\mathrm{of}\mathrm{Traditional}\mathrm{Resource}\left(\mathrm{\Delta }\mathrm{MW}/\mathrm{MW}\right)$

Thus, a performance-adjusted Mileage Cost is calculated in such a way that it is in proportion to Mileage Offer Price and Capacity Offer, but in the inverse proportion to Performance Score and Mileage Correlation Factor. As a faster-moving resource will provide more movement for the same amount of regulating capacity, the cost-per-MW of ACE correction would appear cheaper to the market. PJM will calculate a Lost Opportunity Cost (LOC) per resource, representing the impact of moving the resource out of energy operating limits to regulation operating limits [5]. The Adjusted Total Offer of the resource would be the sum of Adjusted Capacity Cost, Lost Opportunity Cost, and Adjusted Mileage Cost. PJM will rank the resources for clearing as it does today, in $/MW, as:

$\mathrm{Rank}\mathrm{Order}\$/\mathrm{MW}=\mathrm{Adjusted}\mathrm{Total}\mathrm{Offer}\left(\$\right)/\mathrm{Capacity}\mathrm{Offer}\left(\mathrm{MW}\right)$

PJM will clear the market against two requirements, a Capacity Requirement and a Mileage Requirement. The Capacity Requirement sets the amount of regulating reserves that PJM would need in order to achieve a satisfactory frequency regulation performance (PJM has required a capacity of 1% of the projected load for the on-peak/off-peak operating periods). The market will assign units until the capacity requirement constraint is met, or

$\mathrm{Capacity}\mathrm{Requirement}\mathrm{MW}\le {\displaystyle \sum _{i=0}^n{\mathrm{Capacity}\mathrm{Offer}\mathrm{MW}}_i}$

The Mileage Requirement would be defined as the expected amount of resource movement that PJM believes it would need to absorb short-term RTO ACE deviations caused by frequency error, or other sources. The market will assign units until the mileage requirement constraint is met, or

$\mathrm{Mileage}\mathrm{Requirement}\Delta \mathrm{MW}\le {\displaystyle \sum _{i=0}^n\mathrm{Capacity}\mathrm{Offer}{\mathrm{MW}}_i*\mathrm{Expected}\mathrm{Mileage}\mathrm{of}\mathrm{Offered}\mathrm{Resource}\left(\mathrm{\Delta }\mathrm{MW}/\mathrm{MW}\right)}$

When both constraints are satisfied, the last resource assigned (resource n) acts as the marginal unit that sets the Regulation Market Clearing Prices (RMCP) as:

$\mathrm{Regulation}\mathrm{Market}\mathrm{Clearing}\mathrm{Price}=\mathrm{Adjusted}\mathrm{Total}{\mathrm{Offer}}_{\mathrm{n}}/\mathrm{Capacity}{\mathrm{Offer}}_{\mathrm{n}}$

The PJM’s regulation service is featured as follows:

1) The PJM’s proposed accuracy adjustment will assess a resource’s accuracy in responding to AGC set points.

2) The PJM’s market software will co-optimize energy, ancillary services, and mileage to determine awards and market clearing prices based on the marginal unit’s bid. Determining the mileage price and the regulation capacity price is a two-step approach.

The first step is to calculate the expected cost for each resource. The expected cost is the sum of the capacity bid, lost opportunity costs, and mileage bid. The expected cost is adjusted by the resource’s historical performance.

Second, the mileage price is then set at the highest mileage bid from the resources that were awarded regulation. The regulation capacity price is then calculated by selecting the highest expected cost from the resources that were awarded regulation less the mileage multiplier times the mileage price.

5 Fast-Regulation Service – ERCOT as an Example

In 2013, ERCOT implemented a pilot project using batteries for a fast-responding regulation service (FRRS). FRRS was tested as a separate ancillary service and requires full or calculated partial deployment of a resource’s obligated capacity within 60 cycles of a substantial deviation in system frequency or receipt of an ERCOT deployment signal. This faster-responding regulation service has the potential to increase the reliability of the ERCOT system at a lower total cost to load as compared with solely relying on conventional regulation service. One example of FRRS’s response is shown in Figure 4.4.

To participate in FRRS, the resource must meet the following qualification criteria:

• Resources providing FRRS must be able to follow FRRS signal

• Resources requesting FRRS-up/FRRS-down qualification

1) Must be able to respond to large-frequency decay triggered by loss of generation (or high-frequency events).

2) Resources providing FRRS-up (or FRRS-down) must provide full MW response within 60 cycles after frequency hits 59.91 (or 60.09) Hz trigger.

3) This capability must be demonstrated by the resource entity using high-speed frequency and MW data with resolution no less than 32 samples per second.

4) For the purpose of demonstration, the resource entity may use an off-set of –0.2 (or 0.2) Hz to demonstrate the MW response.

Resources must be separately qualified to provide FRRS-Up (increase in output or reduction in consumption provided during certain defined low-frequency conditions) and/or FRRS-down (reduction in output or increase in consumption provided during certain high-frequency conditions), but are not required to seek qualification for both services.

There are two different ways to deploy FRRS at ERCOT according to the following deployment logic. ERCOT has established the frequencies at which a resource will be obligated to deploy due to receipt of a dispatch instruction or due to automatic response.

First, FRRS should be able to be deployed by dispatch instructions. Each Resource must provide 100% of its deployed capacity within 60 cycles of receiving a Dispatch Instruction from ERCOT. ERCOT may deploy FRRS resources for up to two minutes when frequency reaches a deviation of more than +/– .03 Hz from the 60 Hz nominal system frequency (see Figure 4.5).

If the frequency deviation increases to more than +/– .05 Hz within the same frequency excursion (for purposes of this pilot, a single excursion is deemed to end when frequency deviation from 60 Hz decreases to +/– .01 Hz or less at any point and does not increase beyond .01Hz in the same direction of deviation within 12 seconds of reaching +/– .01 Hz, or when frequency deviation exceeds +/– .01 Hz in the direction opposite the excursion for which FRRS had been deployed), then ERCOT may require deployment for an additional two minutes.

If frequency deviation further increases to more than +/– .09 Hz within the same frequency excursion, ERCOT may require deployment for an additional two minutes. If the deployment obligation is extended due to a frequency excursion reaching a greater frequency deviation threshold, any deployment time remaining under the previous threshold will not be added to the deployment. For purposes of FRRS deployed by Dispatch Instruction, ERCOT will not require additional deployment if frequency deviation decreases to a lower threshold before returning to a threshold already reached during the same deployment.

Second, the FRRS should be able to be deployed by trigger frequency. Each resource must deploy 100% of its obligated capacity within 60 cycles of a frequency deviation of more than +/- .09 Hz from the 60 Hz nominal system frequency. ERCOT will issue a Dispatch Instruction, but this will be used only for the purpose of determining when to terminate the deployment. Each Resource shall remain deployed until recalled by ERCOT.

ERCOT shall initiate termination of any deployment whenever frequency deviation is equal to or less than +/– .01 Hz for more than 12 continuous seconds, whenever frequency deviation exceeds +/- .01 Hz in the direction opposite the excursion for which FRRS was previously deployed, or until the maximum deployment time (as described above) has been reached. In all cases, ERCOT will release resources from deployment through a Dispatch Instruction in three steps. In each step, ERCOT will release an additional 1/3 of the originally obligated capacity for each resource by sending a deployment instruction reflecting the reduced obligation. Each step will follow the previous step by no fewer than 12 seconds.

During any hour in which a Resource is obligated to provide FRRS, the resource is subject to deployment by either an ERCOT Dispatch Instruction or by automatic response to a trigger frequency independent of an ERCOT Dispatch Instruction. When required to deploy, a Resource must meet two performance criteria. First, a Resource must deploy within 60 cycles of the Dispatch Instruction or triggering frequency. Second, a Resource must provide 95% to 110% of the obligated capacity during the entire duration of the deployment. Failure of either performance criterion during any deployment for 30% or more of the deployments during any hour shall be deemed failure of the Resource’s obligation for that hour.

ERCOT will conduct a weekly commitment of FRRS capacity. ERCOT will separately commit a maximum of 65 MW for FRRS-Up capacity, and a maximum of 35 MW for FRRS-Down capacity for each hour in the upcoming week. The cost for pilot MWs procured is paid for similar to other ancillary services. Adjustments can be made depending on actual availability and performance. The average cost for REG-Up over the last 6 months has been about $9/MW and for REG-DN has been about $6/MW. The estimated cost for 6 months of FRRS procurement from pilot resources is estimated to be about $3.4 Million.

The benefits of FRRS include 1) promptly arrest frequency decay during unit trips; 2) diminishing use of traditional regulation; 3) reduction in regulation capacity procurement; and 4) better frequency control at a lower overall cost. As seen in Figure 4.6, the simulation result shows that a 36-MW FRRS can improve ERCOT’s ability to arrest frequency decay during unit trips.

6 Summary

The energy-storage resources could provide fast-regulation service for a grid since they can respond to the AGC dispatch instructions more accurately and closely than the fuel-type power plants. The deployment of a properly sized energy storage could improve both the market operational efficiency and the reliability. This is very helpful to accommodate the minute-by-minute fluctuations caused by a high penetration of renewable resources. However, new regulatory changes and market rules should be introduced in order to fully unlock the value from the energy-storage resources. One example of these regulatory changes in North America is the FERC order 755, which requires undue discrimination in the procurement of frequency regulation in organized wholesale electric markets. This has greatly helped to open up the competition of a regulation service market to the energy-storage resources.

This chapter discussed two examples of using energy storage for regulation service in the wholesale electric markets: PJM and ERCOT. Various aspects of these energy-storage applications have been addressed, from the procurement procedure, deployment logic, and market settlement. These examples demonstrated that the ideas for energy-storage applications are still evolving and noticeable differences can be found among different regions on how the market rules could be designed to operate the energy-storage resources.