A Symbiotic Role for Plug-in Hybrid Electric Vehicles in an Electric Delivery System

Transcription

1 Draft 6/16/09 A Symbiotic Role for Plug-in Hybrid Electric Vehicles in an Electric Delivery System by Tim Mount*, Alberto Lamadrid, Surin Maneevijit, Bob Thomas, Max Zhang and Ray Zimmerman Cornell University Abstract The introduction of Plug-in Hybrid Electric Vehicles (PHEV) as replacements for conventional gasoline vehicles has the potential to change future patterns of energy use dramatically. By substituting efficient electric motors for inefficient internal combustion engines, the overall energy efficiency of transportation can be improved and our growing dependence on imported oil reduced. However, to get the full environmental benefits from PHEVs, the electricity used to charge the batteries should be generated from renewable sources of energy such as wind turbines. Earlier research has shown that adding wind capacity to a network can lower the total annual operating cost of meeting the annual pattern of loads. At the same time, the variability of wind generation and the need for higher levels of reserve generating capacity to maintain reliability standards impose additional costs on the system that should not be ignored. In particular, the lower annual earnings of conventional generators when more wind capacity is installed imply that the amount of missing money for these generators will be larger. Although this missing money can be provided through Forward Capacity Markets, for example, the important implication for regulators is that the cost of each MW of peak system load is now much higher. The potential symbiotic role for PHEVs is 1) to offset the variability of wind generation using smart (controllable) chargers for the batteries, 2) to reduce the system peak load by using Vehicle to Grid (V2G) capabilities. These potential benefits will be illustrated in a case study using a test network and the SuperOPF. An important feature of the SuperOPF is that the amount of conventional generating capacity needed to maintain Operating Reliability is determined exogenously, and as a result, it is possible to determine the net social benefit of relying more on an intermittent source of generation, such as wind capacity, that lowers production costs but increases the cost of maintaining System Adequacy. The capabilities of the SuperOPF provide a consistent economic framework for evaluating Operating Reliability in real-time markets and System Adequacy for planning purposes. In the latter type of application, it is possible to consider the capital and investment costs as well as the operating costs. Basically, a socially beneficial investment requires that the reductions in the total annual costs of the existing system should be larger than the annualized cost of financing the addition of, for example, wind generation to a network. The scenarios considered make it possible to determine 1) the amount of conventional generating capacity needed to meet the peak system load and maintain reliability, 2) the amount of missing money paid to generators through a Capacity Market, 3) changes in the congestion rents for transmission that are collected by the system operator, and finally, 4) the total annual system costs paid by customers in the Wholesale Market and indirectly as missing money. The results show that total system costs increase when wind capacity is added unless storage is used to mitigate the variability of wind generation. Additional scenarios are being run to show how PHEVs can be used to mitigate the variability of wind generation and, more importantly, lower the system peak load. * Corresponding author, tdm2@cornell.edu.

2 A Symbiotic Role for Plug-in Hybrid Electric Vehicles in an Electric Delivery System 1. Introduction The new federal administration of President Obama has recognized the importance of addressing climate change by establishing a new energy plan with the ambitious goal of reducing the nation s emissions of greenhouse gases by 80% by the year 2050 (Source: Reaching this objective will imply substituting renewable sources of energy for the conventional fossil fuels used for generating electricity 1 and by electrifying transportation. Adapting lifestyles and production to a new low-carbon future will be a disruptive transformation of the economy, and most of the new non-fossil sources of energy will involve the electric delivery system either directly (e.g. generation from wind) or indirectly (e.g. charging batteries for plug-in hybrid vehicles). Energy from liquid biofuels is one exception. Although the generation of electricity from fossil fuels is relatively inefficient, electricity actually delivers services very efficiently. Hence, electricity generated from renewable sources of energy is intrinsically efficient because it avoids the losses that occur in a typical thermal power plant. In addition, using electric motors for transportation will avoid the losses due to the inherent inefficiency of internal combustion engines. Basically, one unit of electricity from a renewable source replaces three units of energy from coal in a typical power plant and over five units from petroleum in a typical gasoline engine. In a low-carbon economy, electricity will be the dominant way to deliver most energy services to customers. Unfortunately, some of the most promising renewable sources of energy are intermittent (wind and solar), and as a result, policies that simply focus on increasing the penetration of these sources, such as Renewable Portfolio Standards (RPS), may undermine the reliability of the electric delivery system. Major blackouts attributable to unexpected drops in wind generation have already occurred in Texas and Germany. If 1 Carbon Capture and Sequestration (CCS) presents one way to continue using fossil fuels to generate electricity and still meet the goals of climate change. In the European plan for reducing emissions of greenhouse gases, Zero Emissions Power (ZEP) is one of the largest components of the plan. 2

3 this type of disruption of electricity supply becomes more widespread, it is likely to lead to a backlash against renewables by system operators and the public. Hence, policies for a successful transition to a low-carbon economy should ensure that the reliability standards for supplying electricity are maintained and the uncertainty associated with wind and solar generation is mitigated effectively. In a recent study initiated by the US Department of Energy (DOE) 2, the effects of increasing dependence on wind energy to 20% of the total generation of electricity by 2030 are evaluated. The study is relatively positive about this scenario and states: Until recently, concerns had been prevalent in the electric utility sector about the difficulty and costs of dealing with the variability and uncertainty of energy production from wind plants and other weather-driven renewable technologies. But utility engineers in some parts of the United States now have extensive experience with wind plant impacts, and their analyses of these impacts have helped to reduce these concerns wind s variability is being accommodated, and given optimistic assumptions, studies suggest the cost impact could be as little as the current level - 10% or less of the value of the wind energy generated. The DOE study focuses on the initial capital cost of installing the wind capacity and upgrading the transmission network and compares these costs to the lower operating costs when wind energy displaces fossil fuels. The objective of this report is to show that there are other, hidden costs of wind power associated with the need to maintain the Financial Adequacy of conventional generating capacity. Since wind capacity is essentially a non-dispatchable source of energy, it contributes relatively little capacity for meeting the reliability standard of Generation Adequacy. Nevertheless, wind generation, when it is available, is essentially free and it displaces most conventional sources of generation. As a result, the capacity factors of conventional generators are typically reduced when wind capacity is added. This happens even though the total amount of conventional capacity needed to maintain reliability may actually increase. Consequently, these conditions lead to increasing amounts of missing money for generators that are generally paid through some form of Capacity Market in most deregulated markets in the US 3. For example, 2 20% Wind Energy by 2030, US DOE, May Some energy only markets do not have a Capacity Market, and some other way of maintaining the financial viability of conventional generators, such as tolerating high, scarcity prices, is used. 3

4 generating units in New York City have typically been paid over $100,000/MW/year. 4 This paper argues that Financial Adequacy should be treated as an additional criterion for planning purposes that would complement the standard engineering criterion of maintaining System Adequacy. The case study presented shows that the total system costs charged to customers increase if a new wind farm replaces an existing coal unit on a network. With the wind farm in place, the increase in missing money is larger than the decrease in total operating costs in the Wholesale Market. For an investment to be economically viable from the perspective of a social planner, the total annual cost of maintaining the existing system must go down and this decease in cost must be bigger than the annualized cost of financing the investment. If some form of storage capability, such as a battery, mitigates the variability of wind generation, the total annual cost of the existing system does decrease. The battery charges when the wind speed is higher than forecasted and discharges when it is lower than the forecast, and in this way, the effective wind generation is smoothed over time. As a result, there is an effective floor on the amount of generation from wind capacity when the indirect generation from discharging the battery is included. The presence of this floor reduces the total amount of conventional generating capacity that is needed to meet the peak system load and maintain System Adequacy, and as a result, the amount of missing money is also reduced. In addition, the total amount of wind that is spilled (i.e. wasted) is reduced when batteries are coupled with the wind farm 5. The introduction of Plug-in Hybrid Electric Vehicles (PHEV) can be viewed, from the perspective of a system operator, as adding to load when the batteries are being charged, and potentially, could be used to reduce peak load if the batteries are discharged using Vehicle to Grid (V2G) capabilities and not just the Grid to Vehicle (G2V) capabilities needed for storage. Typically, G2V charging would occur at homes overnight and V2G discharging would occur during the day at work. The conclusion from the analysis is that there are likely to be substantial economic benefits from having 4 Figures 10 and 11 on p. 15 of 2007 State of the Market Report, New York ISO. 5 In one of the scenarios, must-take contracts for wind generation are evaluated by making the cost of not using available wind generation expensive, and in this scenario, the total annual system cost of maintaining System Adequacy is substantially higher than the corresponding cost without the wind farm. 4

5 smart G2G and V2G capabilities that couple the charging and discharging of PHEVs to actual wind generation and system needs. PHEVs are simply mobile batteries that can be transformed into another form of dispatchable load. These batteries will be relatively inefficient in terms of the round-trip use of energy because the main purpose of owning a PHEV is for transportation. On the other hand, the capital cost of buying a PHEV is already covered and should not be charged to the network. The only new cost that should be added to total system costs is the investment needed replace the charge-at-will capabilities of the PHEVs with the ability to control charging and discharging in response to system conditions on the network. The overall objectives of the paper are to demonstrate through a case study 1) why Financial Adequacy is an important concept that should be considered by system planners, 2) why the social value of storage and controllable load increases when intermittent sources of generation are added to a network, and 3) how PHEVs can complement the reliability of operating a network and lower the total annual system cost. The structure of the paper includes four additional sections. Section 2 describes the SuperOPF and shows how this analytical framework relates to the NERC standards for Operating Reliability and System Adequacy. Section 3 presents the specifications for a case study that considers the effects of replacing a coal unit by a large wind farm with three times the installed capacity of the coal unit. Since the wind farm causes more congestion on the network when the wind blows, a series of additional scenarios show the effects of upgrading the capacity of a tie line to reduce this congestion. Since congestion rents on the network are treated as one source of income for transmission owners, reducing these rents implies that there will be more missing money that must be paid by customers outside the Wholesale Market to ensure that the transmission owners are financially viable. The missing money for both generators and transmission owners contributes to the total annual system cost of maintaining System Adequacy and Financial Adequacy. The results for all of the scenarios are presented in Section 4 and the implications of these results for adding 1) charge-at-will PHEVS, and 2) controllable PHEVs to the network are discussed in Section 5. The overall conclusions of the analysis and some suggestions for regulators are summarized in Section 6. 5

6 2. Reliability Standards and the SuperOPF Federal legislators have formally recognized the importance of maintaining Operating Reliability in the Energy Policy Act of 2005 (EPACT05), and the major effect of this legislation is to give the Federal Energy Regulatory Commission (FERC) the overall authority to enforce reliability standards throughout the Eastern and Western Inter-Connections (see FERC [5]). The North-American Electric Reliability Corporation (NERC) has been appointed by FERC as the new Electric Reliability Organization (ERO), and NERC has been given the responsibility to specify explicit standards for reliability. A draft set of reliability standards were released for public comment from April 23 rd to May, 25 th, This report contains over 1000 pages and covers a wide range of different topics. Although the important issue of System Adequacy is referenced many times, the report recognizes the important feature and the major complication of how the North American Bulk Power Network is governed. There are many layers of governance, and in general, State regulators determine the rules for maintaining System Adequacy. For example, the NERC report refers in a numerous places to: Any reserve margin or resource adequacy requirements for loads within the Transmission Service Provider s area established by other entities, such as municipalities, state commissions, regional transmission organizations, independent system operators, Regional Reliability Organizations, or regional entities. Given the existing structure of governance of the industry, the NERC report is effectively limited to specifying comprehensive rules for Operating Reliability, and the responsibility of determining how to comply with these rules and how to maintain System Reliability rests with regulators. To date, the FERC has allowed a surprisingly wide range of different approaches to the maintenance of System Adequacy by different State regulatory agencies. As a result, there is no equivalent to the NERC report on reliability that summarizes a set of regulatory guidelines for maintaining System Adequacy. Furthermore, there is no clear winner to date in the ongoing real-time 6 NERC, Reliability Standards for the Bulk Electric Systems of North America, 2009,

7 experiments that are being conducted in different States 7. In fact, there is no general agreement about the whether the creation of new Regional Transmission Organizations (RTO) has benefitted the public overall. A recent call has been made by the Government Accountability Office (GAO) to the FERC to establish standard measures that track the performance of RTO operations and markets, and this report states Experts and industry participants are divided on the benefits of RTO markets and their effects on consumer electricity prices, generator efficiency and infrastructure investment. 8 time for an independent evaluation of the performance of RTO markets with full access to all financial information is long overdue. The NERC uses the following two concepts to evaluate the reliability of the bulk electric supply system: 9 1. Adequacy The ability of the electric system to supply the aggregate electrical demand and energy requirements of customers at all times, taking into account scheduled and reasonably expected unscheduled outages of system elements. 2. Operating Reliability The ability of the electric system to withstand sudden disturbances such as electric short circuits or unanticipated failure of system elements. The To simplify the concept of Operating Reliability, it is convenient to adopt a single measure, and the traditional NERC standard of one day in ten years for the Loss of Load Expectation (LOLE) is still treated by many regulators as the appropriate measure for the reliability of the bulk transmission system (i.e., this does not include outages of the local distribution systems caused, for example, by falling tree limbs and ice storms). Adequacy implies that past investments in the capacity of the electric delivery system must be sufficient to make the real-time operations meet the reliability standards. As a result, the Adequacy standard has important economic and financial implications that 7 In the development of new products, engineering practices require that numerous tests of the products are performed before they are made available to the public. In contrast, regulators in the USA have been willing to adopt major changes in the ownership of the electric delivery system and to introduce new types of Wholesale Market with virtually no testing or formal justification demonstrating that the public would be better off. 8 Pages 46 and 59 in Electricity Restructuring: FERC Could Take Additional Steps to Analyze Regional Transmission Organizations Benefits and Performance, USGAO , September North-American Electric Reliability Corporation (NERC), 2007 Long-Term Reliability Assessment , Princeton NJ, October

8 should be addressed by regulators in a more systematic and transparent way. In this paper, it is argued that a new criterion of Financial Adequacy should be treated as one of the standard measures used by system planners to evaluate the desirability of proposed changes to system capacity. Chen et al. 10 have proposed an alternative way to determine the optimal dispatch and nodal prices in an energy-reserve market using co-optimization (CO- OPT). The proposed objective function minimizes the total expected cost (the combined production costs of energy and reserves) for a base case (intact system) and a specified set of credible contingencies (line-out, unit-lost, and high load) with their corresponding probabilities of occurring. Using CO-OPT, the optimal pattern of reserves is determined endogenously and it adjusts to changes in the physical and market conditions of the network. For example, the amount of reserves needed typically increases after the addition of an intermittent source of generation from a wind farm. This framework corresponds to using a conventional n 1 contingency criterion to maintain Operating Reliability. In practice, the number of contingencies that affect the optimal dispatch is much smaller than the total number of contingencies. In other words, by covering a relatively small subset of critical contingencies, all of the remaining contingencies in the set can be covered without shedding load. In the SuperOPF, the CO-OPT criterion is modified to include the cost of Load- Not-Served (LNS), and it also distinguishes between positive and negative reserves for both real and reactive power. A high Value Of Lost Load (VOLL) is specified as the price of LNS. In a conventional Security-Constrained OPF used by most System Operators, the n-1 contingencies are treated as hard constraints rather than as economic constraints as they are in the SuperOPF. 11 Increasing the stress on a network by, for example, increasing the peak system load over a planning horizon eventually causes load shedding, and typically, load shedding occurs first in one or more of the contingencies. Since the expected cost of LNS in a contingency is determined by multiplying a large VOLL by a small probability, the overall effect on the total expected system cost may be 10 Chen, J., T. D. Mount, J. S. Thorp and R. J. Thomas, Location-based scheduling and pricing for energy and reserves: a responsive reserve market proposal, Decision Support Systems, 40, pp , October A hard constraint is equivalent to specifying the VOLL as plus infinity. 8

9 modest. From a social planner s perspective, the standard of one day in ten years for the LOLE should correspond to equating a reduction in the expected annual cost of operating the system, including changes in the expected cost of LNS, with the annual cost of making an investment in additional capacity. Using the SuperOPF, the System Operator determines the optimal dispatch for energy and reserves for the base case (intact system, k = 0) and for K different contingencies by minimizing the expected cost of meeting load in the (K + 1) states of the system as follows (additional components for reactive power and dynamic VAr are included in the SuperOPF but not shown in the objective function below): min p C k G i (G ik ) + INC i (G ik G i0 ) + + DEC i (G i0 G ik ) + G ik, R ik,lns jk K J K k =0 I i =1 [ ] + p k VOLL j LNS jk + [ C Ri (R + i ) + C Ri (R i )] k =0 j =1 I i =1 Subject to meeting LOAD and all of the nonlinear AC CONSTRAINTS of the network, where k = 0, 1,..,K is a CONTINGENCY i = 1,2,..,I is a GENERATOR j = 1,2,..,J is a LOAD p k is the probability of Contingency k occurring G i is the quantity generated in MWh C G (G i ) is the COST of generating G i MWh INC(G ik G i0 ) + > 0 is the cost of increasing generation from the base DEC(G i0 G ik ) + > 0 is the cost of decreasing generation from the base VOLL j is the VALUE OF LOST LOAD LNS jk is the LOAD NOT SERVED MWh R + i = (Max[G ik ] G i0 ) + < Ramp i is the quantity of UP reserves in MW C R (R + i ) is the COST of providing R + i MW of UP spinning RESERVES R i - = (G i0 Min[G ik ]) + < Ramp i is the quantity of DOWN reserves in MW C R (R i - ) is the COST of providing R i - MW of DOWN spinning RESERVES 9

10 The objective function above represents a simplification of the objective function used in the SuperOPF because the reactive components of generation, reserves and load are not all identified explicitly, and for the remainder of the paper, the discussion only refers to real power (energy) even though reactive power is actually included in the case study 12. It would also be straightforward to include more flexible forms of load response, such as interruptible contracts and price responsive demand, but capabilities of this type were not used for this case study. The reserve capacity for each generator is defined as the difference between the maximum (minimum) of the K +1 optimal levels of dispatch (which may be less than the true physical maximum) and the optimal level of dispatch for the intact system (k = 0). The maximum (minimum) dispatched capacity for every generator is realized for generating real energy in at least one contingency. The amount of reserve capacity is limited for each generator by a specified Ramp Rate. This ensures that each generator has the physical capacity to meet the optimal dispatch in all of the specified contingencies. The Max[G ik ] over the K + 1 contingencies for the system peak load determines the amount of capacity needed for System Adequacy. Adding wind capacity to a network with no mitigation of the variability of the amounts generated by wind typically increases both Up and Down Reserves, and as a result, the total amount of conventional, non-wind capacity needed for System Adequacy may actually increase. Paying the full cost of this extra capacity is potentially a major additional system cost of adding wind capacity, and in the case study, reducing the need for additional capacity is a benefit of using storage batteries to mitigate the variability of wind generation. The objective function used in the SuperOPF implies that the optimum quantities of spinning reserves for k = 0 are contracted and paid ahead of real time in, for example, a day-ahead market. In this case study, the generators are compensated for the actual energy generated using real-time prices but the payments could be made in two parts. One part paying the contracted dispatch for k = 0 using day-ahead prices, and the other part paying the difference between the actual dispatch and the contracted dispatch using real-time prices. In addition, extra payments for shifting generation away from the base 12 It is assumed that each generator has contracted in advance to provide VArs at no additional cost up to the limits of a specified Capability Curve, and loads are specified to have a constant power factor, implying that the nodal prices cover the cost of both real and reactive power. 10

11 dispatch can also be specified (the INCs and DECs), and these costs can be used to reflect the inefficiencies associated with ramping generators. However, the INCs and DECs are all zero in this case study. The level of reserve capacity for any generator is determined endogenously, and it responds to conditions on the network, such as the pattern of forecasted load. This feature is important for the case study due to the wide range of wind conditions that affect the actual generation from a wind farm and the difficulty in forecasting wind speeds accurately. If the regulated standard of Operating Reliability corresponds to avoiding load shedding in any of the K + 1 contingencies, finding optimal values of LNS jk > 0 is equivalent to violating this reliability standard, and it signals a failure of System Adequacy in a planning application. In practice, such violations would be corrected by increasing the system capacity in some way. Since the VOLL is specified to be very large compared to typical market prices, it is important to note that a major part of the total benefit of many components of the grid comes from avoiding unscheduled load shedding when contingencies occur. The real-time optimization using the SuperOPF treats the actual state of the system as the new base (k = 0) and includes all of the remaining contingencies in the same way as they are used in the day-ahead optimization. This implies that the optimum dispatch in real time still attempts to maintain Operating Reliability. However, if a major failure has already occurred, it may not be possible to meet the load in all situations if a second failure occurs. This would not be a violation of the typical standard of Operating Reliability assuming that the day-ahead dispatch did not shed load in any of the K + 1 contingencies. For example, if the contingencies cover all single failures of equipment, there is no guarantee that the system can accommodate the relatively rare events when two or more failures occur simultaneously. Following any major contingency, bringing the system back into compliance with Operating Reliability would require adding existing resources that were not committed (needed) in the day-ahead dispatch. The key to deriving the economic value of maintaining a given reliability standard is to consider the benefits of not shedding load. In the empirical simulations discussed in 11

12 Section 4, a VOLL of $10,000/MWh is used 13. Consequently, reducing the probability of shedding load by 0.1%, for example, still saves $10/MWh. Some components of a network may only have a positive economic value when contingencies occur because they reduce the amount of LNS 14. These components affect only reliability. Other components, such as a new baseload unit, may reduce the cost of generation when the system is intact and have little affect on reliability. More generally, components will affect both the operating costs for the intact system and the reliability of supply. For an intermittent source such as wind power, there is a fundamental tension between providing an inexpensive source of generation and making the existing network more vulnerable to outages. The solution to this predicament is to add new capabilities to the network that can compensate for the intermittent nature of wind power, such as dispatchable load and storage capacity. Evaluating the net-benefits of a portfolio of assets is the type of problem that can be evaluated effectively using the analytical framework supported by the SuperOPF, and the case study described in the following two Sections illustrates the economic symbiosis that occurs when wind capacity is coupled with storage. The following section describes the characteristics of the 30-bus test network used in the case study and the specifications for the simulations. The basic objective of this analysis is to evaluate the effects of replacing an existing coal unit by a large new wind farm. The initial amounts of installed capacity are sufficient to meet the standard for System Adequacy, and since wind generation is inherently intermittent, the installed capacity of the wind generators is substantially larger than the capacity of the coal unit. In addition, the analysis determines the economic benefit of upgrading a transmission tie line to transfer more wind generation from a remote location to an urban center. 13 Using the costs for a peaking unit specified in Section 4, the current NERC reliability standard of 1 day in 10 years corresponds to a VOLL of $36,762/MWh ( ,000/2.4, based on an operating cost of $95/MWh and an annual capital cost of $88,000/MW for a peaking unit). This high value is actually at the low end of the range of the costs of unexpected outages for different economic sectors (Table 3-3, p.13 in Lawton, Leora, Michael Sullivan, Kent Van Liere, and Aaron Katz and Joseph Eto, A Framework and Review of Customer Outage Costs: Integration and Analysis of Electric Utility Outage Cost Surveys, Office of Electric Transmission and Distribution, U.S. Department of Energy, Washington DC, November 2003). These outage costs range from negligible for Construction to $168,000/MWh for Finance, Insurance and Real Estate, and the average cost for all sectors is $20,000/MWh. 14 Having a generator provide VArs is a typical example because VArs can be provided at no cost within the boundary limits of a capability curve. However, when a generator operates on the capability curve, the nodal price of providing one more MVAr may be much higher than the highest offer for generating real energy. 12

13 3. The Specifications for the Case Study The Test Network The case study is based on a 30-bus test network that has been used extensively in our research to test the performance of different market designs using the MATPOWER platform. The one-line-diagram of this network is shown in Figure 1 below. The 30 nodes and the 39 lines are numbered in Figure 1 and this numbering scheme provides the key to identifying the locations of the specific contingencies described in the following discussion. In addition, the six generators are also identified. The network is divided into three regions, Areas 1 3, and Area 1 represents an urban load center with a large load, a high VOLL and expensive sources of local generation from Generators 1 and 2. The other two regions are rural with relatively small loads, low VOLLs and relatively inexpensive sources of generation from Generators 3 6. Consequently, an economically efficient dispatch uses the inexpensive generation in Areas 2 and 3 to cover the local loads and as much of the loads in Area 1 as possible. Figure 1: A One-Line-Diagram of the 30-Bus Test Network. 13

14 The capacities of the transmission tie lines linking Areas 2 and 3 with Area 1 (Lines 12, 14, 15 and 36) are the limiting factors. Since lines and generators may fail in contingencies, the generators in Area 1 are primarily needed to provide reserve capacity. The general structure of the network poses the same type of problem faced by the system operators and planners in the New York Control Area. Most of the load is in New York City (i.e. Area 1) and the inexpensive sources of baseload capacity (hydro, coal and nuclear) are located upstate (i.e. Areas 2 and 3). For this case study, the peak system load is specified at a level that causes some load shedding in some contingencies, and the expected amount of LNS varies in the different scenarios. This load shedding occurs even though the maximum amount of generating capacity dispatched is lower than the total installed capacity. In other words, the physical limitation of the existing transmission network makes is impossible or excessively expensive to transfer additional real energy from generators with surplus capacity to the locations where it is needed. The installed capacities and locations of the different types of generating units are shown in Table 1 together with their production costs. Table 1: Installed Generating Capacity of the Initial System by Type and Location and the Production Costs Area Nuclear/ Hydro Coal Oil Combined Cycle Gas Gas Turbine Total by Area MW 0 45MW 110MW 2 50MW 70MW MW 3 65MW MW 0 105MW Total by Type 115MW 70MW 65MW 40MW 45MW 335MW Production Cost $5/MWh $25/MWh $95/MWh $55/MWh $80/MWh - The Contingencies Considered The specific contingencies included in the SuperOPF are listed in Table 2. These contingencies include the failures of individual generators and transmission lines, and also the uncertainty about the actual level of load associated with the errors of the forecasted load. When wind capacity is added to the network, this type of uncertainty about actual outcomes is exacerbated, and for this case study, the combined forecasting 14

15 errors for load and wind generation are represented by four possible outcomes (Wind 1-4 in Table 2). Each outcome specifies a level of generation from a wind farm and the corresponding pattern of loads, and each outcome has a specified probability of occurring. This information is summarized in Table 3. Table 2: The Contingencies Used in the Case Study* * The probabilities for Contingencies 0-3 are summarized in Table 3 The probabilities for each one of Contingencies 4-18 is 0.2% For generators and lines, there are only two possible outcomes. The first outcome is to perform as required in the optimum dispatch, and the second is to fail completely. However, the probability of failure is very small (0.2% for each failure in this case study), and as a result, the probability that each piece of equipment will perform as required is 99.8%. Since there are 15 failures identified in Table 2, the expected number of failures is 3 in 100 periods because the individual failures and periods are specified to be statistically independent. In other words, the system is expected to be intact 97% of the time. In contrast to the high probabilities that equipment will not fail, the probability that a forecasted level of wind generation will actually be realized is relatively small. Hence, it is unlikely that anyone of the four different outcomes for wind generation (Wind 1 4) will have a dominant probability. In addition, the range of possible outcomes is large, and as a result, one would expect that the quantity of reserves needed 15

16 to maintain Operating Reliability should increase as more wind capacity is installed. This phenomenon has already been observed in Europe and the levels of operating reserves have increased by almost 10% in some countries. The Realizations of Wind Generation There are three different forecasts of the level of wind generation (high, medium and low), and each forecast has four possible outcomes, summarized in Table 3. With no wind capacity installed, the contingency k = 0 corresponds to the intact system using the forecasted level of load (i.e. the network shown in Figure 1). The analysis that underlies the information presented in Table 3 has three components that are described in a paper by Anderson and Cardell. 15 The first component is a set of time-series data for hourly wind speeds at a specific location (in New England for this case study). The second component is an ARMA model for predicting wind speed (one hour ahead for this case study), and finally, there is a power curve for a wind turbine that converts a given wind speed to the amount of energy generated (wind turbines are also specified as a potential source of positive reactive power in the SuperOPF). Table 3: Specifications of the Wind Contingencies in the Case Study Forecasted Wind Speed LOW (0-5meters/second) MEDIUM (5-13meters/second) HIGH (>13meters/second) Probability of Forecast Occurring 11% 46% 43% Wind Generation (% of Installed MW) Probability of Actual Generation 0% 66% 7% 26% 33% 5% 73% 3% 6% 24% 38% 20% 62% 18% 93% 38% 0% 14% 66% 4% 94% 3% 100% 79% 15 Anderson, Lindsay and Judy Cardell, Reducing the Variability of Wind Power Generation for Participation in Day Ahead Electricity Markets, Proceedings of the 41st Annual Hawaii International Conference on System Sciences, January

17 The data set of observed wind speeds provides the information needed to 1) derive a probability distribution of the wind speed, 2) estimate a forecasting model, and 3) compute the forecasts of wind speed for each hour. The forecasted wind speeds are then assigned to three bins by dividing them into three different ranges (Column 1 in Table 3). For each bin of forecasted wind speeds, the probability of being in a bin is calculated (Column 2). For each bin of forecasted wind speeds, the corresponding observed wind speeds for each bin of forecasts are collected. The next step is to divide each collection of observed wind speeds into four new bins, and then to determine the average wind speed and the corresponding probability (Column 4) for each bin of observed wind speeds. The final step is to convert the average observed wind speeds in the four bins to the corresponding levels of generation using the power curve for a wind turbine. These levels of generation are reported as percentages of the maximum output (rated capacity) of a wind farm (Column 3). Figure 2: The Power Curve for a Vestas 1.8MW Wind Turbine 17

18 The levels of generation reported in Table 3 are derived from the power curve for a Vestas 1.8MW turbine shown in Figure This power curve implies that generation increases from zero for wind speeds above 4 meters/second and reaches a maximum level of generation of 1.8MW at wind speeds above 14 meters/second. However, the turbine cuts out at wind speeds above 25 meters/second to avoid damage to the turbine. The generation from the wind farm in this case study is determined by scaling the generation from a single turbine. This makes the effects of having the turbines cutout at high wind speeds more severe than they would be if the spatial diversity of the turbines were considered. However, the actual characteristics of the generation from a wind farm are very sensitive to the specific location. Scaling the power curve for a single turbine is a convenient simplification for illustrating the importance of determining the level of reserve generating capacity needed to maintain reliability as an endogenous output from the analysis. An important point that underlies the levels of generation in Column 3 of Table 3 is that the ranges of observed wind speeds for each wind forecast (Low, Medium or High) are much larger than the range of the forecasted wind speeds that define each bin. Consequently, the ranges of generation for a given wind forecast are also very large. For a Low wind forecast, the distribution of outcomes is skewed to the right. The modal outcome is zero generation but there is still a small probability that the actual generation from wind will be 73% of the maximum. The outcomes for a Medium wind forecast are more uniform in distribution but have a wide range from 6% to 93% of the maximum. The outcomes for a High wind forecast are the most challenging for System Operators because there is a substantial probability of 14% that the turbines will cut out at very high wind speeds (>25meters/second) to avoid damaging the equipment. Hence, the distribution of wind speeds is bimodal with modes at 0% and 100% of the maximum level of generation Discussed in Anderson, Lindsay and Judy Cardell, Reducing the Variability of Wind Power Generation for Participation in Day Ahead Electricity Markets, Proceedings of the 41st Annual Hawaii International Conference on System Sciences, January These extreme specifications are used on purpose in this case study to provide stress on the network when wind capacity is installed. In practice, there may be a substantial amount of smoothing of total wind generation from wind farms located in different regions of a network. 18

19 The Wind Scenarios For this case study, the wind farm is located at the same node as Generator 6, and before any wind capacity is installed, the generating capacity at this node represents the combination of a coal plant with a capacity of 35MW and a nuclear plant with a capacity of 30MW. The rationale for the case study is that policy makers are interested in evaluating the effects of replacing some coal capacity by wind capacity as a way to address growing public concerns about climate change, and the 35MW coal unit at Generator 6 is replaced by 105MW of new wind capacity. The justification for choosing 105MW of wind capacity is that it approximates the current procedures used by regulators to de-rate the amount of installed wind capacity to the expected amount of capacity that will be available to meet the peak system load. A coal plant may have an availability rating of over 90% but a wind farm is likely to have a rating of less than 20%. Hence, the implicit use of 33% for the relative availability of wind capacity compared to a coal unit in the simulation is relatively high, and this value was chosen to avoid having the wind farm overwhelm the capacity of the network. In spite of adopting this strategy, it is still does not pay to use all of the potential generation from the wind farm even if this source of generation is submitted into the wholesale auction at an offer price of zero. The following four different Cases for wind generation are considered: 1) NO Wind; with a 35 MW coal unit installed at Generator 6, 2) NORMAL Wind; with the coal unit replaced by 105MW of wind capacity, using the wind speed specifications in Table 3 and zero for the offer price in the wholesale auction, 3) NICE Wind; the same as Case 2 with a different set of specifications for the realizations of wind generation (shown in Table 4) that represent the net effect of coupling wind generation with storage batteries 18, 4) NASTY Wind; the same as Case 2 with the offer price in the wholesale auction set to -$1,500/MWh to force the acceptance of 18 The rationale for Case 3 is that the batteries are charged when the realized wind speed is high and they are discharged when the wind speed is low, and most importantly, when the wind turbines cutout for safety reasons at very high wind speeds. 19

20 wind generation in the auction and represent a Must-Take form of contract between the wind farm and the System Operator. In addition, the same four Cases are rerun after making an upgrade to Transmission Line L15 in Figure 1. This upgrade doubles the transfer capacity of the main tie line linking the wind farm in Area 2 to the urban center in Area 1. These Cases are referred to as Case 1UP, Case 2UP, etc. Table 4: Specifications of the Wind Contingencies for NICE Wind* Forecasted Wind Speed LOW (0-5meters/second) MEDIUM (5-13meters/second) HIGH (>13meters/second) Probability of Forecast Occurring 11% 46% 43% Wind Generation for NICE Wind (% of Installed MW) Probability of Actual Generation 35% 66% 35% 26% 35% 5% 38% 3% 41% 24% 55% 20% 55% 18% 58% 38% 35% 14% 70% 4% 70% 3% 70% 79% * The realized levels of net wind generation in Column 3 are different from the levels for NORMAL and NASTY Wind shown in Table 3, and they represent the effect of coupling batteries with the wind farm to reduce the range of outcomes for a given forecast of wind speed and to provide a floor > 0 to the amount generated. Annual Load Duration Curves An important goal of the analysis is to feature the concept of Financial Adequacy, and for this reason, it is necessary to simulate the annual earnings of individual generators. Hourly system loads are ranked from high to low to form an annual load duration curve. The same ranking is used to form the corresponding load duration curves for Area 1 and for Areas 2 and 3 in Figure 1 using scaled data for New York City and Long Island in 2005 to represent Area 1 and scaled data for Upstate New York in 2005 to represent Areas 2 and 3. Each load duration curve is divided into 100 bins with the average load for 87 hours in each bin. All loads within Area 1 are scaled by the same factor to represent the different loads in the 100 bins and a similar procedure is used to scale all of the loads in Areas 2 and 3. The two load duration curves are shown in Figure 20

21 3, and they illustrate the effects of the relatively high demand for air conditioning in Area 1, the urban center. The high loads are higher and the low loads are lower in Area 1 than they are in Areas The peak system load in the case study is 223MW and the lowest load is 103MW. Since the amount of installed generating capacity is 335MW (see Table 1), there is plenty of generating capacity to meet the load and System Adequacy is not likely to be a major issue. However, one result of the case study is to determine the minimum amount of generating capacity that is really needed for Generation Adequacy in each scenario. To determine the net social benefits for a specific scenario, it is assumed that it is only necessary to consider the costs of maintaining the Financial Adequacy for the minimum amount of conventional capacity needed to maintain the Physical Adequacy of the network. Figure 3: Annual Load Duration Curves for Two Regions 21

22 For each one of the 100 load bins, the corresponding probabilities of having a High, Medium or Low forecast of the wind speed are determined (The probabilities of the three wind forecasts shown in Column 2 of Tables 3 and 4 are the average annual values). The probabilities for the 100 bins are shown in Figure 3 and they imply that, in general, the forecasted wind speeds are slightly negatively correlated with the total system loads in the 100 bins 19. The most important feature for evaluating Financial Adequacy is that the probability of having a Low wind forecast is over 10% at the System Peak Load, and this has a direct effect on the total amount of capacity needed to maintain System Adequacy 20. For all intents and purposes, there must be enough conventional generating capacity installed to meet the peak load and cover the contingencies with virtually no wind generation. In each one of the different scenarios, the contingencies corresponding to equipment failures in Table 2 are always specified for the lowest realization of wind speed for any one of the three wind forecasts. Figure 4: The Probabilities of a High, Medium or Low Forecast of Wind Speed 19 At the highest loads, the forecasts are positively related with the system load but there is no reason to expect that this specific relationship will hold for wind farms in other locations. 20 The realizations of wind generation and the corresponding probabilities for the Low wind forecast in Table 3 imply that the expected wind generation is 5.66% of the maximum at the peak system load. 22

23 The Amount of Missing Money The final component of the analytical framework is to describe how much missing money is required by generators above their annual earnings in the Wholesale Market to ensure that they are financially viable. This is the main implication of requiring that an electric delivery system should maintain Financial Adequacy as well as Physical Adequacy. In addition, changes in the amount of congestion rents collected by the System Operator in the Wholesale Market affect how much additional money is needed to pay transmission owners a regulated rate of return on their capital investment in transmission and distribution. Even though the total payment made to transmission owners is the same in all scenarios, the proportion of this amount coming from congestion rents is determined in the analysis. Under a regulated regime, the rates charged to customers are set so that utilities receive enough revenue to cover all operating costs and a fair rate of return of and on the depreciated book value of the capital assets that are considered by the regulators to be used and useful. This procedure is assumed to still be the appropriate method of paying for transmission and distribution assets in a deregulated market. For merchant generators, however, their revenue in a deregulated market comes from 1) being paid the nodal prices for their generation and ancillary services in the Wholesale Market 21, and 2) payments for capacity in a Capacity Market if such a market exists. These generators expect to earn a market rate of return on the market value of their generating assets. Given the physical durability of conventional generating units relative to the standard regulatory accounting rates of depreciation, the market value of conventional capacity is typically substantially higher than the book value would have been under continuous regulation 22. This is a major additional cost that should be compared with any gains in economic efficiency in the Wholesale Market that lower operating costs. The amount of missing money for a generator is determined by specifying the minimum annual earnings needed to maintain the Financial Adequacy of the 21 These payments may also be made through forward contracts, but the contract prices will still reflect the expectations of traders about future prices in the Wholesale Market. In addition, there may be bilateral contracts that include two-part payments for energy, or an ancillary service, and for capacity. 22 The late Mike Rothkopf was one of the few economists to raise this issue as an important reason for being skeptical about the widely held belief among academics and regulators that deregulating electric utilities would benefit customers. See Michael H. Rothkopf, Dealing with Failed Deregulation: What Would Price C. Watts Do? The Electricity Journal 20(7), pp.10-16, July-August

24 conventional generating units. It is assumed that the minimum annual earnings above the annual operating costs correspond to the replacement value for each generating unit. The values used in the analysis are shown in Table 5. As long as the annual earnings in the Wholesale Market are bigger than the minimum earnings, the generating unit meets the standard of Financial Adequacy and there is no missing money. On the other hand, if the annual earnings in the Wholesale Market are less than the minimum earnings, the difference between the minimum earnings and the actual earnings measures the amount of missing money. Dividing the amount of missing money by the minimum generating capacity gives the minimum price of capacity ($/MW/Year) needed for Financial Adequacy. In other words, as long as the price paid in a Capacity Market is larger than this minimum price, it is big enough to ensure the Financial Adequacy of all of the generating units. Table 5: Minimum Annual Earnings for the Financial Adequacy of Generators Source: EIA Annual Energy Outlook 200***** The final step in the calculation of the total amount of missing money for conventional generators is to specify a structure for the Capacity Market. It is assumed, following the structure of the market in New York State, that the Capacity Market is divided into two regions; 1) Area 1, the urban region, and 2) Areas 2 and 3, the rural region. Each region sets its own capacity price and this price is equal to the highest of the minimum capacity prices needed for Financial Adequacy for all of the generating units in a region. The market price is paid for all of the generating capacity in a region 24

25 that is needed to meet the Peak system load and maintain System Adequacy. This procedure follows the standard practice used to make payments in a uniform price auction. 4. Results of the Simulations The basic structure of the case study is to evaluate the system effects of replacing 35MW of installed coal capacity at Node 13 (Generator 6) by 105MW of wind capacity. This change represents the type of policy option that regulators implicitly consider when setting a target for the increased penetration of wind capacity using, for example, a Renewable Portfolio Standard. Earlier research 23 has shown that the main effects of making this change are to 1) increase the amount of reserve generating capacity needed to maintain reliability, 2) reduce the total operating cost of meeting load, and 3) reduce the annual capacity factors for most of the conventional generating units. The implications of adding the wind capacity show that 1) it is important to determine the amount of capacity needed for Generation Adequacy endogenously, 2) the average wholesale prices are lower, and 3) the net revenues of conventional generating units above operating costs are reduced substantially. The third implication relating to the earnings of conventional generators poses a serious problem in deregulated markets and this is the main problem that is addressed in this paper. The new results demonstrate the potential value to the network of coupling storage capacity with wind generation to reduce the variability of the net generation from a wind farm. Since PHEVs can be viewed as mobile batteries, the results also have implications for how the electrification of transportation can complement the electric delivery system and lower the overall cost of maintaining System Adequacy. The results presented in this section summarize the economic costs for the network shown in Figure 1 of meeting the same annual pattern of load (shown in Figure 3) for the eight different scenarios discussed in the previous section. For this analysis, it is assumed that the wholesale market is deregulated. The main questions of interest are 23 See the conclusions in Tim Mount, Alberto Lamadrid, Surin Maneevitjit, Bob Thomas and Ray Zimmerman, Evaluating the Net Benefits of Investing in New Wind and Transmission Capacity on a Network, Proceedings of the 42 nd IEEE HICSS Conference, Waikoloa HI, Jan

26 1) how much generating capacity is needed to maintain System Adequacy, and 2) how much missing money will have to be paid to conventional generators and transmission owners. When the results for any two scenarios are compared, a reduction in the total annual cost of meeting load and maintaining the Financial Adequacy of the system provides the increased economic surplus that should be compared to the annualized cost of financing the investment that characterizes the difference between the two scenarios. In other words, the cost of making an investment in wind turbines, for example, should be less than the reduction in the total cost of running the existing network if the investment is to be economically beneficial. The basic economic question about adding wind capacity to a network is to determine the extent to which the higher amounts of missing money offset the lower cost of operating the network. The four different types of wind generation considered are NO Wind, NORMAL Wind, NICE Wind (i.e. wind generation coupled with storage batteries) and NASTY Wind (i.e. must-take contracts for wind generation). For each one of these four types of wind generation, a scenario is run with and without an upgrade of the main tie line linking the wind farm to an urban center. The order of the eight scenarios is: 1. Case 1; NO Wind 2. Case 1UP; NO Wind + Upgrade 3. Case 2; NORMAL Wind 4. Case 2UP; NORMAL Wind + Upgrade 5. Case 3; NICE Wind 5. Case 3UP; NICE Wind + Upgrade 7. Case 4; NASTY Wind 8. Case 4UP; NASTY Wind + Upgrade Comparing the total annual system cost of any one of the four wind cases with the corresponding scenario with a tie line upgrade gives the information needed to evaluate the economic benefit of making the investment in upgrading the tie line. For the different wind cases, the most interesting comparisons are between NO wind and each one of the other three cases. The basic results from making these comparisons show that there is little change in the total system cost from adding NORMAL Wind, a large reduction in the system cost from adding NICE Wind and a large increase in the system cost from adding NASTY wind. These differences in total system costs are determined largely by differences in the amount of missing money rather than by differences in the operating cost, and the total amounts of conventional generating capacity needed to maintain System Adequacy are the key determinants of the amount of missing money. 26

27 Before presenting the empirical results, some additional qualifications should be made. The overall implications of the results are specific to this case study and they should not be extrapolated to make inferences about the effects of adding wind capacity to a network in general. First, the amount of wind capacity added to the network is at a single location and is relatively large. The wind capacity corresponds to roughly one quarter of the total installed generating capacity. Hence, the variability of wind generation has a much more serious effect on system reliability than it would when wind generation provides only a small part of total generation. Second, the minimum amount of annual earnings needed by conventional generators to maintain Financial Adequacy is also relatively large because it is based on the replacement costs of generating capacity. This specification exaggerates the economic effects of changing the amounts of generating capacity needed for System Adequacy, but it also demonstrates clearly the danger of reaching potentially misleading conclusions about the economic benefits of adding wind capacity when the effects on missing money are ignored. Table 6: Summary of Key Results For each scenario, the reported annual costs are the sums over the 100 load levels shown in Figure 3 of the expectation of the costs for the three forecasts of wind speed shown in Tables 3 and 4 using the second-stage optimization of the SuperOPF. 24 The key results for the eight scenarios are presented in Table 6. The first row (Load Paid) shows that the annual payments made by customers in the wholesale market are substantially lower than the NO Wind scenario (Cases 1 and 2) for all of the wind scenarios except Case 4 (NASTY Wind). These cost reductions represent the displacement of fossil fuels by wind generation whenever the wind blows, and at low loads, the expected generation 24 In other words, the expected costs are computed for the different contingencies in Table 2 and for the different forecasts of wind speed. 27

28 from wind is the dominant source. For NICE Wind with the upgraded tie line (Case 3UP), the customers only pay one sixth of the corresponding cost with NO wind (Case 1) in the wholesale market. Wind also displaces fossil fuels in the Case 4, and the high wholesale prices paid by customers are caused by the increased cost of dealing with congestion on the network. 25 The generally lower wholesale costs of purchases with wind generation in Table 6 contrast with the amounts of conventional generating capacity needed for System Adequacy (Gen Capacity Needed). The capacity needed is roughly the same with NORMAL Wind (Cases 2 and 2UP) as it is with NO Wind (Cases 1 and 1UP), and it is even higher with NASTY Wind (Cases 4 and 4UP). It is only with NICE Wind that the capacity needed is substantially lower (Cases 3 and 3UP). The underlying reason is that the storage capacity coupled with wind generation in Cases 3 and 3UP provides a floor on the minimum generation of 35 MW from the wind farm, including the effect of the cutout contingency at very high wind speeds. In addition, the ranges of the uncertainty about actual wind realizations for a given forecast of wind speed are also reduced (see Table 4), and this reduction also increases the capacity value of the wind farm for meeting the peak system load. Even though the production cost and price offer of wind generation is set to zero for NORMAL Wind, the maximum amount of wind capacity dispatched (Max Wind Committed) is only 35MW compared to the true maximum of 105MW. The reasons are 1) the size of the contingency when the wind turbines cutout at high wind speeds is set by the maximum capacity dispatched, and 2) the additional cost of congestion on the network can overwhelm the low production cost of wind generation. More wind generation can be used economically when the tie line is upgraded (Case 2UP) or if storage capacity is coupled with the wind generation (Cases 3 and 3UP). However, the amount of wind capacity that is dispatched is not really limited by the physical capacity of the network. In Cases 4 and 4UP with must-take contracts, all 105MW of wind capacity are dispatched but the consequence is that customers have to pay a lot more in 25 When wind generation is the dominant source at low loads, it is difficult to accommodate this generation on the network because so much of the total generation is produced at a single node. With a must-take contract in Case 4, there is an economic penalty for not using all of the available generation from wind even though this source increases the marginal production cost substantially and this marginal cost sets the market price in a uniform price auction. 28

29 the wholesale market for congestion. The maximum amount of wind capacity dispatched is inversely related to the proportion of total generation from conventional sources (Conventional Generation), and in Case 4UP, conventional generation provides only 55% of the total. In contrast, conventional generation provides 88% of the total with NORMAL Wind in Case 2, implying that almost 75% of the potential wind generation is spilled. Spilling less wind in Cases 3 and 3UP leads to lower wholesale costs by displacing more fossil fuels, but in Cases 4 and 4UP, the additional costs of reserve capacity and congestion overwhelm the lower cost of fossil fuels. The final row in Table 6 shows the number of hours in a year in which some load is shed (Load Not Served). Strictly speaking, none of the scenarios meet the NERC standard of one-day-in ten-years (less than 2.4 Hours/Year). However, the amounts of load shed are very small and the annual economic costs of LNS are trivial compared to the total system cost. The basic economic choice is to decide whether to add more reserve capacity at a certain small cost to avoid shedding load at a very high cost in a contingency that has a very small probability of occurring. 26 However, comparing the values of Load Not Served in the different scenarios does demonstrate that load shedding is determined endogenously by the SuperOPF, and it also shows that load shedding occurs much more frequently with NASTY Wind, as expected, in Cases 4 and 4UP. Annual Costs to Customers in the Wholesale Market The wholesale market is operated by an Independent System Operator (ISO) and customers pay the real-time nodal prices for purchasing real energy throughout the year. The results reported in Table 7 are the same as the expected payments shown in Table 6. Generators are also paid by the ISO the real-time nodal prices for generating real energy and the day-ahead prices for providing reserve generating capacity. The difference between the money paid by customers to the ISO and the money paid to generators by the ISO measures the congestion rents that are transferred to transmission owners and these 26 A certain payment of $10/MWh for 1MW of reserve capacity is equal to the expected cost of shedding 1MW of load at a VOLL of $5,000/MWh in a typical contingency with a probability of 0.2%. If the load shedding occurs in a contingency for only one of the three wind forecasts, the expected cost of shedding load is even lower. 29

30 rents are called the Congestion Surplus in Table 7. The Congestion Surplus may be negative, and in this situation, it is another source of missing money for the network. Some of the money paid to the conventional generators covers their operating costs, primarily for fuels. The difference between the money paid to the generators and their operating costs measures the annual Net Revenue of generators. Since, the offer prices submitted to the wholesale auction are equal to the true operating costs, the annual Net Revenue for a generating unit is never negative. The annual payments by the ISO to the wind farm are equal to the annual Net Revenue because the operating cost is set to zero for wind generation. The Net Revenues for conventional and wind generators are reported in Table 7 and the Total Producer Surplus for operating the network is the sum of these two Net Revenues. Table 7: Expected Annual Operating Costs and Benefits The results in Table 7 include an additional cost corresponding to the cost of shedding load (Expected LNS x VOLL). This cost is treated as a real cost of operating the system even though planners usually ignore it and treat System Adequacy as a physical requirement rather than economic issue. 27 Since the demand for real energy is perfectly inelastic for every level of the system load, the consumer surplus is the difference between the system load evaluated at the appropriate VOLL and the sum of the expected payments to the ISO and the expected cost of shedding load. In Table 7, an arbitrary large number is used for the maximum possible consumer surplus for free electricity ($2 billion/year) to keep the magnitudes of the reported consumer surplus in line with the magnitudes of other sources of surplus. Consequently, it is more 27 Identifying this cost explicitly is equivalent to paying compensation to the customers who lose power. 30

31 informative to interpret the values of Consumer Surplus in Table 7 by evaluating the change in this value from one scenario to another rather than to draw conclusions from the individual values. The Total Operating Surplus is the sum of the Producer Surplus, the Congestion Surplus for transmission owners and the Consumer Surplus. The change in the Total Operating Surplus from one scenario to another scenario is the negative of the corresponding change in the Total Real Operating Costs. In other words, given the assumption that demand is perfectly inelastic, reducing the Total Real Operating Cost is the only way to increase the Total Operating Surplus. However, the allocation of this surplus among consumers, generators and transmission owners varies substantially from scenario to scenario. Using NO Wind (Case 1) as the base for comparison, the Total Operating Surplus increases for NORMAL and NICE Wind (Cases 2 and 3) but not for NASTY Wind (Case 4). The Operating Cost in Case 4 is high because the mix of generating capacity dispatched is very different from the least-cost merit order due to network constraints and congestion. In all four cases, upgrading the tie line increases the Total Operating Surplus by reducing the Operating Cost and making it possible to rely more on the sources of generation with the lowest operating costs. The benefits from the increased Total Operating Surplus when NORMAL and NICE wind is compared to NO Wind go to customers, wind generators and transmission owners. Conventional generators are the big losers even though their actual operating costs are lower. In contrast, the winners with NASTY Wind are conventional generators as well as wind generators, and for wind generators, their annual net revenue is actually higher with NASTY Wind than it is for any other scenario. Consumers and transmission owners are the losers with NASTY Wind. Adding the tie line upgrade to any one of the four wind scenarios increases the surplus for consumers and reduces it for transmission owners, as expected. With NO Wind, conventional generators benefit slightly from the upgrade, but they lose for the other three wind scenarios because upgrading the tie line makes it economic to spill less of the potential wind generation and displace more conventional generation. Wind generators benefit from the upgrade with NORMAL and NICE Wind but not with NASTY Wind because the upgrade reduces the high nodal prices caused by congestion that are paid to all generators. 31

32 The Missing Money for Conventional Generators and Transmission Owners The simplest type of missing money is for transmission owners. It is assumed for all scenarios that transmission owners received $30 million/year to cover all of their costs for the existing network including the annualized capital costs. Part of this total is paid by the Congestion Surplus in the wholesale market (see Table 7) and the remaining part corresponds to the missing money. For scenarios in which the ISO pays more to generators than the amount received from customers, the Congestion Surplus is negative and the corresponding missing money will be larger than $30 million/year. However, customers still pay the same total cost of transmission and the only feature that changes from scenario to scenario is the amount of money contributed in the wholesale market. For generators, the situation is more complicated because the amount of conventional generating capacity needed for reliability purposes and the total annual operating costs vary from scenario to scenario. Nevertheless, the total annualized capital cost of the conventional generating units is paid in a similar way to the payments for transmission. Some of the money comes from the wholesale market (Gen Net Revenue in Table 7) and the rest is paid as missing money through a Capacity Market. The calculations for determining the amount of missing money are illustrated in Table 8 for the NO Wind scenario. Table 8: The Missing Money for Conventional Generators with NO Wind (Case 1) Column 1 in Table 8 shows the minimum annual earnings/mw for the individual generating units presented in Table 5, and Column 2 shows the amounts of capacity needed to meet the System Peak Load and maintain System Adequacy (the total capacity corresponds to the value shown in Table 6). The product of these two columns in 32

33 Column 3 determines the minimum annual earnings for each generating unit needed to maintain Financial Adequacy. The values in Column 5 are the differences between the actual earnings in Column 4 and the minimum earnings in Column 3 (the total corresponds to the Gen Net Revenue in Table 7). The negative values in Column 5 indicate that there is missing money, and all generating units but one face this situation. 28 When a generating unit has a positive difference in Column 5, it indicates that the unit is earning some excess profits and these profits are kept by the generator. The values for the missing money in Column 6 are equal to the Maximum[0, -Column 5], and the implicit prices of capacity in Column 7 are calculated by dividing Column 6 by Column 2. Since the capacity market is specified to have two regions, one for Gen 1 and Gen 2 and another for Gen 3 6, the price paid for capacity is the highest price needed by any one of the generating units in each region (bold in Column 7). The payments for capacity in Column 8 are determined by multiplying Column 2 by $81,000/MW/Year for Gen 1 and Gen 2, and by $170,000/MW/Year for Gen 3 6. The total of $40 million/year is the amount paid by customers to cover the missing money for the conventional generators. This total includes some excess profits for generators as well as the money that is actually needed to maintain Financial Adequacy. In an ideal regulated market, the regulators would set the rates charged to customers to pay for all of the operating costs and capital costs. The main differences in the procedures used for a deregulated market in Table 8 are 1) the Minimum Earnings/MW in Column 1 would be calculated from the book value of capital rather than the market value, and 2) the capacity payments would cover the missing money for each unit in Column 6 and would also confiscate excess profits (i.e. positive differences in Column 5). In other words, even if the lower values for Column 1 are ignored, the customers would only pay $23 million/year (Column 6) under regulation rather than $40 million/year (Column 8) in a Capacity Market. For this case study, the capacity payments in Column 8 are the ones presented but it would be also be sensible to evaluate the scenarios using the sum of the differences in Column 5 instead of the Total for 28 The value for the second unit of Generator 4 in Column 5 is actually positive but it appears to be zero because the amount of capacity needed is trivially small. However, this unit does account for the slight difference in the totals in Columns 5 and 6. 33

34 Column 8. A comparison of the costs to customers in a fully regulated market and in a deregulated market will be the considered in future research. The steps used to calculate the payments for capacity outlined in Table 8 were completed for each scenario. These payments together with the payments to transmission owners are added to the Total Real Operating Costs in Table 6 to give the Total Annual System Cost charged directly in the wholesale market and indirectly as missing money to customers. These values are shown in Table 9. Before comparing the results in Table 9 for the different scenarios, it is important to understand the implications of the results in Table 8. Since all but one of the generating units do not receive enough net revenue in the wholesale market for Financial Adequacy, the mechanism for paying the cost of capital for generating units is similar to the way that transmission owners are paid 29. Table 9: The Annual Net Social Surplus for the Existing Network The money needed for Financial Adequacy comes from both the Gen Net Revenue in Table 6 and the Missing Money in Table 9. If the levels of the different types of generating capacity needed for System Adequacy remain the same, the total amount paid for the generators capital costs stays the same. Lower earnings in the wholesale market are offset by higher payments in the capacity market. 30 Hence, the only effective ways to reduce the total annual payments to generators are 1) to reduce the amount of 29 The fact that most of the generating units in Table 8 do need additional earnings in the Capacity Market is consistent with a competitive wholesale market. If all offer prices in the wholesale auction equal the true marginal operating costs and the combination of different types of generation is economically efficient, all generating units would have the same amount of missing money/mw equal to the capital cost of the most expensive peaking unit. 30 Strictly speaking, the equivalent of the negative sum of the differences in Column 5 of Table 8 offsets the earnings in the wholesale market, and the procedures used to make payments in the capacity market may introduce a difference in the total. 34

35 installed capacity needed for System Adequacy, and to a lesser extent, 2) to change the mix of generating units needed for System Adequacy. Using the NO Wind scenario (Case 1) as the base, the values of the annual Net Social Surplus in Table 9 show that this surplus is slightly lower with NORMAL Wind, is substantially higher with NICE Wind (Case 3) and substantially lower with NASTY Wind (Case 4). Consequently, there is no economic justification for adding wind capacity unless the variability of wind generation is mitigated as it is in Case 3. There is no additional economic surplus with NORMAL Wind in Case 2 to cover the cost of the investment in wind capacity, and with NASTY Wind in Case 4, the cost of the paying for the existing network increases substantially even if adding the wind capacity incurs no additional costs to customers. For all four of the wind cases upgrading the tie line increases the Net Social Surplus, but for Case 4UP with NASTY Wind, the Net Social Surplus is still lower than Case 1 with NO Wind and no upgrade to the tie line. Figure 5: Total Payments by Customers in the Wholesale Market Figure 5 shows the dramatic reduction of the payments made by customers in the wholesale market with NORMAL and NICE Wind (Cases 2 and 3) compared to NO Wind (Case 1). It also shows how these payments increase with NASTY Wind (Case 4). The effects of upgrading the tie line are small for NO Wind and NORMAL Wind, but the reductions in wholesale payments are relatively large for NICE Wind and for NASTY 35

36 Wind. The reason for NICE Wind is that it is possible to spill less of the wind generation with the upgrade, and wind generators are paid more as a result. For NASTY Wind, the upgrade makes the problem of congestion on the network less severe and lowers and makes it more economic to accommodate the wind generation. Figure 6: The Total Annual System Costs Paid by Customers Figure 6 summarizes the overall results of the analysis by showing the Total Annual Systems Costs for the different scenarios and where the money goes. The first three components (the lowest three) are the Operating Costs, the Generator Net Revenue and the Wind Net Revenue. The sum of these three components represents the payments made in the wholesale market shown in Figure 5, and the large reductions in these payments with NORMAL and NICE Wind (Cases 2 and 3) compared to NO Wind (Case 1) are still obvious. However, this reduction for NORMAL Wind is more than offset by the large increase in the Capacity Payments made to the conventional generators to maintain Financial Adequacy. These Capacity Payments are lower for NICE Wind because less generating capacity is needed to maintain System Adequacy, and as a result, 36

37 the Total System Cost is also lower. 31 For NASTY Wind (Case 4), both the Operating Costs and the Capacity Payments to conventional generators are higher than the corresponding values for NO Wind. The main conclusion is that focusing on a reduction in the average wholesale price when wind capacity is introduced into a deregulated market in Figure 5 can be very misleading unless the effects on the missing money needed by the conventional generators and transmission owners for Financial Adequacy is also considered. The relatively low Total Annual System Cost for NICE Wind (Case 3) is caused by the combination of storage batteries coupled with the wind generation. Charging the batteries when wind speeds are high and discharging them when wind speeds are low reduces the range of net generation from the wind farm substantially and the uncertainty of generation for a given forecast of the wind speed is also reduced. The most important economic effect of having the batteries is to increase the minimum level of net generation from 0MW to 35MW (see Table 4), and this accounts for most of the reduction of 46MW of conventional generation capacity needed for System Adequacy from 288MW with NORMAL Wind (Case 2) to 242MW with NICE Wind (see Table 6). The remaining reduction is caused by limiting the size of the contingency when the wind turbines cutout at very high wind speeds. Instead of a possible maximum drop in generation from 105MW to 0MW, the range with NICE Wind is only 35MW, from a maximum of 70MW to 35MW 32 (see Table 4). As a result, less reserve generating capacity is needed to cover the loads when the wind turbines cutout. The underlying reason for calculating the Total Annual System Costs is to determine the viability of making investments in wind capacity and in upgrading a tie line, and the economic justification for an investment comes from the resulting reduction in the Total Annual System Costs. Table 10 summarizes the results for investing in 1) the three different types of wind generation compared to NO Wind (Case 1), and 2) upgrading the tie line for Cases 1 4. For both NORMAL Wind (Case 2) and NASTY Wind (Case 4) the changes are negative (shown as red) and imply that the investments in 31 The total payments made to transmission owners are the same in all scenarios and the payments to wind generators and the costs of shedding load are always relatively small. 32 A lot of the potential wind generation is wasted in the NORMAL Wind scenario in order to reduce the size of the reduction in wind generation when the turbines cutout in high wind conditions. 37

38 wind capacity are not financially viable. For NICE Wind (Case 3), over $9 million/year is available to offset the cost of the investment. Considering the upgrade of the tie line, the Total Annual System Cost is lower in all four cases. The reduction for NASTY Wind is by far the largest but it is still not large enough to make the Combined Total positive. It is interesting to compare the differences between the Combined Total for NORMAL Wind and NICE Wind. For NORMAL Wind, upgrading the tie line provides all of the reduction in cost, but for NICE Wind, adding the wind capacity is the dominant source of the reduced cost and the reduction from upgrading the tie line is relatively small. This difference suggests that it may be interesting for planners to compare adding wind capacity with storage (Case 3) to adding wind capacity with an upgraded tie line (Case 2UP). The results in Table 6 show that the primary source of the reduction in the Total Annual System Cost for Case 2UP comes by displacing conventional generation and thereby reducing Operating Costs. For Case 3, there are two sources of the reduction in the cost. More conventional generation is displaced causing a larger reduction in Operating Costs, and in addition, some conventional generating capacity is also displaced causing a reduction in capital costs. Table 10: Savings in the Total Annual System Cost Available to Offset Investments 38

39 5. A Potential Symbiotic Role for PHEVs on a Network The main question addressed in this section is how will Plug-in Hybrid Electric Vehicles (PHEV) affect the Total Annual System Cost on a network? No precise answers will be given because the new scenarios with PHEVs have not yet been completed. Hence, the discussion will be limited to identifying a set of conjectures that need to be tested. The basic effects of PHEVs used for commuting are to increase the load at night when batteries are charging, and possibly, to lower the peak load in the day by discharging excess energy from the batteries using Vehicle to Grid capabilities (V2G) 33. The PHEVs could also be used with smart charging capabilities to support the network at night by responding to changes in wind generation and mitigating its variability during the charging cycle. In other words, the PHEVs can substitute to some extent for dedicated, utility-scale storage capacity. Clearly, using the capabilities of these mobile batteries in PHEVs effectively is more technically challenging than it is for dedicated batteries, and the main advantages of using PHEVs to support the network are 1) the capital costs of the batteries are essentially zero for the network, and 2) dealing with PHEVs in a passive way that allows for charging-at-will may exacerbate problems on the network by, for example, increasing the peak system load. The new scenarios for PHEVs will be based on determining the optimal patterns of dispatch and reserve capacity for different hours in representative days of the year. A long-term objective is to develop new capabilities of the SuperOPF to solve the dynamic programming problem of unit commitment in a day-ahead market for the whole of the following day. Without this capability, the individual hours of a day will have to be solved in sequence step by step. With storage capacity on the network, some form of exogenous strategy for charging and discharging batteries will be imposed until the unit commitment problem can be solved. However, even with this limitation, some important new analytical features will be supported. The most important of these is that it will be 33 The existence of excess energy implies that the batteries in some PHEVs store more than enough energy to complete their round-trip commutes. Discharging the battery during the day using V2G will increase the load for charging the batteries at night. During the highest peak system loads, it may also be reasonable to consider fully discharging batteries and using the fossil fuel engine to charge the batteries for the trips home. Determining the environmental effects of using PHEVs in different ways on urban air quality is an important issue for policy makers, and this topic is the focus of a current project being conducted by our research group at Cornell. 39

40 possible to evaluate the shifting of load explicitly from peak to off-peak during a day. In the analysis presented in the previous section for NICE Wind with storage capacity (Case 3), the realizations of wind generation for a given forecast of wind speed in Table 4 are specified exogenously and the resulting effect on the peak system load is implicit. A second feature is that it will be possible to evaluate the economic effects of the variability of wind generation from hour to hour. By setting relatively high values for INCs and DECs for some generating units (e.g. the nuclear units) compared to peaking units and batteries, the cost of modifying the pattern of dispatch from one hour to the next can be calculated by treating the actual amounts of generating capacity dispatched in the previous hour as the contracted amounts for the next hour (i.e. the values of G i0 in the objective function presented in Section 2). Changing the pattern of dispatch from hour to hour will result in an additional cost caused by the higher heating values associated with ramping generating units 34. Table 11: The Distribution of PHEVs on the Network and Needs for Commuting % of PHEVs Round Trip Miles % of Full Charge Used for Commuting % of Full Charge Available for V2G Urban Center 20% 0 0% 100% 20% 10 25% 75% Rural 20% 20 50% 50% 20% 30 75% 25% 20% >40 100% 0% Average 50% 50% In order to evaluate the effects of the PHEVs on the performance of the network, it is assumed for illustrative purposes that the fleet of PHEVs is uniformly distributed among five different commuting distances and that 2/5 of the vehicles are in an urban center and the other 3/5 of the vehicles are in rural areas and have longer distances to 34 Changing the pattern of dispatch to cover contingencies will also cause additional costs of ramping but the expected cost of making these changes is generally small because they are weighted by their relatively small probabilities of occurrence. In contrast, the changes of the intact system from one hour to the next have high probabilities of occurring. 40

41 commute. The specifications of the fleet of PHEVs are summarized in Table 11. A fully charged battery is specified to provide electric power for 40 miles. Hence, commuters with round trips less than 40 miles can provide V2G discharging, including the vehicles in the urban center that are not used for commuting at all. With these assumptions, half of the total capacity of the batteries is needed for commuting, and potentially, the other half of the charge could be used with V2G capabilities to reduce the peak load. Fully discharging all of the batteries using V2G would double the total amount of electric energy needed for recharging the batteries over night, and typically, the rate of discharging for V2G to reduce the peak system load would be more rapid than the rate of charging at night. Figure 7: Average Daily Load Profiles for July and the Corresponding Wind Speeds The average daily load profiles for weekdays in July 2005 in New York City plus Long Island (NYC + LI) and in Upstate New York are shown in Figure 7 together with the daily pattern of average wind speeds for the wind data used in the previous section. As expected, average wind speeds are not closely related to the daily load profiles, and the additional problems of wind variability still have to be dealt with on the network. 41