reference integration of variable generation into the bulk ... of variable... · 2.7.1 facility...

RReeffeerreennccee MMaannuuaall

Integration of Variable Generation into the Bulk Power System

July 15, 2008

1

2

June 25, 2008

TTaabbllee ooff CCoonntteennttss Reference Manual ........................................................................................................................... 1

Characteristics of Variable Generation........................................................................................... 8

2.1 Brief Description of Variable Generation Technologies ................................................ 8 2.1.1 Solar Thermal.......................................................................................................... 8 2.1.1.1 Parabolic Trough Systems ...................................................................................... 9 2.1.1.2 Parabolic Dish-Engine Systems............................................................................ 10 2.1.1.3 Power Tower Systems........................................................................................... 11 2.1.1.4 Compact Linear Fresnel Reflector (CLFR) .......................................................... 11 2.1.1.5 Solar Chimney ...................................................................................................... 12 2.1.1.6 Performance Characteristics ................................................................................. 12 2.1.2 Solar Photovoltaic................................................................................................. 13 2.1.2.1 Operating Principles.............................................................................................. 13 2.1.2.2 Concentrating Solar Photovoltaic Systems........................................................... 13 2.1.2.3 Performance Characteristics ................................................................................. 14 2.1.3 Wind...................................................................................................................... 14 2.1.3.1 Performance Characteristics ................................................................................. 15 2.1.4 Ocean Power ......................................................................................................... 15 2.1.4.1 Tidal Power........................................................................................................... 16 2.1.4.1.1 Clean Current ................................................................................................... 16 2.1.4.1.2 Marine Current Turbines - Seagen ................................................................... 17 2.1.4.1.3 Performance Characteristics ............................................................................ 17 2.1.4.2 Wave ..................................................................................................................... 17 2.1.4.2.1 Pelamis.............................................................................................................. 18 2.1.4.2.2 PowerBuoy ........................................................................................................ 18 2.1.4.2.3 Performance Characteristics ............................................................................ 18

2.2 Drivers for Future Growth of Renewable Generation in the United States.................. 19 2.2.1 Summary of US Drivers for Future Growth ......................................................... 19 2.2.1.1 Renewable Portfolio Standards (RPS) Legislation (State and Federal)................ 19 2.2.1.1.1 State Legislated Renewable Portfolio Standards (RPS) ................................... 19 2.2.1.1.2 Renewable Portfolio Standards (RPS) at the Federal Level ............................ 29 2.2.1.2 Carbon constraints ................................................................................................ 30 2.2.2 Drivers for Future Growth of Renewable Generation in Canada ......................... 30 2.2.2.1 Carbon Constraints................................................................................................ 30 2.2.2.2 Sustainable Energy Policy .................................................................................... 31 2.2.2.3 Characteristics of feed-in tariffs............................................................................ 32 2.2.2.4 Green Power Program and Certificates................................................................. 33

2.3 Existing Variable Generation Installed Capacity Levels .............................................. 33 2.4 Major Technical Characteristics of Variable Generation Related to Power System Operation................................................................................................................................... 34

2.4.1 Wind Generation................................................................................................... 34 2.4.1.1 Basic Types of Wind Turbine-Generators ............................................................ 34 2.4.1.1.1 Type 1: Fixed speed Induction Generator ........................................................ 34

3

2.4.1.1.2 Type 2: Variable-Slip Induction Generator ..................................................... 36 2.4.1.1.3 Type 3: Double-Fed Asynchronous Generator................................................ 37 2.4.1.1.4 Type 4: Full Power Conversion variable speed .............................................. 37 2.4.1.2 Voltage and Reactive Power Control.................................................................... 38 2.4.1.2.1 Power factor control ......................................................................................... 38 2.4.1.2.2 Voltage Control................................................................................................. 39 2.4.1.2.3 Zero-Power Voltage control ............................................................................. 42 2.4.1.3 Low-Voltage Ride-Through.................................................................................. 42 2.4.1.3.1 High Voltage Ride-Through.............................................................................. 43 2.4.1.4 Active Power Control Functions........................................................................... 44 2.4.1.4.1 Curtailment ....................................................................................................... 44 2.4.1.4.2 Power Ramp Rate Control ................................................................................ 44 2.4.1.4.3 Frequency Regulation and Reserve Functions ................................................. 45 2.4.1.4.4 Over-Frequency Response ................................................................................ 45 2.4.1.4.5 Under-Frequency and Power Reserve Response.............................................. 46 2.4.2 Concentrating Solar Generation (by Eric John)................................................... 48

2.5 Geographic Diversity Impacts of Variable Generators Diversity on Power System Operation................................................................................................................................... 48

2.5.1 Geographic Diversity of Wind Power................................................................... 48 2.5.2 Distribution of Wind Power Step Changes........................................................... 55 2.5.3 Wind Plant Ramping Behavior ............................................................................. 56 2.5.4 Rare Situations ...................................................................................................... 58 2.5.5 Summary and Conclusions ................................................................................... 61 2.5.6 Diversity between Wind Generation and Load..................................................... 61 2.5.7 Diversity across Technology Families.................................................................. 71

2.6 Technical Power System Operation Challenges with Variable Generation and Current Operations Practices to Address Such Challenge ..................................................................... 73

2.6.1 Seconds to Minute Time Frame............................................................................ 74 2.6.1.1 Automatic Generation Control (AGC), Area Control Error (ACE), and Frequency Regulation:............................................................................................................................ 74 2.6.1.2 Voltage/Reactive Power Control .......................................................................... 76 2.6.1.3 Contingency Reserves........................................................................................... 77 2.6.1.4 System Stability and Dynamic Response Characteristics..................................... 77 2.6.1.5 System Restoration ............................................................................................... 78 2.6.2 Minute to Hours Time Frame ............................................................................... 78 2.6.2.1 Near-Term Reserves ............................................................................................. 79 2.6.2.2 Load Balancing / Energy Balancing, and Sustained Ramp Events ...................... 80 2.6.2.3 Transmission Constraints and Transfer Limits ..................................................... 82 2.6.2.4 Excess Energy (Over-Generation Condition) ....................................................... 83 2.6.3 Longer Time Frames............................................................................................. 84

2.7 Technical Power System Planning Challenges with Variable Generation ................... 84 2.7.1 Facility Connection Requirements “Grid Code” .................................................. 86 2.7.2 Variable Generation Models ................................................................................. 86 2.7.3 Facility Connection Studies .................................................................................. 87

2.8 Variable Generation Forecasting Methods ................................................................... 88 2.8.1 Wind Generation Forecasting ............................................................................... 88

4

2.8.1.1 Capacity Value Analysis....................................................................................... 89 2.8.1.2 Near Term and Real Time Wind Forecasting:...................................................... 90 2.8.2 Solar generation forecasting ................................................................................. 90 2.8.2.1 Forecasting by Technology................................................................................... 91 2.8.3 Other technologies ................................................................................................ 93 2.8.4 Challenges Ahead ................................................................................................. 93

Transmission and Generation Planning Impacts........................................................................... 94

3.1 Introduction (DB).......................................................................................................... 94 3.2 Representation of Wind Generator Output for Planning Studies (MM, DO) ............... 95

3.2.1 Some Data Requirements are Common for Transmission Planning and Wind Integration Studies ................................................................................................................ 95 3.2.2 Hourly Wind Power Simulated or Actual Data Can be Used for Wind Integration Studies and Long-term Transmission Planning Studies. ...................................................... 98 3.2.3 Wind Integration Studies Also Need Sub-Hourly Data........................................ 98 3.2.4 Caution Should Be Used With Existing Wind Plant Data.................................... 98

3.3 Long-Term Resource Adequacy Planning (MM, DO, KP, DB)................................... 99 3.3.1 Current Supply Adequacy Planning Approaches ................................................. 99 3.3.2 Wind Plants and Supply Adequacy Evaluations................................................. 100 3.3.3 Methods for Calculating the Capacity Credit Assigned to Wind Plants............. 100 3.3.3.1 Effective Load Carrying Capability (ELCC) ...................................................... 101 3.3.3.2 Simplified Risk-Based Methods........................................................................... 110 3.3.3.3 Time Period Methods .......................................................................................... 110 3.3.4 Recommendations for reliability assessments needed to ensure variable/dispatchable resources influence on reliability/adequacy is reasonably assessed. 111

3.4 Transmission Planning (A. Ellis, M. Fecteau, Y. Kazachkov, S. Paquette, P. Pourbeik, D. Schooley) ........................................................................................................................... 112 3.2 Transmission Planning (A. Ellis, M. Fecteau, Y. Kazachkov, S. Paquette, P. Pourbeik, D. Schooley) ........................................................................................................................... 112 3.2.1 A Review of Present Approaches to Planning and Current Grid Codes (AE)........ 112

Modeling standards............................................................................................................. 112 Generator performance standards ....................................................................................... 112

3.2.2 Transmission Expansion Required for Integrating Wind (PP) ............................... 113 3.2.3 Dynamic Line Ratings and Other Considerations for Maximum Utilization of Transmission Capacity (PP).................................................................................................... 114 3.2.4 Identify Improvements to Transmission Planning Approaches (PP)...................... 116 3.2.5 The Need to Consider Interactions Issues (PP)....................................................... 118

Sections Missing ................................................................................................................. 119 3.2.6 Wind Generation Modeling for Power System Stability, Power Flow, and Short-Circuit Analyses [18] (YK, PP) .............................................................................................. 120 3.2.7 How Should Models be Utilized for Traditional System Impact Studies (AE, PP) ...... 127 3.2.8 Modeling requirement for positive-sequence models (SP)............................................ 128

3.2.9 Modeling wind plants for transient studies (MF, SP, PP) ................................ 129 3.2.10 Recommendations for Model Development .................................................... 130

3.2.11 Section Summary and Recommendations (PP) ...................................................... 131 References:.............................................................................................................................. 132

5

Appendix: Current Methods Utilized in U.S. Jurisdictions to Calculate Capacity Credit of Wind Plants............................................................................................................................. 145

Pennsylvania-New Jersey-Maryland Regional Transmission Organization....................... 145 New York ISO .................................................................................................................... 146 ISO New England ............................................................................................................... 146 Southwest Power Pool ........................................................................................................ 148 Minnesota Department of Commerce/Xcel ........................................................................ 148 PacifiCorp ........................................................................................................................... 149 Electric Reliability Council of Texas (ERCOT)................................................................. 149 Mid-Continent Area Power Pool (MAPP).......................................................................... 149 Portland General Electric (PGE)......................................................................................... 150 Nebraska Public Power District .......................................................................................... 150 Idaho Power ........................................................................................................................ 150 Puget Sound Energy (PSE) ................................................................................................. 150 California ............................................................................................................................ 151 PNM.................................................................................................................................... 151 Tri-State Generation and Transmission .............................................................................. 152 Colorado PUC/Xcel Energy................................................................................................ 152 Rocky Mountain Area Transmission Study........................................................................ 152

Operational Planning & System Operations ............................................................................... 153

5. Other and Future Considerations ............................................................................................ 181

Introduction:............................................................................................................................ 181 Geographical diversity and aggregation of wind plants: ...........Error! Bookmark not defined.

Wind Variability and Forecasting..........................................Error! Bookmark not defined. Balancing Area Size and Variability......................................Error! Bookmark not defined. Market Size and Flexibility....................................................Error! Bookmark not defined. Transmission expansion- a tool for aggregation and variability reduction.Error! Bookmark not defined.

Dealing with variability through demand response: ..................Error! Bookmark not defined. Price responsive load markets and demand response ............Error! Bookmark not defined. Demand response as a wind-specific ancillary service..........Error! Bookmark not defined. Responsive Load characteristics ............................................Error! Bookmark not defined.

Future outlook for dealing with uncertainty through forecasting: ........... Error! Bookmark not defined.

Value of forecasts ..................................................................Error! Bookmark not defined. Different forecasts for different time periods ........................Error! Bookmark not defined. Forecast error and uncertainty ...............................................Error! Bookmark not defined. Forecasting data requirements ...............................................Error! Bookmark not defined. Central vs plant forecasts .......................................................Error! Bookmark not defined. Moving forecasts into the operating environment. ................Error! Bookmark not defined.

Energy Storage considerations for wind energy: .......................Error! Bookmark not defined. Sources of system flexibility..................................................Error! Bookmark not defined. Dedicated vs. system storage .................................................Error! Bookmark not defined. Findings regarding storage in existing wind integration studies ......... Error! Bookmark not defined. Modeling storage technologies ..............................................Error! Bookmark not defined.

6

7

Outlook for storage technologies ...........................................Error! Bookmark not defined. Plug-in hybrid electric vehicles .............................................Error! Bookmark not defined.

Future wind generator/plant behavior........................................Error! Bookmark not defined. Inertial response and primary frequency control ...................Error! Bookmark not defined. No-load var production and voltage control ..........................Error! Bookmark not defined. Wind Plant Participation in AGC ..........................................Error! Bookmark not defined. SCADA Data Requirements and Communications Protocol.Error! Bookmark not defined. High speed cut-out .................................................................Error! Bookmark not defined.

Continued development of generic turbine models ...................Error! Bookmark not defined. Dealing with new wind turbine architectures ........................Error! Bookmark not defined. Long term model development, maintenance and validation Error! Bookmark not defined. Model data reporting requirements for turbine manufacturers ............ Error! Bookmark not defined.

Cumulative Impacts of Distributed Generation on Bulk System Behavior....Error! Bookmark not defined.

IEEE 1547 anti-islanding requirements and conflict with LVRT requirements............Error! Bookmark not defined. Cell architecture of Energinet.dk for high wind penetration .Error! Bookmark not defined.

Conclusions and Recommendations ..........................................Error! Bookmark not defined. References..................................................................................Error! Bookmark not defined. Appendix 1.................................................................................Error! Bookmark not defined.

Glossary ...................................................................................................................................... 225

CChhaarraacctteerriissttiiccss ooff VVaarriiaabbllee GGeenneerraattiioonn

This chapter broadly presents the characteristics of various variable generation technologies which are commonly intended to harness renewable energy.1 Variable generation technologies generally refer to those generating technologies where the primary energy source cannot be reasonably stored leading to a generally uncontrolled primary source. Although the main focus of this NERC report is on wind generation, in this chapter we have tried to present some background information on most variable generation technologies.

2.1 Brief Description of Variable Generation Technologies

Variable generation technologies are diverse and include wind, solar, biomass, biogas, geothermal, hydroelectric, and ocean energy. Steady advances in equipment and operating experience spurred by government incentives have lead to many mature renewable technologies. The technical feasibility and cost of energy from nearly every form of renewable energy have improved since the early 1980s and the field is rapidly expanding from the niche markets of the past to making meaningful contributions to the world’s electricity supply.

The technologies presented in this chapter are:

Solar Thermal Generation

Solar PV Generation

Wind Generation

Ocean Power Generation

Generally, each technology is described with respect to its principles of operation and resource characteristics and performance.

It should be noted that the characteristics provided in this section are general, and have been developed for the purposes of providing high-level information on these technologies. It should also be noted that a few of the technologies presented here are not commercially viable at this time.

2.1.1 Solar Thermal

The technologies discussed include:

Parabolic trough

Parabolic dish

Power tower

Compact Linear Fresnel Reflector (CLFR)

Solar Chimney

1 This chapter heavily draws on Chapter 5 of the Phase 1 A Report prepared for the California Public Utility Commission’s Renewable Energy Transmission Initiative prepared by Black & Veatch (will add site address once finalized)

8

Thermal plants consist of two major subsystems: a collector system that collects solar energy and converts it to heat, and a power block that converts heat energy to electricity. Concentrating solar thermal power plants (CSP) produce electric power by collecting the sun’s energy to generate heat using various mirror or lens configurations. For solar thermal electric systems, the heat is transferred to a turbine or engine for power generation. Other solar thermal systems, like the solar chimney, collect solar heat without the aid of concentrators.

All CSP systems make use of the direct normal insolation (DNI) component of solar radiation, that is, the radiation that comes directly from the sun. Global radiation, which is reflected radiation, is present on sunny and cloudy days but is unusable by CSP systems. Since all CSP systems use DNI and concentration of DNI allows a solar system to achieve a high working fluid temperature, there is a need for the collector systems to track the sun. Parabolic trough and CLFR systems use single-axis trackers to focus radiation onto a linear receiver, while dish-Stirling and power tower CSP systems use two-axis trackers.

Trough, power tower, CLFR, and chimney systems collect heat to drive central turbine-generators making them best suited for relatively large plants—50 MW or larger. Trough, tower and CLFR plants, with their large central turbine generators and balance of plant equipment, have a cost advantage of economy of scale. Dish systems are modular in nature, with single units producing power in the range of 5 kW to 35 kW making them ideal for distributed or remote generation applications. Dish systems can also be sited as large plants by aggregating many units. Dish systems have the potential advantage of mass production of individual units, similar to the mass production of automobiles.

Trough and tower systems have the potential advantage over dish systems in that an amount of dispatchability can be designed into the system with thermal storage or the use of hybrid fossil fuel. Storage for CLFR systems, while being explored in concept, has not been developed. Dispatchability allows the solar plant to generate electricity during short duration cloudy periods or to generate electricity into the evening after sunset. This gives the plant potential to receive capacity credit, and provides the ability to more closely match the utility peak load profile. At this time, dish-Stirling systems have not been configured to provide hybrid fossil capability.

Solar chimney systems behave differently from the other solar technologies in that they can continue to produce electricity beyond sunny periods without the use of thermal storage systems or fossil fuels. Only a residual heat difference is needed.

2.1.1.1 Parabolic Trough Systems

Parabolic trough solar thermal systems have been the dominant solar thermal technology installed to date. Parabolic trough systems concentrate DNI using single axis tracking, parabolic curved, trough-shaped reflectors onto a receiver pipe or heat collection element (HCE) located at the focal line of the parabolic surface. A high temperature heat transfer fluid (HTF) picks up the thermal energy in the HCE. Heat in the HCE is then used to make steam in the steam generator. The steam drives a conventional steam-Rankine power cycle to generate electricity. Figure 2.1-1 shows trough collectors. A collector field contains many parallel rows of troughs connected in series. Rows are typically placed on a north-south axis, allowing the single-axis troughs to track the sun from east to west during the day.

9

10

Figure 2.1-1. Kramer Junction Trough Plant (NREL)

Parabolic trough systems are considered commercially available. The currently planned technology for thermal storage is the molten salt two-tank system. This provides a feasible storage capacity of up to 12 hours and is considered to have a low-to-moderate associated technology risk.

2.1.1.2 Parabolic Dish-Engine Systems

A solar parabolic dish-engine system comprises a solar concentrator (or “parabolic dish”) and the power conversion unit (PCU). The concentrator consists of mirror facets which combine to form a parabolic dish. The dish redirects DNI to a receiver mounted on a boom at the dish’s focal point. The system uses a two-axis tracker such that it points at the sun continuously.

A parabolic dish-engine system using a Stirling engine is shown in Figure 2.1-2. The PCU includes the thermal receiver and the engine-generator. In the solar receiver, radiant solar energy is converted to heat in a closed hydrogen loop, driving the Stirling engine-generator. Because the PCUs are air cooled, water cooling is not required. This is important because water cooling is necessary for the large, central power blocks associated with trough and power tower technologies. Thermal storage is not currently considered to be a viable option for dish-Stirling systems.

Individual dish-Stirling units range in size from 5 to 25 kW. Because they can operate independent of power grids, they can be used for remote

Figure2.1-2. Dish-Stirling System (NREL)

applications as well as grid connected applications. At the present time, there are no operating commercial dish-Stirling power plants.

2.1.1.3 Power Tower Systems

A power tower uses thousands of sun-tracking mirrors called heliostats to redirect DNI to a receiver at the top of a tower. The receiver at the top of the tower either generates steam directly, or heats a molten nitrate salt HTF to generate steam. The steam is used in a conventional turbine generator to produce electricity. Molten nitrate salt has superior heat transfer and energy storage capabilities, but is more expensive and difficult to work with. Systems with air as the working fluid in the receiver or power system have also been explored in international research and development programs. Commercial power tower plants can be sized to produce anywhere from 50 to 200 MW of electricity. Figure 2.1-3 is a photograph of the 10 MW Solucar PS 10 plant in Spain, a direct steam generation system.

An advantage of power tower plants is that molten salt can be heated to 1,050°F, with steam generation at 1,000°F, which is utility-standard main steam temperature. This results in slightly higher cycle efficiency than is achievable with the lower temperature (about 700°F) steam produced in a trough system. Furthermore, power towers have the advantage that the molten salt is used both as the HTF and as the storage medium, unlike the trough system which uses high temperature oil as the HTF, and requires oil-to-salt and salt-back-to-oil heat exchange for thermal storage. The result is that storage is less costly and more efficient for power towers than for troughs.

2.1.1.4 Compact Linear Fresnel Reflector (CLFR)

The compact linear Fresnel reflector (CLFR) is a solar thermal technology in which rows of mirrors reflect solar radiation on a linear receiver located on towers above the mirror field. A CLFR system is shown in Figure 2.1-4.

Figure 2.1-4. Liddell Phase 1 CLFR Demonstration System

Figure 2.1-3. Solucar PS 10 Tower (Solucar)

11

In the CLFR, collector mirrors rotate on the linear axis parallel to the receiver, following the sun’s movement throughout the day. The CLFR is similar to the more commercially mature solar parabolic trough systems in that it uses one-axis tracking to focus solar radiation on a linear receiver. However, the CLFR has major difference from the trough system. These include:

The CLFR optics is less stringent than optics of a trough. This allows a less expensive collector/receiver system.

The CLFR receiver does not move, such that no flexible hoses or ball joints are required as in a trough system.

The CLFR is more compact in terms of land use. A CLFR may have a ground cover ratio (GCR), which is the ratio of mirror area to land area, of about 70 percent versus a GCR of about 30 percent for a trough.

The saturated steam generated by the CLFR is relatively low temperature and being saturated, rather than superheated, results in less efficient power generation.

The overall CLFR solar to steam efficiency is substantially lower than trough.

2.1.1.5 Solar Chimney

Unlike other solar thermal technologies, solar chimneys do not generate power using a thermal heat cycle. Instead, they generate and collect hot air in a large (several square miles) greenhouse. A tall chimney is located in the center of the greenhouse. As the air in the greenhouse is heated by the sun, it rises and enters the chimney. The natural draft produces a wind current that rotates a collection of dozens of ground mounted air turbines.

A prototype solar chimney was constructed in Spain in the early 1980’s and operated for seven years. The tower height was 195 meters with a diameter of 10 meters and a greenhouse collection area of 46,000 m2 or 11 acres. It generated 50kW. The first large-scale solar chimney project was proposed in Australia. This 200MW facility would have a chimney 1 km tall and a collector 5 km in diameter.

2.1.1.6 Performance Characteristics

The only solar thermal technology commercially available today is parabolic trough. In addition, much of the commercial development interest appears to be for trough technology. Representative characteristics for a parabolic trough system without energy storage are shown in Table 2.1-1.

12

Table 2.1-1. Parabolic Trough Basic Performance Information

Typical Duty Cycle Peaking-Intermediate

Typical Plant Capacity (MW) 200 MW

Integrated Storage None

Capacity Factor (location dependent) 26% to 29%

2.1.2 Solar Photovoltaic

Due to its high cost, high variability, and low capacity factor, solar photovoltaics (PV) have had little penetration into the bulk electricity market. While solar, in general, represents a very small portion of the overall electricity generated in the NERC region, solar PV represents an even smaller fraction. However, there is recent strong growth being observed in the PV industry.

2.1.2.1 Operating Principles

Solar PV converts sunlight (also known as insolation) directly into electricity. The power produced depends on the material involved and the intensity of the solar radiation incident on the cell. Single or polycrystalline silicon cells are most widely used today. Single crystal cells are manufactured by growing single crystal ingots, which are sliced into thin cell-size material. The cost of the crystalline material is significant. The production of polycrystalline cells can cut material costs, but with some reduction in cell efficiency. Thin film solar cells are made from layers of semiconductor materials only a few micrometers thick. These materials make applications more flexible, as thin film PV can be integrated into roofing tiles or windows. Thin film cells significantly reduce cost per unit area, but also result in lower efficiency cells. Gallium arsenide cells are among the most efficient solar cells and have other technical advantages, but they are also more costly and typically are used only where high efficiency is required even at a high cost, such as space applications or in concentrating PV applications. Additional advanced technologies are under development including dye sensitized solar cells (DSSC) and organic light emitting diodes (OLED). Developers of these technologies hope to achieve dramatic reductions in cell cost, but likely will have efficiencies on the lower end of the range for PV cells.

A new development in the solar market has been the growth of larger, utility-scale systems. In the past, photovoltaics had been seen as a distributed technology suitable for rooftops and industrial applications.

2.1.2.2 Concentrating Solar Photovoltaic Systems

Concentrating photovoltaic (CPV) plants provide power by focusing solar radiation onto a photovoltaic (PV) module, which converts the radiation directly to electricity – Figure 2.1-5. Either mirrors or lenses can be used to concentrate the solar energy for a CPV system. Most of the CPV systems use two axes tracking to achieve point focus images on PV cells. Single axis,

13

line focus CPV systems have been built, but do not appear to have the long term commercial potential that the two axis tracking CPV systems have.


Table 2.1-2 shows the costs and performance for solar photovoltaic systems.

Figure 2.1-5. Flat Acrylic Lens Concentrator with Silicon Cells (NREL)

Table 2.1-2. Photovoltaic Performance

Typical Duty Cycle Peaking-Intermediate

Typical Plant Capacity (MW) 20 MW

Integrated Storage None

Capacity Factor (location dependent) 25% to 30%

2.1.3 Wind

Wind power systems convert the movement of air to power by means of a rotating turbine and a generator. Wind power has been among the fastest growing energy sources over the last decade, with around 30 percent annual growth in worldwide capacity over the last five years. Cumulative worldwide wind capacity is now estimated to be more than 94,000 MW.

Typical utility-scale wind energy systems consist of multiple wind turbines that range in size from 1.5 to 2.5 MW. Wind energy system installations commonly total 5 to 300 MW, although the use of single, smaller turbines is also common in the United States for powering schools, factories, water treatment plants, and other distributed loads. Furthermore, offshore wind energy projects are now being built in Europe and are planned in the United States, encouraging the development of larger turbines (up to 5 MW) and larger wind farm sizes.

Wind generation average capacity factors generally range from 25 to 40 percent. The capacity factor of an installation depends on the wind regime in the area and energy capture characteristics of the wind turbine. Capacity factor directly affects economic performance; thus, reasonably strong wind sites are required for cost-effective installations. Figure 2.1-6 shows a wind farm in the Palm Springs area of California.

14

15

Given that the focus of the rest of this NERC report will be heavily on wind generation, we will be very brief in our presentation of wind technologies.


Table 2.1-3 provides typical characteristics for a 100 MW wind farm. The low end of the capacity factor range represents moderate class 3-4 wind sites, while the higher estimates are representative of class 5-6 wind sites. Significant gains have been made in recent years in identifying and developing sites with better wind resources and improving turbine performance and reliability. As a result, the average capacity factor for all installed wind projects in the United States has increased from about 24 percent in 1999 to over 32 percent in 2005.

Figure 2.1-6. Wind Farm near Palm Springs, California (Black & Veatch)

Table 2.1-3. Wind Technology Characteristics

Type Onshore Offshore

Typical Duty Cycle As Available As Available

Typical Plant Capacity (MW) 100 200

Capacity Factor (location dependent) 25% to 40% 35% to 45%

2.1.4 Ocean Power

Ocean based renewable energy is still in early stages of concept design and development in comparison to other established renewable energy options. A number of large scale devices have been tested in the offshore environment; however there have currently been no commercial installations. Ocean power technologies may be divided into two main categories: Tidal power and Wave power.

2.1.4.1 Tidal Power

Tidal power sources have a general advantage that they are generally very predictable especially when compared to wind, wave or solar generation. Significant tidal stream currents generally occur where large tidal flows are forced through relatively narrow boundaries. Thus both high tidal ranges and narrow channels are generally required to cause significant tidal stream currents. However, due to local site conditions a high tidal range does not always indicate high tidal currents and similarly low tidal ranges do not always indicate low tidal currents.

The four main categories that characterize tidal stream devices currently under development, as determined by the “prime-mover” (or principle defining characteristic) are as follows:

Horizontal Axis Axial Flow Turbine (HAA)

Vertical Axis Cross Flow Turbine (VAC)

Oscillating Hydrofoil (OH)

Venturi Devices (V)

The mechanical energy from the prime-mover may be converted to electricity via a number of conversion steps (e.g. hydraulic, direct electrical, mechanical) embodied in a “power-train”.

There are in the region of 50 developers worldwide at varying stages however it is beyond the scope of this project to describe them all therefore; however, a couple of examples of horizontal axis axial flow (HAA) turbines are included - which have both been tested offshore.

2.1.4.1.1 Clean Current

Clean Current has been developing tidal technology for 6 years. Their tidal stream device is a bi-directional ducted horizontal axis turbine. It has a direct drive variable speed permanent magnet generator and therefore only incorporates one moving part. The support type is not specified although, as the device is fully submerged, it is likely to be a monopile, support frame, or gravity base.

Since inception, Clean Current has followed a defined development plan, which began with the testing of two prototypes in 2002 and 2003 which were used to validate the concept, see Figure 2.1-7.

Figure 2.1-7. Clean Current (Black & Veatch)

16

2.1.4.1.2 Marine Current Turbines - Seagen

The Marine Current Turbines (MCT) “Seagen” device is a commercial demonstrator that has twin axial flow rotors, between 15 and 20m in diameter (Figure 2.1-8) which drive the generator (via a gearbox). Each rotor consists of two blades which are pitch controlled to optimize the efficiency of the device. The rotors are fixed onto a horizontal bridge which is attached to a surface piercing monopile and the movement of the bridge up and down the monopile allows the rotors to be raised and lowered for maintenance.

Figure 2.1-8. Marine Current Turbines, SeaGen (Black & Veatch)

2.1.4.1.3 Performance Characteristics

Table 2.1-4 provides typical characteristics for a 100 MW tidal farm. Generic data has been provided at this stage due to the lack of commercially available data.

Table 2.1-4. In-stream Tidal Technology Characteristics

Type Generic offshore

Typical Duty Cycle As Available

Expected Net Plant Capacity (MW) 100

Capacity Factor 25% to 45%

2.1.4.2 Wave

Serious research into the use of wave energy as a viable form of power generation dates back to the 1970s with a large number of Wave Energy Converter (WEC) devices having been

17

18

developed since. There are a considerable number of wave energy devices in design and development stages; however, they cannot all be covered in this report. We have included information on two offshore devices which have been tested in the offshore environment.

2.1.4.2.1 Pelamis

Pelamis is an attenuator device and consists of four tubular sections, connected by three hinged modules. As a wave passes, the four tubular sections move relative to each other causing movement in the hinge modules. The modules convert this motion by means of an internal hydraulic power conversion system. The design has inherent survivability with a very small frontal area subjected to the hydrodynamic forces of incident waves. The Pelamis is anchored by a slack mooring system which allows the device to weathervane into the dominate wave direction. The device is 120m long with a 3.5m diameter, and weight of 700 tones when fully ballasted. The rated power of the device is 750kW (i.e. 250kW per module).

Figure 2.1-9. Pelamis

Wave Generator

(Black & Veatch)

2.1.4.2.2 PowerBuoy

PowerBuoy is a free floating point absorber device which is moored to 3 buoys. The device can be deployed in relatively deep waters, from 35m to 60m in depth. Each unit has a rated power of 40kW, a diameter of 5m, a stroke of 3 m, and weighs 26 tons, see Figure 2.1-10.

2.1.4.2.3 Performance Characteristics

Table 2.1-5 provides typical characteristics for a 100 MW wave farm. Generic data has been provided at this stage due to the lack of commercially available technology today.

Figure 2.1-10. PowerBuoy Wave Generator (Black & Veatch)

Table 2.1-5. Wave Energy Converter Characteristics.

Type Generic offshore

Typical Duty Cycle As Available

Expected Net Plant Capacity (MW) 100

Capacity Factor 25% to 45%

2.2 Drivers for Future Growth of Renewable Generation in the United States

This section covers drivers for future growth of renewable generation in the Unites States and Canada. Northern Baja California, Mexico which is part of NERC, is not considered in this section.

2.2.1 Summary of US Drivers for Future Growth

State RPS policies are one of a number of drivers for renewable energy capacity expansion. Other significant motivators include federal and state tax incentives, state renewable energy funds, utility integrated resource planning, voluntary green power markets, and the economic competitiveness of renewable energy relative to other generation options. Disentangling these various drivers is – to put it mildly – challenging. This task is further complicated by the fact that experience with state RPS policies remains somewhat limited.2

2.2.1.1 Renewable Portfolio Standards (RPS) Legislation (State and Federal)

2.2.1.1.1 State Legislated Renewable Portfolio Standards (RPS)

Considerable RPS legislation has been enacted at the state level. Figure 2.2-1, below, shows a state-by-state breakdown of RPS in the US. In terms of renewable energy growth, state RPS requirements have been the primary driver of new projects in the US.

2 Source: “Renewable Portfolio Standards in the United States”, Wiser and Barbose, Lawrence Berkeley National Laboratory, April 2008.

19

WI: 10% by 2015

NV: 20% by 2015

TX: 5,880 MW by 2015

PA: 8% by 2020

NJ: 22.5% by 2021CT: 23% by 2020

MA: 4% by 2009

ME: 40% by 2017

NM: 20% by 2020 (IOUs)10% by 2020 (co-ops)

CA: 20% by 2010

MN: 25% by 2025Xcel: 30% by 2020

IA: 105 MW by 1999

MD: 9.5% by 2022

RI: 16% by 2019

HI: 20% by 2020

AZ: 15% by 2025

NY: 24% by 2013

CO: 20% by 2020 (IOUs)10% by 2020 (co-ops and munis)

MT: 15% by 2015

DE: 20% by 2019

DC: 11% by 2022

WA: 15% by 2020

NH: 23.8% by 2025

OR: 25% by 2025 (large utilities)5-10% by 2025 (smaller utilities)

NC: 12.5% by 2021 (IOUs)10% by 2018 (co-ops and munis)

IL: 25% by 2025

Mandatory RPS

Non-Binding Goal

VT: Load growth by 2012up to cap of 10%

ND: 10% by 2015

VA: 12% by 2022

MO: 11% by 2020

WI: 10% by 2015

NV: 20% by 2015

TX: 5,880 MW by 2015

PA: 8% by 2020

NJ: 22.5% by 2021CT: 23% by 2020

MA: 4% by 2009

ME: 40% by 2017

NM: 20% by 2020 (IOUs)10% by 2020 (co-ops)

CA: 20% by 2010

MN: 25% by 2025Xcel: 30% by 2020

IA: 105 MW by 1999

MD: 9.5% by 2022

RI: 16% by 2019

HI: 20% by 2020

AZ: 15% by 2025

NY: 24% by 2013

CO: 20% by 2020 (IOUs)10% by 2020 (co-ops and munis)

MT: 15% by 2015

DE: 20% by 2019

DC: 11% by 2022

WA: 15% by 2020

NH: 23.8% by 2025

OR: 25% by 2025 (large utilities)5-10% by 2025 (smaller utilities)

NC: 12.5% by 2021 (IOUs)10% by 2018 (co-ops and munis)

IL: 25% by 2025

Mandatory RPS

Non-Binding Goal

VT: Load growth by 2012up to cap of 10%

ND: 10% by 2015

VA: 12% by 2022

MO: 11% by 2020

Figure 2.2-1: State RPS Policies and Non-binding Renewable Energy Goals

As shown in Figure 2.2-2, over 50% of (non-hydro) renewable capacity additions in the United States from the late 1990s through 2007 have occurred in states with mandatory RPS policies, totaling roughly 8,900 MW. Since 2002, this percentage rises to over 60%. In 2007 alone, approximately 76% of non-hydro renewable capacity additions came from RPS states.3

These numbers should be viewed with some caution, because they do not assess whether any given facility was constructed because of a state RPS or was, in fact, even eligible for a given state’s RPS. In some RPS states, such as Texas and Iowa, a substantial amount of renewable capacity has been added in recent years that are not directly motivated by those states’ RPS policies, and the data presented in Figure 2.2-2 may therefore overestimate the importance of state RPS policies. On the other hand, most states allow out-of-state generation to count toward their RPS requirements, so renewable capacity built in a non-RPS state may be used to meet another state’s mandate; since the data presented in Figure 2.2-2 are unable to take account of this effect, those data may understate the importance of state RPS policies. Finally, because RPS policies have often been established in states with reasonably strong renewable resource potential, it is perhaps not surprising that a good fraction of the renewables development in the U.S. has occurred in those states.

3 Source: Wiser and Barbose

20

Cumulative Capacity

13,000

16,000

19,000

22,000

25,000

28,000

31,000

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Na

me

pla

te C

ap

ac

ity

(M

W)

RPS

non-RPS

Annual Capacity Additions

0

1,000

2,000

3,000

4,000

5,000

6,000

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Na

me

pla

te C

ap

ac

ity

(M

W)

RPS

non-RPS

Figure 2.2-2: Cumulative and Annual Renewables Capacity in RPS and Non-RPS States4

Though it is somewhat unclear whether the data presented here under- or over-estimate the importance of state RPS policies, it is evident that these policies are beginning to have an important impact on new resource development.5 And, because many of these policies have only recently been enacted, renewable energy contracting has only just begun. In California alone, for example, the state’s investor- and publicly-owned utilities have contracted for more than 7,000 MW of new renewables capacity since the RPS was enacted in 2002 through 2007, but just 1,100 MW of this capacity was online at the end of 2007. Future Impacts of State RPS Policies Are Projected to be Relatively Sizable6

The impacts of state RPS programs are expected to expand in the long term as renewable purchase obligation increase, though the magnitude of that growth depends on whether the RPS policies are implemented fully, whether cost caps are limiting, whether entities elect to make alternative compliance payments, and whether new renewable energy projects would have come on line absent the support of state RPS policies.

Ignoring these complexities, and assuming that full compliance is achieved, Berkeley Lab estimates that over 60 GW of cumulative, new renewable energy capacity may be needed by 2025 to fully meet the existing state RPS policies (see Figure 2.2-3), including 4 GW already required by 2007, a cumulative 14 GW by 2010, and a cumulative 32 GW by 2015. The 60 GW figure increases to over 62 GW if one also includes the non-binding renewable energy targets established in Missouri, North Dakota, Vermont, and Virginia, and to nearly 77 GW if one includes the longer-term, non-binding renewable energy goals in California, Iowa, and Texas.

The largest markets, in terms of capacity growth requirements, are projected to be California, Illinois, Minnesota, Texas, New Jersey, and Arizona, each of which would require over 3,000 MW of additional renewable energy capacity by 2025 to achieve full compliance. As a proportion of expected statewide retail sales in 2025, however, leading states are somewhat

4 Non-solar data for 1998-2006 were sourced from EIA Form-860; wind data for 2007 were from AWEA; biomass and geothermal data for 2007 were from Global Energy Decisions; and solar data for all years were from Larry Sherwood (Interstate Renewable Energy Council) and known installations of solar thermal electric facilities. 5 Research at Berkeley Lab confirms this to some degree. In particular, Berkeley Lab estimates that, from 2001 through 2007, roughly 65% of the total wind additions in the U.S. were motivated, at least in part, by state RPS policies. 6 Source for this sub-section: Wiser and Barbose

21

22

different, and include Minnesota, Oregon, Connecticut, New Jersey, New Hampshire, New Mexico, and Delaware, each of which will require 15% or more of statewide load in 2025 to come from new renewable generation. Some of the leading states in terms of required capacity additions, including Texas, require rather moderate additions on a percentage-of-load basis.

Figure 2.2-3: New Renewable Energy Required Meeting Existing RPS Policies7

Though the eventual market impacts of existing state RPS policies are uncertain, there is little doubt that the aggregate amount of new renewable energy generation required under these policies is significant. The estimated 58 GW of new renewable energy capacity equates to an additional 4.8% of total projected electricity generation in 2025, for example, compared to a non-hydro share of 2.1% in 1999 and 2.4% in 2006. In this scenario, 16% of the roughly 1,434 TWh of demand growth expected by EIA from 2000 though 2025 would come from new renewable generation required under existing state RPS policies. Even with this growth, however, non-hydro renewables would continue to provide a relatively modest contribution to U.S. electricity supply: adding the estimate of new renewable generation required by existing state RPS programs from 2000 to 2025 to the 1999 base amount of non-hydro renewables sums to just 6% of total projected electricity generation in the U.S. by 2025. 7 Data used to generate this figure were derived by applying RPS percentage obligations in each state to LBL’s projection of obligated retail sales, and deducting expected contributions from existing renewable generation. The figure may overstate new renewables needed to fully meet state RPS policies to the extent that more-aggressive energy efficiency programs reduce load growth, or if LSEs use out-of-state existing renewable generation to a greater extent than assumed here. Note that the new renewable generation required under the Maryland and Washington, D.C. RPS policies is assumed to come exclusively from those states’ solar set-asides, with all remaining RPS requirements in those two states projected to be met by existing resources.

23

State RPS policies are primarily supporting wind power, but some resource diversity is apparent.8 Of the more than 8,900 MW of new non-hydro renewable energy capacity that has come on line in RPS states from 1998 through 2007, roughly 93% has come from wind power, with biomass (4%), solar (2%), and geothermal (1%) playing lesser roles (see Figure 2.2-4).

Annual Capacity Additions

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Na

me

pla

te C

ap

ac

ity

(M

W)

Total Capacity Additions (1998-2007)

Solar

Geothermal

Biomass

Figure 2.2-4: Renewable Energy Capacity Additions in RPS States9

Though renewable resource diversity has so far been limited, there is some evidence that diversity may increase over time as RPS policies expand, at least in some states. In California, for example, of the more than 7,000 MW of contracts for new or repowered renewable energy projects signed from 2002 through 2007 by the state’s IOUs and POUs, 58% of the total capacity are wind, 23% solar, 12% geothermal, 7% biomass/MSW, and less than 1% small hydro and ocean energy, demonstrating a greater level of diversity than historical, national trends. Additionally, largely because of technology tiers that exist in a number of states, a growing amount of solar energy is being motivated by RPS obligations (see following section).

Solar-specific RPS designs are becoming more prevalent.10 Because of concerns that traditional RPS programs – in which all eligible renewable technologies compete – are likely to benefit only the least-cost projects, an increasing number of states have begun to design their RPS programs to provide differential support to promising but (currently) higher-cost renewable technologies or applications. Typically, this support has been provided through either credit multipliers, in which favored renewable technologies are given more credit towards meeting RPS requirements than are other technologies, or through set-asides, in which some fraction of the RPS must be met with favored technologies.

Set-asides and credit multipliers have been used to support an array of favored technologies, applications, project locations, and vintages. The most popular use of these mechanisms,

8 Source for this sub-section: Wiser and Barbose 9 Non-solar data for 1998-2006 were sourced from EIA Form-860; wind data for 2007 were from AWEA; biomass and geothermal data for 2007 were from Global Energy Decisions; and solar data for all years come from Larry Sherwood (Interstate Renewable Energy Council) and known installations of solar thermal electric facilities. 10 Source for this sub-section: Wiser and Barbose

Solar2%

Geothermal 1%

Wind93%

Biomass4%

Wind

24

WA: 2x multiplier for DG

however, has been to support central and distributed solar energy specifically, or customer-sited distributed generation (DG) more generally.11

Set-asides for solar or DG exist within 12 of the 26 U.S. RPS programs (see Figure 2.2-5). Four of these states combine credit multipliers of some form with these set-asides. Credit multipliers have become somewhat less popular in recent years, and only two states – Texas and Washington – now use credit multipliers without an accompanying mandatory set-aside. The popularity of set-asides for solar or DG, on the other hand, has increased dramatically in recent years. In 2007 alone, new solar or DG set-asides were created in Delaware, Maryland, New Hampshire, New Mexico, and North Carolina, and the previously-established solar set-aside in Colorado was effectively expanded though an increase in that state’s overall RPS target.

Among those states with set-asides, two are restricted to photovoltaic (PV) applications, nine also allow solar-thermal electric technologies to qualify, three allow solar heating and/or cooling to qualify12, and three states have DG set-asides in which solar PV can compete with other forms of renewable DG (see Table 2.2-1). The policies also differ in their targets and timeframes, geographic scope of project eligibility, use of cost caps and alternative compliance mechanisms, and degree of regulatory oversight on solar contracting. Most of these set-asides have yet to take effect, however, and only Arizona, Nevada, and New Jersey have more than three years of operational experience.

Figure 2.2-5: Differential Support for Solar Energy in State RPS Policies

Table 2.2-1: Design Elements of State Solar and DG Set-Asides

11 Those states seeking to support solar within an RPS will also need to address issues of eligibility (Are all forms of solar electricity eligible? Are customer-sited generators eligible? Are metering and tracking systems in place?) and REC ownership (Do owners of solar systems own their RECs? Do mechanisms exist to trade small quantities of RECs?). 12 In addition to Arizona, Nevada, and North Carolina, which allow solar heating and/or cooling to qualify for their solar/DG set-asides, a number of other states allow solar heating and/or cooling to qualify for their overall RPS target, including: Delaware, Hawaii, Illinois, New Hampshire, Pennsylvania, and Texas.

NV: 1% solar by 20152.4x multiplier for central PV2.45x multiplier for distributed PV

PA: 0.5% PV by 2020

NJ: 2.12% solar electric by 2021

AZ: 4.5% customer-sited DG by 2025 (half from residential)

NY: 0.1542% customer-sited DG by 2013

CO: 0.8% solar electric by 2020 (half from customer-sited projects)1.25x multiplier for in-state projects3x multiplier for co-ops and munisfor solar installed before July 2015

DC: 0.386% solar electric by 20211.1x multiplier for solar 2007-09

NM: 4% solar electric by 2020, 0.6% DG by 2015

DE: 2.005% PV by 2019 3x multiplier for PV installed before 2015

MD: 2% solar electric by 2022

Set-aside

Multiplier

NC: 0.2% solar by 2018

NH: 0.3% solar electric by 2014

Set-aside with multiplierTX: 2x multiplier for all non-wind

WA: 2x multiplier for DG

NH: 0.3% solar electric by 2014

NY: 0.1542% customer-sited DG by 2013NV: 1% solar by 20152.4x multiplier for central PV2.45x multiplier for distributed PV NJ: 2.12% solar electric by 2021 DE: 2.005% PV by 2019

3x multiplier for PV installed before 2015

PA: 0.5% PV by 2020

MD: 2% solar electric by 2022

DC: 0.386% solar electric by 20211.1x multiplier for solar 2007-09

CO: 0.8% solar electric by 2020 (half from customer-sited projects)1.25x multiplier for in-state projects3x multiplier for co-ops and munisfor solar installed before July 2015

AZ: 4.5% customer-sited DG by 2025 (half from residential) NC: 0.2% solar by 2018

NM: 4% solar electric by 2020, 0.6% DG by 2015Set-aside

Multiplier

Set-aside with multiplierTX: 2x multiplier for all non-wind

Resource Eligibility

State First Compliance Year

Photovoltaics

Solar Thermal Electric

Solar Heating and/or Cooling

Non-PV Dist. Generation

Arizona 2001 ● ● ● ● Colorado 2007 ● ● Delaware 2008 ● Maryland 2008 ● ● Nevada 2003 ● ● ● New Hampshire

2010 ● ●

New Jersey 2004 ● ● New Mexico 2011 ● ● ● New York 2006 ● ● North Carolina

2010 ● ● ●

Pennsylvania 2006 ● Washington D.C.

2007 ● ●

Despite the nascent state of these policies, solar and DG set-asides, in combination with state and federal incentives, have already begun to have a significant impact on the grid-connected PV market in the United States, as shown in Figure 2.2-6. Since 2000, New Jersey has represented the largest solar set-aside-driven PV market in the United States, although Nevada and Colorado emerged as equally-significant solar set-aside markets in 2007. Additional contributions to grid-connected PV additions in states with solar set-asides have come from Arizona and, more recently, New York. In total, from 2000 through 2007, 102 MW of grid-connected PV capacity was added in states with solar set-asides, representing 22% of all grid-connected PV installations in the U.S. over this period, and 75% of all grid-connected PV additions outside of California, the country’s largest market.

The impact of solar and DG set-asides is not restricted to PV. In fact, the nation’s only two solar-thermal electric plants built since 1991 – a 1 MW facility in Arizona commissioned in 2006 and a 64 MW plant in Nevada commissioned in 2007 – have also been motivated by state solar set-asides. More generally, solar-thermal electric development does not, in some states, appear to require a solar set-aside. In California, for example, a number of such projects are in development, driven by a more –traditionally designed RPS, without a solar set-aside13 (Table 2.2-2).

13 California’s RPS has resulted in 15-29 MW of utility-scale PV contracts (range reflects expansion options). Separately from the RPS, California has enacted aggressive financial incentive programs that intend to support 3,000 MW of customer-sited solar PV by 2017. Solar-thermal electric development does not, in some states, appear to require a solar RPS set-aside. Solar-thermal

25

26

0

10

20

30

40

50

60

70

2000 2001 2002 2003 2004 2005 2006 2007

An

nu

al G

rid

-Co

nn

ec

ted

PV

Ins

talla

tio

ns

O

uts

ide

of

Ca

lifo

rnia

(M

W)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Pe

rce

nt

of

An

nu

al G

rid

-Co

nn

ec

ted

P

V In

sta

llati

on

s (

%)

DelawareMarylandDistrict of Columbia Percent of annual US PV installations

(excluding California) [right axis]PennsylvaniaColoradoNevadaNew YorkArizonaNew Jersey

Percent of annual US PV installations [right axis]

Figure 2.2-6: Early Impacts of State RPS Solar and DG Set-Asides for Grid-Connected PV Installations14

The impacts of RPS solar set-asides on solar development will continue to grow as a greater number of the existing set-asides take effect and as targets increase over time.

Figure 2.2-7 and Table 2.2-2 present Berkeley Lab estimates of the solar electric capacity (including PV and solar thermal electric) that would be required to fully achieve existing state solar and DG RPS set-aside policies. Changes in federal tax incentives, binding RPS cost caps, force majeure events, and other barriers will – in reality – challenge the full achievement of these policies.15 As such, the estimates presented here should be considered a reasonable, if uncertain, estimate of the potential impact of these set-asides under an aggressive assumption of full compliance.

electric developments in California, for example, are occurring under a more-traditionally designed RPS, while activity in Florida appears to be driven by other factors altogether. 14 PV installation data from 2000-07 were provided by Larry Sherwood (Interstate Renewable Energy Council). For the purpose of assigning state PV installations to set-asides, we include installations in the year before the first set-aside compliance date. Data are presented in direct-current units, at Standard Test Conditions. 15 Actual impacts will be affected not only by whether full compliance is achieved, but also by future load growth, the competitiveness of solar energy in broader DG set-asides, the relative contribution of different types of eligible solar technologies, and other factors.

Table 2.2-2: Status of Utility-Scale Solar-Thermal Electric Facilities Proposed in the U.S.

Power Purchaser

Developer State Project Size

Status Motivation

Nevada Power Acciona Nevada 64 MW Operational Solar set-aside

Arizona Public Service

Acciona Arizona 1 MW Operational Solar set-aside

Pacific Gas & Electric

Solel California 554 MW Contracted General RPS


Ausra California 177 MW Contracted General RPS


BrightSource California 500 – 900 MW

Contracted General RPS

Southern California Edison

Stirling Energy Systems

California 500 – 850 MW


San Diego Gas & Electric

Stirling Energy Systems

California 300 – 900 MW



Bethel California 49 MW Contracted General RPS


Bethel California 49 MW Contracted General RPS

Florida Power & Light

Ausra Florida 10 – 300 MW

Announced Not stated

Arizona Public Service

Abengoa Arizona 280 MW Announced General RPS

Notes: Table does not include facilities announced by developers, unless a purchaser of the power has been identified. In addition to the specific facilities listed here, a number of utilities in the Southwest have issued a 250 MW RFP for central station solar power, and Colorado’s major IOU (Xcel Energy) has announced preliminary plans for a 200 MW facility.

27

28

Figure 2.2-7: Solar Capacity Required by State RPS Solar and DG Set-Asides16

Even with these caveats, the estimates presented here demonstrate the potential importance of these set-aside policies for the solar market in the coming decades. As shown, a cumulative 550 MW of solar capacity may be required by these policies by 2010, growing to 2,200 MW by 2015, 5,300 MW by 2020, and 6,700 MW by 2025. Annual solar additions on the order of 100 MW may be required from 2008 through 2010, rapidly ramping up to nearly 300 MW a year from 2011 through 2014, and then to over 500 MW a year from 2015 to 2021, if full compliance is to be achieved.

The largest set-aside driven solar markets in the long-term, based on required capacity to fully meet state targets, are projected by Berkeley Lab to include Arizona, New Jersey, Maryland, and Pennsylvania. In the next several years, however, significant growth in solar capacity will also be required in New Mexico, Nevada, and Colorado. Finally, as a proportion of expected statewide load in 2025, these set-aside policies are projected to require solar generation shares as high as 3.1% in New Mexico and 2% or more in Arizona, Maryland, and New Jersey again assuming that full compliance is achieved.

Achieving these targets is not assured, however, and a number of policy design issues may constrain the market’s growth. States have developed various types of cost caps, for example, many of which may ultimately become binding, limiting future solar market expansion to levels below those estimated here.

16 Berkeley Lab prepared these estimates using a number of input assumptions regarding expected load growth, capacity factors, compliance exemptions, the share of solar used to meet broader DG obligations, the share of PV and solar-thermal electric used to meet solar requirements, and other factors. Data are presented in direct-current units, at Standard Test Conditions.

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

20

08

20

09

20

10

20

11

20

12

20

13

20

14

20

15

20

16

20

17

20

18

20

19

20

20

20

21

20

22

20

23

20

24

20

25

Cu

mu

lati

ve

So

lar

Ca

pa

cit

y (

MW

)

0

100

200

300

400

500

600

700

800

An

nu

al S

ola

r A

dd

itio

ns

(M

W)

NJ

MD

AZCumulative Capacity (left axis)

PA

NM

Annual Capacity (right axis) NC

DE

CO

NV

DC

NH

NY

Table 2-8: Cumulative Solar Capacity Required by State RPS Solar and DG Set-Asides, by State

State 2010

Capacity2025

Capacity

2025 Solar Generation as a %

of State Load

Arizona 110 MW 1,600 MW 2.0%

Colorado 29 MW 160 MW 0.4%

Delaware 0.5 MW 190 MW 1.4%

Maryland 14 MW 1,500 MW 2.0%

Nevada 76 MW 180 MW 0.6%

New Hampshire 4 MW 35 MW 0.3%

New Jersey 210 MW 1,600 MW 2.1%

New Mexico 64 MW 420 MW 3.1%

New York 10 MW 15 MW 0.0%

North Carolina 5 MW 280 MW 0.2%

Pennsylvania 25 MW 690 MW 0.5%

Washington D.C. 0.5 MW 54 MW 0.4%

Total 550 MW 6,700 MW n/a

Additionally, some states – especially those in which retail electric competition exists – continue to struggle with how to encourage appropriate contracting for solar generation, given the political risk of future policy changes. In 2007, New Jersey sought to address this concern by developing plans to transition away from a rebate-based solar market and towards a market primarily supported by solar renewable energy credits. To provide some encouragement for longer-term REC contracting, however, New Jersey established, in advance, an eight-year schedule for solar alternative compliance payment (ACP) levels, therefore removing some uncertainty in the level of future ACPs. Other states, such as Maryland, North Carolina, Colorado, and Nevada, simply require long-term contracting for solar energy or RECs. In other cases, states have mandated or encouraged the use of up-front financial incentives for at least smaller-scale PV systems (and sometimes larger commercial installations as well); this is true in Colorado, Nevada, Arizona, New Jersey, New York, and Maryland.

2.2.1.1.2 Renewable Portfolio Standards (RPS) at the Federal Level

The federal government has supported renewable energy projects with tax relief incentives such as a 30% production tax credit (PTC) that applies to wind generation projects and a 30% investment tax credit (ITC) that applies to solar projects. The limitation of these federal

29

incentives has been the limited duration. Due to the project development cycle of renewable energy projects (typically years) and intensive capital requirements, the limited duration of tax credits (and political maneuvering for extensions) and subsequent uncertainty has led to delays in developing new projects as investors wait for the legislative cycle to provide economic certainty.

2.2.1.2 Carbon constraints

As a non-signatory to the Kyoto Protocol, the US currently does not have any national policy with respect to carbon emissions. However, individual states have taken actions that consequently impose constraints on carbon emissions. Notably, California has restricted electric utilities from signing long term contracts with resources that emit CO2 at a rate greater than a typical combined cycle gas generator, effectively eliminating the possibility of utilities contracting for new coal plants using conventional technology (not having some form of carbon emission control). California also has a pending form of cap and trade system for carbon emissions in the works. In 2007, Kansas regulators rejected an application for a new coal fired station at Holcomb on the grounds that CO2 is a pollutant and existing regulations thereby govern CO2 as they regulate emissions such as NOX and SOX.

The potential for a state-by-state patchwork of regulations concerning carbon emission has led large businesses to call on the Federal government to pass uniform nationwide legislation concerning carbon emissions. Furthermore, many large US corporations with overseas operations in Kyoto-signatory countries have adopted carbon reduction programs for worldwide operations, including facilities located in countries not signed on to the Kyoto Protocol. The United States Climate Action Partnership (http://www.us-cap.org/index.asp) is one such organization that includes both electric utilities and large corporate consumers of electric power “that have come together to call on the federal government to quickly enact strong national legislation to require significant reductions of greenhouse gas emissions”17. The candidates in the 2008 Presidential Election (as of April 2008) have all proposed plans for “cap and trade” systems that would effectively limit US carbon emissions18.

While the US currently does not have federal regulations that limit carbon emissions, the general expectation is that limitations will be enacted in the relatively near-term. With the expectation of carbon constraints in the future, utilities and regulators are pursuing renewable energy sources at increasing rates as one form of hedge against the risk of increased costs for carbon emissions.

2.2.2 Drivers for Future Growth of Renewable Generation in Canada19

2.2.2.1 Carbon Constraints

The Canadian government has set a target of 20% reduction in greenhouse gas emissions by 2020 using a 2006 baseline. However, they’ve left it up to the provinces to determine individual

17 http://www.us-cap.org/index.asp 18 www.hillaryclinton.com, www.barackobama.com, www.johnmccain.com 19 Contributions by the Canadian Wind Interconnection Working Group and Canadian Wind Energy Association are gratefully acknowledged.

30

http://www.us-cap.org/index.asp

http://www.hillaryclinton.com/

http://www.barackobama.com/

http://www.johnmccain.com/

policies. Alberta plans on reducing greenhouse gas emissions by 14% below 2005 levels by 2050.

BC wants to reduce emissions by 33% below 2007 levels by 2020. British Columbia has allocated $1 billion to a variety of green measures. Some of the initiatives include making the public sector carbon neutral by 2010 and expanding biodiesel production and solar thermal energy systems. They’ve introduced a consumer-based carbon tax that applies to virtually all fossil fuels. The proposed rates are based on CAD $10 per ton of carbon dioxide equivalent emissions. Carbon dioxide equivalency is a quantity that describes, for a given mixture and amount of greenhouse gas, the amount of CO2 that would have the same global warming potential over a given period, typically 100 years. Green house gases include carbon dioxide, methane and nitrous oxide. The rates will increase over the next four years to CAD $30 per ton.

2.2.2.2 Sustainable Energy Policy

The government of Canada is committed to renewable energy sources. Towards the end of 2002, the federal government introduced the Wind Power Production Incentive (WPPI) that provides a one cent per kilo-watt hour (kWh) payment for the first 10 years of operation for projects commissioned prior to April 2007. The original target of 1000 MW is fully subscribed. Currently, the federal government has remodeled WPPI as the ecoENERGY for Renewable Power program. Renewable sources, such as wind, biomass, low-impact hydro, geothermal, solar and ocean energy projects commissioned between April 2007 and March 2011 are eligible. This program also provides a one cent per kilo-watt hour (kWh) payment for the first 10 years of operation. The government plans on investing more than CAD $1.5 billion to encourage an increase in renewable supply by up to 4000 MW. To maximize the number of projects supported, the program limits support up to the maximum capacity factors per technology as follows:

biomass energy: 80 percent

hydro energy: 60 percent

wind energy (offshore): 42 percent

wind energy (onshore): 35 percent

photovoltaic energy: 20 percent

Currently, there is close to 10000 MW of projects that have pre-registered for this program. Roughly, 82% of the projects are wind, 12% are small hydro, 5% biomass and 1% geothermal.

Canadian income tax regulations allow taxpayers an accelerated write-off of 50% of the eligible capital costs on a declining basis. Canadian Renewable and Conservation Expenses (CRCE) is a category of fully deductible expenditures, including test wind turbines, associated with the start-up of a renewable energy project. The CRCE expenses are 100% deductible in the year incurred and can be carried forward indefinitely.

In addition to federal incentives, each province in Canada is experimenting with their own wind policies. Most provinces are issuing requests for proposals (RFPs) for specific volumes of capacity and then signing power purchase agreements with the bidder offering the best proposal.

31

Manitoba is evaluating an RFP for 300 MW. The provincial government in Manitoba would like to see 1000 MW in Manitoba with future RFPs proposed to target 200 MW in 2014, 200 MW in 2016 and 200 MW in 2018. A 5 MW community based renewable energy RFP is under consideration. Quebec has set a target of 4000 MW and has issued an RFP for 2000 MW. Another 250 MW RFP for First Nation’s projects and 250 MW for municipalities are in development. The Ontario government has set capacity targets of 5% renewable by 2007 and 10% by 2010. A 500 MW RFP is expected in 2008 in Ontario. The Ontario Power Authority’s 20 year resource plan has set a target of 4685 MW to be installed by 2020. Between 2011 and 2020, 1148 MW is assumed to be installed on the distribution system and 1891 MW on the transmission system.

The provincial governments in the Maritimes are mandating or targeting a certain percentage of demand energy be supplied from new renewable resources. Nova Scotia’s and New Brunswick’s mandates are 20% by 2013 and Prince Edward Island’s is 15% by 2010 (2004 Renewable Energy Act). The PEI government is considering a renewable energy target of 30% by 2016.

British Columbia and Ontario have also introduced feed-in tariffs (standard offer program) for wind projects less than 10 MW connected to the distribution system.

Alberta has a wholesale energy market. There are no specific renewable energy targets and the independent system operator plays no role in centralized generation planning. However, the Alberta government has set a target of 3.5% of total electricity supply be from renewable resources by 2008. All generators may request to connect and compete on the wholesale market. There was a 900 MW wind capacity cap but this has been lifted and replaced with a Market and Operational framework plan that is supposed to allow integration of wind beyond 2000 MW. In the Alberta Electricity System Operator's 20 year generation plan, various generation scenarios have been investigated to help identify potential transmission needs. In the first 10 years, they've assumed between 1600 and 3400 MW of new wind projects and between 2000 and 4000 MW in the next 10 years could be connected.

In general, the support mechanisms have been sufficient to stimulate significant growth in the wind industry in Canada.

2.2.2.3 Characteristics of feed-in tariffs

British Columbia is offering between 6.5 and 7.9 cents/kWh depending on the location and Ontario is offering 11 cents/kWh to all generators except photovoltaic (PV). PV projects are paid 42 cents/kWh and are not eligible for inflation indexation or peak-hour premiums. In BC20, the price is escalated annually based on the consumer price index (CPI). The price is also adjusted based on the time of day and month of delivery. Peak hours in the winter have around a 125% price adjustment and off-peak hours in the summer have around a 75% price adjustment. An additional $3.05/MWh will be added to the price if the project receives EcoLogo certification as low impact electricity. The developer can choose a term of between 20 and 40 years.

20 http://www.bchydro.com/rx_files/info/info54026.pdf

32

In Ontario21, 20% of the base rate of 11 cents/kWh will be indexed based on the CPI. Projects that can reliability operate during on-peak hours (11 am to 7 pm) are eligible for an additional 3.52 cents/kWh. Intermittent (variable) generators are not eligible for this payment. WPPI and ecoENERGY payments are to be shared 50:50 between the developer and the Ontario Power Authority. The contract term is 20 years.

Minimum purchase price regulations have been set in Prince Edward Island that establish the price utilities must pay for power produced by large-scale renewable energy generators (those capable of producing more than 100 kilowatts of energy). The PEI government has set this rate at 7.75 cents per kilowatt-hour, with 5.75 cents of that a fixed rate and 2.0 cents a variable rate that may be adjusted annually to reflect changes in operating costs. The variable rate will be tied to the Consumer Price Index.

2.2.2.4 Green Power Program and Certificates

There are green pricing programs in Canada for direct sales of green power. Enmax offers a Greenmax program built on its McBride Wind plant at a premium of $6.25/month (50% support) or $12.50 month (100% support). Nova Scotia Power offers a similar program that sells green power for $5.00 per 125 kWh. The premium will be used to further develop renewable energy in the province. The Toronto Renewable Energy Cooperative operates wind generators along the Toronto waterfront. An additional $5 per month can be added to Toronto Hydro’s bill for investment in this cooperative. SaskPower offers a green power program at $2.50 per 100 kWh block, with proceeds going towards operating its 33 wind turbines. Bullfrog Power offers Green Power to homes in Ontario and Alberta from EcoLogo certified renewable resources. In Ontario the resources are comprised of 80% low impact hydro and 20% wind and the current price premium is 8.9 cents/kWh. In Alberta, the price premium is 2.0 cents/kWh and the resources are all wind.

There are also green certificate programs offered by various companies in Canada for supporting the development of new green initiatives. Renewable energy credits (RECs) in the U.S. are better defined in states that have set a renewable portfolio standard. Prince Edward Island has taken a first step at exporting both wind power energy and renewable energy credits to the U.S. The PEI government maintains ownership of any carbon credits or offsets based on the wind being owned by the people of PEI and any greenhouse gas mitigation benefits would also belong to the people of PEI. The value of RECs varies depending on the market between $5 and $90/MWh with a value of $20 being typical. Canadian Hydro Developers in Alberta, for example, offer 1 MWh blocks of RECs at $20/MWh.

2.3 Existing Variable Generation Installed Capacity Levels

[TO BE ADDED - ERIC]

21 http://www.powerauthority.on.ca/sop/Storage/32/2804_RESOP_Program_Rules_Version_2.0.pdf

33

2.4 Major Technical Characteristics of Variable Generation Related to Power System Operation

2.4.1 Wind Generation

2.4.1.1 Basic Types of Wind Turbine-Generators

2.4.1.1.1 Type 1: Fixed speed Induction Generator

The simplest form of wind turbine-generator (WTG) in common use is comprised of an induction generator that is driven through a gearbox, as shown in Figure 2.4-1. This type operates within a very narrow speed range dictated by the speed-torque characteristic of the induction generator, as illustrated in Figure 2.4-2. As wind speed varies up and down, the electrical power output also varies up and down per the speed-torque characteristic of the induction generator.

In its simplest form, this type of WTG does not include a pitch control system. The blades have a fixed pitch and are aerodynamically designed to stall (i.e., naturally limit their maximum speed). These are called “stall-regulated” turbines. However, more advanced models include a variable blade pitch control system. The stall regulation feature may be implemented passively (blades stall naturally a wind speeds above a certain magnitude) or actively with action by the blade pitch control system.

If the wind speed increases to a level where steady-state electrical power output would exceed the rated power output of the turbine generator, the pitch-angle of the rotor blades is adjusted to limit power output to the rated value. However, the pitch control system is not fast enough to respond to fast wind gusts. If the wind increases rapidly, the electric power output would temporarily increase above rated power (per the torque-speed characteristic), until the pitch control adjusts the blade pitch angle and reduces power output to the rated value.

One advantage of this type of fixed-speed induction generator WTG is its simplicity. A disadvantage is the significant variation in real and reactive power output as wind speed changes. Simple induction generators always consume reactive power, “under-excited” in the convention of grid connected synchronous generators, with the reactive consumption being primarily dependent on the active power production. Thus, management of reactive power must consider this under-excited behavior as well as the reactive power requirements of the grid.

Figure 2.4-3 shows the reactive power at the terminals of a typical induction generator WTG as a function of real power output. The blue trace shows the reactive power consumed by the induction generator. It ranges from about 0.18 p.u. at no load to nearly 0.50 p.u. at full load. It is common practice to compensate for the reactive power consumption of the induction generator by installing capacitors at the WTG. One approach is to compensate for the no-load reactive power consumption with a fixed capacitor, as shown by the gray curve. Another approach is to use several capacitors and switch them as a function of load. This type of “step compensation” keeps the net reactive power of the WTG near zero or some other desired value.

34

3

GEAR BOX

INDUCTION GENERATOR TRANSFORMER

GRID

Figure 2.4-1. Type 1 WTG; Induction Generator

Figure 2.4-2. Speed-torque and speed-current characteristics for an induction generator (Source: BEW report for CEC, May 2006)

35

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0 0.2 0.4 0.6 0.8 1

Real Power (p.u.)

Rea

ctiv

e P

ow

er (

p.u

.)

Uncompensated

No-Load Compensation

StepCompensation

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0 0.2 0.4 0.6 0.8 1

Real Power (p.u.)

Rea

ctiv

e P

ow

er (

p.u

.)

Uncompensated

No-Load Compensation

StepCompensation

Figure 2.4-3. Reactive Power as a function of Real Power for and Induction Generator WTG, with and without compensation using shunt capacitors

2.4.1.1.2 Type 2: Variable-Slip Induction Generator

The variable-slip induction generator WTG is similar to the Type 1 induction generator machine, except that the generator includes a wound rotor and a mechanism to quickly control the current in the rotor (see Figure 2.4-4). The operating characteristics are similar to the Type 1 induction generator WTG, except that the rotor-current control scheme enables a degree of fast torque control, which improves the response to fast dynamic events and can damp torque oscillations within the drive train.

3

3GEAR BOX

WOUND ROTORINDUCTION GENERATOR TRANSFORMER

GRID

IGBT RControl

Rectifier

Figure 2.4-4. Type 2 WTG; Wound Rotor Induction Generator with Variable Slip (Vestas Opti-Slip®)

36

2.4.1.1.3 Type 3: Double-Fed Asynchronous Generator

The double-fed asynchronous generator (DFAG) type of WTG includes a mechanism that produces a variable-frequency current in the rotor circuit (see Figure 2.4-5). This enables the WTG to operate at a variable speed (typically about 2:1 range from max to min speed), which improves the power conversion efficiency and controllability of the WTG. Since the power converters need only be rated to carry a fraction of the total WTG power output, this design is also attractive from an economic perspective.

Although the original incentive for this scheme was variable speed power conversion, the power converters have since evolved to perform reactive power and voltage control functions, similar to those in conventional thermal and hydro power plants. The fast response of the converters also enables dynamic features such as low-voltage ride-through and governor-type functions.

3

3GEAR BOX

WOUND ROTORINDUCTION GENERATOR TRANSFORMER

GRID

IGBT POWER CONVERTORS

3

Figure 2.4-5. Type 3 Double-Fed Asynchronous Generator, Variable Speed WTG (GE 1.5)

2.4.1.1.4 Type 4: Full Power Conversion variable speed

Another approach to variable speed WTGs is to pass all turbine power through a ac-dc-ac power electronic converter system (see Figure 2.4-5). This system has many similar operating characteristics to the DFAG system, including variable speed, reactive power and voltage control, and fast control of power output. It has an additional advantage of totally decoupling the turbine-generator drive train from the electric power grid, which means that dynamics during grid disturbances can be better controlled (LVRT, governor-type functions, etc.). It also reduces dynamic stresses on drive train components when grid disturbances occur.

37

GEAR BOX

SYNCHRONOUSGENERATOR TRANSFORMER

GRID

Rectifier

33

IGBTInverter

Figure 2.4-6. Type 4; Full Power Conversion, Variable Speed WTG (Enercon, GE Multi-MW)

2.4.1.2 Voltage and Reactive Power Control

FERC order 2003a (and/or?) 661-A22 - article 9.6.1 requires -.95 to +.95 power factor at the Point of Interconnection (POI). This is a recent step in an evolution of generator standards (need the appropriate NERC documents to reference) that define power range requirements at the terminals of individual synchronous generators. Since wind plants consist of multiple WTGs and may include other reactive power equipment, definition of required power factor range at the POI allows non-technology specific means of meeting system performance objectives.

It should be noted that currently, the intent of the this power factor range requirement, has been open to multiple interpretations. Specifically, one widely used interpretation of the rule is that wind plants satisfy the requirement, if the plant power factor remains anywhere within this range during operations. The other interpretation, which we believe is consistent with the intent of the requirement, is that wind plants must be able to deliver controlled reactive power, such that the power factor can be set or controlled to any level within the specified range. This second interpretation is consistent with conventional synchronous generator interconnection. Many wind plants are presently being designed and commissioned subject to the first interpretation in North America.

The other key distinction is that order 2003a places the onus on the host system to prove the need for wind generation to deliver reactive power. System studies must show that delivery of reactive power from proposed wind plants is necessary for system reliability and operation, before requiring such capability of prospective new wind generators. [the subtext being that there is no established mechanism by which host systems can prove such a need and this is starkly at odds with the requirements imposed on other more conventional generators.]

2.4.1.2.1 Power factor control

Power factor control of wind plants is standard practice in much of Europe. In North America, power factor control is usually limited to smaller generators or generation in vicinity of larger

22 Standardization of Generator Interconnection /Interconnection for Wind Energy and other Alternative Technologies {need to get current citations}

38

39

Wind Plant Active Power, PWP (MW)

Win

d P

lant

Rea

ctiv

e P

ow

er,

QW

P (

MV

Ar)

generation responsible for controlling voltage. In weak systems, power factor control can be detrimental to voltage stability.

As noted above, some types of wind generation equipment have no inherent reactive power control. Wind plants with these types of WTG equipment will commonly have a combination of switched shunt capacitors, reactors, and/or SVC-type reactive compensation equipment that is controlled at the substation level to maintain a constant power factor at the POI. Figure 2.4-7 illustrates the net reactive power flowing out of a wind plant into the POI as a function of plant power output. The plot on the left shows a design where the net reactive power increases with plant load, while the plot on the right illustrates a design where net reactive power output decreases with plant load.

Desired Q tomaintain constant

Power Factor

ToleranceBand forSwitchingCapacitors

Net Q Desired Q tomaintain constant

Power Factor

Win

d P

lant

Rea

ctiv

e P

ow

er,

QW

P (

MV

Ar)

Net Q

ToleranceBand forSwitchingCapacitors

PWP (MW)

Figure 2.4-7. Typical reactive power profiles for wind plants with power factor control using switched shunt reactive compensation

2.4.1.2.2 Voltage Control

Power plants are normally required to regulate bus voltage at the point of interconnection. This is normally the high-side bus of the plant’s step-up transformer. Conventional plants with synchronous generators regulate bus voltage by controlling field current with an excitation system.

Wind plants are different. There are two basic schemes for regulating voltage with a wind plant:

By using controlled reactive compensation devices (capacitors, reactors, SVC, STATCOM) in the plant substation, or

By controlling the reactive power output of individual wind turbines

Figure 2.4-8 shows a typical wind plant with induction generators WTGs (Type 1 or 2). These types of WTGs often operate with each WTG holding a constant power factor. The reactive power exchange at the point of common coupling (POI) is controlled by reactive compensation equipment in the substation, usually connected to the low-voltage bus (a combination of switched capacitors, switched reactors, SVC or STATCOM, depending on interconnection requirements).

Figure 2.4-9 shows a typical wind plant with DFAG or full conversion WTGs (Type 3 or 4). These types of WTGs have the capability to quickly and continuously adjust their reactive power output and thereby contribute to regulating voltage at the POI. The scheme depicted in Figure 2.4-9 includes a reactive power controller in the substation that measures voltage at the POI and adjusts the reactive power output of the WTGs to regulate the voltage at the POI. Depending on the requirements of the specific plant, this basic control scheme can be supplemented by switched reactors or capacitors, or LTC into and integrated voltage and reactive power control system. Figure 2.4-10 shows an example of the performance of this type of voltage control scheme at a 160 MW wind plant in the western US.

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QL

QC

SVC

QSVC

HV Bus

LV Bus

Reactive Compensation

PWPQWP

Substation

Point of Interconnection(POI)

ReactivePower

Controller

LTC

Figure 2.4-8. Wind plant with WTGs that operate with constant power factor. Voltage or power factor at POI are controlled by reactive compensation devices in the substation

40

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QWTG

PWTG

QL

QC

HV Bus

LV Bus

Reactive Compensation(if required)

PWPQWP

Substation

Point of Interconnection(POI)

ReactivePower

Controller

LTC

Figure 2.4-9. Wind plant with WTGs that can control reactive power output and regulate voltage

41

42

Voltage at POI

Wind Plant Power Output

Wind Plant Voltage

Figure 2.4-10. Performance of Wind Plant voltage control scheme for a 1-hour period with significant variation in wind power output

2.4.1.2.3 Zero-Power Voltage control

This is a new feature being introduced on full conversion variable speed wind turbines. Turbines equipped with this feature are able to regulate their terminal voltage by controlling the reactive power flowing in their electronic power converters, regardless of whether or not the turbines are rotating and producing real power. This feature is described in Chapter 5.

2.4.1.3 Low-Voltage Ride-Through

LVRT requirement evolved over the past 5+ years, starting with a history of deliberate tripping on low voltage. European practice and experience lead North American practice and resulted in depth-duration curves of the type shown as the blue and green curves in Figure 2.4-11. The current standard requires that wind generation not trip for zero voltage (i.e. bolted 3-phase fault) at the POI for 9 cycles. This latest version of the requirement is often called “zero-voltage ride-though.” The standard also requires tolerance of arbitrarily long duration backup cleared single-phase-to-ground faults.

2.4.1.3.1 High Voltage Ride-Through

There has been considerable recent discussion on high voltage ride-through requirements. These discussions have not had the depth, nor the technical sophistication, of several years of debate about low voltage tolerance.

The proposed limit high-voltage limit (red curve) in Figure 2.4-11 is reasonably interpreted as starting when the voltage exceeds 110% of normal (not when a system fault occurs and initiates a voltage depression). The required HVRT tolerance would reasonably be specified as a cumulative duration of withstand, as is the common and accepted practice for other power system equipment, and not be specified as an "envelope" defined by elapsed time from some initiating event. (A realistic overvoltage event typically has multiple short excursions into the overvoltage domain, and only these excursions are relevant for overvoltage performance.) Such a standard would more appropriately reflect the stress that must be endured by the equipment in terms consistent with overvoltage withstand standards applied to other power system equipment.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

-1 0 1 2 3 4

Time (seconds)

Vo

ltag

e a

t th

e P

oin

t O

f In

terc

on

nec

tio

n (

PU

)

Generator not required to remain on li

Generator not required to remain on line

WECC and NERCLow voltage limit

NERC and ERCOT High voltage limit

Generator required to remain on line

ERCOTLow voltage limit

Pre-Fault Period

Figure 2.4-11. Voltage ride-through criteria from recent proposals by NERC, ERCOT, and WECC

43

2.4.1.4 Active Power Control Functions

2.4.1.4.1 Curtailment

Turbines without pitch control cannot limit their power output. However, wind plants with multiple wind turbines can limit or reduce total plant power output by shutting down some of the turbines in the plant.

Turbines with pitch control are capable of curtailing power in response to a real-time signal from an operator by adjusting the pitch of the turbine blades (i.e., “spilling wind”). Wind plants with such turbines are able to limit or regulate their power output to a set level by controlling the power output on individual turbines, as shown by the multiple red traces in Figure 2.4-12.

Figure 2.4-12. Curtailment of WTG output using blade pitch control (Source: BEW report for CEC, May 2006)

2.4.1.4.2 Power Ramp Rate Control

Double-fed and full conversion wind turbines are also capable of controlling the rate of change of power output in some circumstances, including:

Rate of increase of power when wind speed is increasing

Rate of increase in power when a curtailment of power output is released

Rate of decrease in power when a curtailment limit is engaged

These functions could be implemented either at an individual turbine level or at a plant level.

44

2.4.1.4.3 Frequency Regulation and Reserve Functions

Double fed and full conversion wind turbines are capable of adjusting their power output in real time in response to variations in grid frequency. This would be an optional control feature, implemented in wind plants where participation in grid frequency regulation is deemed necessary.

If frequency increases, the frequency regulation function would reduce power output from the wind turbine, similar to a droop-type governor function in a thermal or hydro generating plant. A wind turbine would always be able to respond to increased grid frequency, since it is always possible to reduce power output below the total available power in the wind.

The frequency regulation function is also capable of increasing power when grid frequency decreases, provided that the turbine’s nominal operating point at nominal frequency is below the total available power in the wind. When operating in this mode (power output curtailed below total available power), the wind turbine would be contributing spinning reserve to the grid.

Examples of overfrequency and underfrequency regulation performance are described below, utilizing data from staged tests at a 60 MW wind farm with forty 1.5 MW double-fed wind turbines.

2.4.1.4.4 Over-Frequency Response

Figure 2.4-13 illustrates the power response of the wind plant due to a grid over-frequency condition. For this test, the controller settings correspond to a 4% droop curve and 0.02Hz dead band. During this test, the site was operating unconstrained at prevailing wind conditions. It was producing slightly less than 23MW prior to the over-frequency condition. The system over-frequency condition was created using special test software that injected and added a 2% controlled ramp offset into the measured frequency signal. The resulting simulated frequency (the red trace in Figure 2.4-13) increased at a 0.25Hz/sec rate from 60Hz to 61.2 Hz. While the frequency is increasing the plant power (the dark trace in Figure 2.4-13) is observed to drop at a rate of 2.4MW/sec. After 4.8 seconds the frequency reaches 61.2 Hz and the power of the plant is reduced by approximately 50%.

The over frequency condition is removed with a controlled ramp down to 60Hz at the same 0.25Hz/sec rate. In response, the plant power increases to its unconstrained power level. This is slightly higher than the unconstrained level prior to the test, due to an increase in the wind speed. The droop and deadband settings for this test are typical values. Settings can be adjusted to meet specific grid and application requirements.

45

Figure 2.4-13. Power response of wind plant to overfrequency condition.

2.4.1.4.5 Under-Frequency and Power Reserve Response

An under frequency condition is simulated using the same test software and the results are presented in Figure 2.4-14. In order to allow for an increase of wind plant active power output in response to an under-frequency condition, some active power production must be kept in reserve. Unlike a conventional power plant, the maximum power production of the wind plant is constrained to that possible with the prevailing wind. For this test, the output of the plant was constrained to 90% of prevailing wind power during nominal frequency conditions, allowing a 10% increase in power with a 4% decrease in frequency. The plant controller continuously calculates the available plant power based on average wind conditions and turbine availability. The controller regulates the output power to 90% (12.4MW) of this calculated value and operates the plant at this level while the system frequency is within +/- 0.02 Hz of nominal frequency (60Hz).

As the system frequency decreases, the control increases the plant power according to the droop schedule. At 57.6 Hz, 4% under frequency, 100% of the calculated available power of the plant is produced (13.8 MW). The power of the plant will remain at this value until either wind conditions reduce or the system frequency increases.

46

47

Figure 2.4-14. Power response of plant to underfrequency condition.

Table 2.4-1. Functional Control Capabilities of Wind Generators by Type

Type 1

Fixed Speed Induction Generator

Type 2

Variable Slip Induction Generator

Type 3

Double-Fed Asynchronous Generator

Type 4

Full Conversion Variable Speed

Constant power factor control Requires SVC or STATCON

Requires SVC or STATCON

Yes Yes

Automatic voltage regulation Requires SVC or STATCON


Yes Yes

Voltage regulation at zero power



Yes Yes

Low-voltage ride-through Requires SVC or STATCON


Yes Yes

Power curtailment Yes, if WTG has pitch control

Yes, if WTG has pitch control

Yes Yes

Power ramp rate control No No Yes Yes

Grid frequency regulation No No Yes Yes

2.4.2 Concentrating Solar Generation (by Eric John)

[TO BE ADDED - ERIC]

2.5 Geographic Diversity Impacts of Variable Generators Diversity on Power System

Operation

This section describes the impact of wind power when it is spread over larger geographic areas. There is considerable diversity among wind turbines within a wind plant, and diversity among geographically disperse wind plants. When this is combined with the diversity in load, the variability of the remaining load to be served after accounting for wind will exceed the variability of load alone, but will be less than the variability of wind alone. Because wind has a larger penetration than other variable generation sources, and because it has been studied more than other technologies, we focus on wind. The difference between wind and solar may be significant, but the approaches for analyzing them are substantially the same with respect to the impact on power system operations.

2.5.1 Geographic Diversity of Wind Power

Wind turbines within a wind power plant are spread apart and therefore do not experience the same wind at the same time. Similarly, different wind plants do not experience the same wind simultaneously either. The impacts of the resulting smoothing of power output, both within a wind plant, and among wind plants, has been studied extensively at the National Renewable Energy Laboratory, using 1-second wind power data from operating wind plants in several parts of the United States. The data from this project has been used to quantify the behavior of wind plants, and this section of the report illustrates the spatial variation and smoothing impacts of larger wind plants and multiple wind plants.

Wind conditions over a short distance are generally similar, therefore outputs from wind turbines and groups of wind turbines that are close to each other often subject to same weather systems with similar wind conditions. The resulting wind output from these turbines would be expected to be similar. For example, if one turbine is operating at 80% of its capacity, other turbines in close vicinity are also expected to operate at about the same level. However, despite close proximity, instantaneous outputs from individual turbines of a large wind power plant are not synchronized. Physical separations and differences of local terrains cause wind speeds at each turbine to vary. Even adjacent wind turbines within the same wind power plant do not experience the same wind conditions during short time frames and their instantaneous outputs are not likely to be synchronized (e.g., outputs from some turbines are increasing while others are decreasing). Such spatial variation in wind speed makes the combined outputs from more turbines less volatile, especially in short time frames. Similar wind conditions will eventually sweep over the entire wind power plant. In longer time frames, outputs of individual turbines or groups of turbines in close vicinity are likely to move in the same direction.

The wind spatial variation and its impact on wind turbine outputs can be analyzed in several ways. Table 2.5-1 lists correlation coefficients among short and longer time frame output fluctuations (step changes) from individual turbines installed along a ridgeline in a small wind

48

plant to illustrate wind power spatial variation. The turbines are identified as #6 through #17 (data from turbine # 11 are not available) in column and row headings. The data used in the table are daily 3-second, 1-minute, 10-minute, and 1-hour output series. For example, the 3-second power step changes from turbine #7 and turbine #12 have a correlation coefficient of –0.003 (negative numbers in the table are enclosed in parentheses). It can be seen that, for 3-second data series, even outputs from turbines next to each other have very poor correlation. When turbines are further apart, their outputs actually move in opposite directions at the same time (numbers in parentheses denote negative correlation coefficients). As the time interval increases, the outputs from those turbines are more in sync. The high correlation coefficients for the 1-hour data series suggest that changes of hourly output from these turbines are practically in lockstep.

49

Table 2.5-1. Individual Turbine Output Correlation Coefficients

3-second #7 #8 #9 #10 #12 #13 #14 #15 #16 #17 #6 0.019 0.002 (0.014) 0.013 (0.027) (0.002) (0.002) (0.009) (0.009) 0.015 #7 0.000 0.012 (0.018) (0.003) 0.016 (0.005) (0.003) (0.007) (0.002)#8 0.023 0.005 (0.018) 0.009 (0.006) 0.001 0.015 (0.001)#9 0.042 (0.011) (0.010) (0.001) 0.007 (0.010) 0.008 #10 0.003 (0.008) 0.009 (0.002) 0.019 0.006 #12 0.014 0.001 0.013 (0.008) (0.018)#13 (0.023) 0.003 (0.010) 0.019 #14 0.012 0.011 (0.005)#15 (0.018) (0.012)#16 (0.009) 1-minute #7 #8 #9 #10 #12 #13 #14 #15 #16 #17 #6 0.492 0.188 0.055 0.043 0.053 0.071 0.055 0.014 0.012 (0.005)#7 0.372 0.116 0.064 0.018 0.067 0.096 0.020 0.043 (0.001)#8 0.459 0.285 0.102 0.111 0.141 0.112 0.072 0.008 #9 0.608 0.163 0.126 0.129 0.117 0.055 0.002 #10 0.164 0.108 0.114 0.124 0.016 0.013 #12 0.003 0.128 0.097 0.018 0.047 #13 0.195 0.109 0.152 0.092 #14 0.360 0.234 0.223 #15 0.501 0.300 #16 0.493 10-minute #7 #8 #9 #10 #12 #13 #14 #15 #16 #17 #6 0.767 0.619 0.555 0.485 0.425 0.362 0.418 0.309 0.400 0.377 #7 0.842 0.659 0.558 0.423 0.373 0.462 0.357 0.493 0.502 #8 0.835 0.685 0.572 0.441 0.449 0.406 0.519 0.486 #9 0.903 0.671 0.443 0.568 0.530 0.571 0.514 #10 0.750 0.497 0.624 0.595 0.619 0.550 #12 0.538 0.578 0.570 0.597 0.433 #13 0.691 0.596 0.615 0.553 #14 0.769 0.716 0.666 #15 0.831 0.776 #16 0.848 1-hour #7 #8 #9 #10 #12 #13 #14 #15 #16 #17 #6 0.953 0.915 0.922 0.918 0.889 0.794 0.910 0.940 0.940 0.916 #7 0.968 0.965 0.966 0.946 0.791 0.937 0.940 0.924 0.882 #8 0.993 0.992 0.972 0.802 0.947 0.942 0.918 0.881 #9 0.996 0.964 0.777 0.933 0.936 0.921 0.889 #10 0.966 0.787 0.938 0.943 0.921 0.885 #12 0.816 0.946 0.926 0.889 0.853 #13 0.856 0.885 0.882 0.839 #14 0.965 0.937 0.919 #15 0.987 0.967 #16 0.980

The smoothing effect of power outputs from multiple wind turbines can easily gauged by examining the wind power changes at different time steps. These step changes are power level differences between two consecutive time steps. Table 2.5-2 lists the average magnitude

50

(absolute value) and the standard deviation values of wind power step changes from four wind installations for different time steps, calculated from a one year data stream. The average magnitudes of step change (absolute) values are calculated because the step changes contain both positive and negative values, and the normal averaging process over a long time series tends to yield zero or very small values (and thus provide no meaningful information). These values are also expressed in the table as respective plant generating (nameplate) capacities to show their relative magnitudes.

Table 2.5-2. Wind Power Step Change Magnitude and Standard Deviation Values

14 turbines 61 turbines 138 turbines 250+ turbines (kW) (%) (kW) (%) (kW) (%) (kW) (%) 1-second Mean 24 0.2 107 0.1 87 0.1 111 0.1 Std

Deviation 42 0.4 160 0.2 154 0.1 200 0.1

1-minute Mean 130 1.2 612 0.8 494 0.5 730 0.3 Std

Deviation 225 2.1 1,038 1.3 849 0.8 1,487 0.6

10-minute Mean 329 3.1 1,655 2.1 2,116 2.0 3,713 1.5 Std

Deviation 548 5.2 2,877 3.6 3,617 3.5 6,418 2.7

1-hour Mean 736 7.0 3,813 4.8 6,081 5.9 11,088 4.6 Std

Deviation 1,124 10.7 6,223 7.9 9,282 9.0 16,746 7.0

It can be seen in Table 2.5-2 that short-term wind power fluctuations are very small. The average magnitudes of second-by-second wind power changes are less than 0.2% of plant capacity. The average 1-minute wind power step changes are only about 0.3%~1.2% of the plant capacity. Table 2.5-2 also shows that as the size of wind power generating capacity increases (more wind turbines in the wind power plant), the magnitudes of step changes do not increase proportionally, and as a percentage of the total generating capacity, the magnitudes of step changes decrease when outputs from more turbines are included. The effect of aggregating wind turbines is clear from the step change value statistics. As more and more wind generating turbines are installed, their combined output becomes less and less variable because of significant spatial diversity in wind conditions.

The relationship between wind plant output power fluctuations is similar. Figure 2.5-1 plots the correlation coefficients of wind power step changes from five pairs of wind power plants with distances between them ranging from 3 km (two turbine strings within the same wind power plant) to about 1,580 km (between Midwest and Texas wind power plants). Figure 2.5-2 plots the correlation coefficients of wind power from the same five pairs of wind power plants.

51

52

Figure 2.5-1 clearly displays the temporal and spatial relationships of wind power. For very short time intervals (1 second to 1 minute), power output fluctuations from all wind power plants are not correlated regardless of the distances that separate them. As the time interval increases, outputs from nearby wind installations tend to move in the same direction, becoming more correlated, while output fluctuations from distant wind power plants are still independent from each other.

Figure 2.5-1 shows the correlation of wind power fluctuations among wind plants that are separated by different distances. Wind power levels from these wind plants have a different correlation. General wind conditions over short distances are likely to be similar while wind conditions over long distances are different. Wind power plants that are hundreds of kilometers apart will have significantly different wind conditions with more spatial variations. Output power levels from wind plants near to each other are more correlated, even in short time frames, while output levels from wind plants far away from each other are not correlated even in longer time frames. Figure 2.5-2 confirms these relationships. In fact, the available data show that the correlation of wind power levels is almost exclusively determined by the distance between wind power plants.

-0 .20

0.00

0.20

0.40

0.60

0.80

1.00

3 40 200 290 490 1580

D istance betw een W ind Farm s (km )

Co

rrel

atio

n C

oef

fici

ents

1 hour

10 m inu te

1 m inute

1-second

Figure 2.5-1. Correlation of wind power fluctuations vs. distance

The data show a high correlation of longer-term power outputs between adjacent wind power plants, whereas high-frequency power variations are statistically independent. These facts facilitate the forecasting of wind power output of large wind plants and lessen the burdens caused by large wind power plants on electric system operations. Figure 2.5-3 gives such an example. Figure 2.5-3 plots 1-minute average power profiles from two wind power plants (Lake Benton and Storm Lake) that are about 200 km apart for a 7-day period (168 hours). The similarity between the two profiles in the figure clearly points out how outputs from these two plants are related during this 7-day period. Daily output correlation coefficient for this period is

0.851. In addition Figure 2.5-3 suggests that a certain temporal relationship exists between the outputs of these two wind power plants. Calculation of cross-correlation coefficients between these two wind power plants reveals more information about this relationship.

Figure 2.5-4 plots the cross-correlation coefficients between these two wind power plants for the first 4 days of the data series. The figure shows a time shift of the Storm Lake power signal from –720 minutes (i.e., advancing the Storm Lake data series 12 hours relative to that of Lake Benton) to +1080 minutes (delaying the Storm Lake data series 18 hours relative to Lake Benton) in 1-minute increments. It can be seen that a strong correlation at +240 minutes (4 hours). This corresponds to the distance between Lake Benton and Storm Lake and wind speed and direction as recorded at the time. Figure 2.5-5 shows the hourly average wind speeds of both locations for the first 96-hour period in Figure 2.5-3.

0 .0 0

0 .1 0

0 .2 0

0 .3 0

0 .4 0

0 .5 0

0 .6 0

0 .7 0

0 .8 0

0 .9 0

1 .0 0

3 40 200 29 0 490 1580

D is tance b etw een W ind Farm s (km )

Co

rrel

atio

n C

oef

fici

ents

1 hou r

10 m inu te

1 m inu te

1 -second

Figure 2.5-2. Correlation of wind power levels vs. distance

53

0

20

40

60

80

100

120

0 24Hours

Pow

er (M

W)

Lake Benton Storm Lake

48 72 96 120 144

he output powerprofiles shown in

Figure 2.5-4. Cross correlation between Lake Benton II and Storm Lake

T

Figure 2.4-3 and the plot of average hourly wind speeds in Figure 2.5-5 show that well-defined weather systems passes through both wind power plants. The weather systems experienced at Lake Benton moved on through Storm Lake after a delay that can be estimated from a cross-correlation plot. Because the same types of turbines are installed at Lake B er plants are similar, the resulting power output from both wind power plants are similar. Meteorologists can predict how fast a weather front travels and when it will reach a certain point. With this knowledge and knowledge of the wind power plant characteristics, the output of the downwind wind power plant can be predicted from the output power of the upwind wind power plant.

enton and Storm Lake, and because the layouts of the wind pow

-1.00

-0.75

-0.50

-0.25

0.00

0.25

0.50

0.75

1.00

-720 -480 -240 0 240 480 720 960

Delay (minutes)

Cro

ss

Co

rre

lati

on

Figure 2.5-4. Cross correlation between Lake Benton II and Storm Lake

54

55

2.5.2 Distribution of

Wind Power Step

Changes

Table 2.5-3 shows that, in addition to small average wind power step changes, standard deviation values are small relative to the mean values. This indicates that step change distributions are concentrated around the mean values. Table 2.5-3 lists the cumulative frequencies of wind power step changes in terms of its standard deviation values (σ). The table shows that large step changes (those step changes with magnitude greater than 5σ) increase as time step lengthens. This is expected because wind can experience bigger changes during longer periods of time. For shorter time steps, such as 1-second and 1-minute, the σ values are small. At 5σ level, step changes are still only about 2% to 10% of the plant capacity. Further investigation shows that none of the 1-second step changes that exceed 10% of plant capacity were wind speed related. It is not easy to determine the cause of large step changes for data series of longer time steps. Many of these large changes were the result of abnormal network conditions, such as forced outages within and without the wind power plant. It is extremely unlikely that all of turbines or even a large number of turbines in a large wind plant would develop a fault at the same time that would cause very a large power level change. These rare but large changes of power level are more critical to system operations because the grid must balance loads with generation at all times.

0.0

2.0

0

6.0

8.0

10.0

12.0

14.0

16.0

0 6 12 18 0 6 12 18 0 6 12 18 0 6 12 18

Hour of Day

pee

d (

m/s

ec)

Lake Benton

Storm Lake

4.W

ind

S

Figure 2.5-5. Average hourly wind speed for the 96-hour period in Figure 3-33

Table 2.5-3. Cumulative Frequency of Wind Power Step Changes

14 turbines 1-second

(σ=42kW) 1-minute (σ=225kW)

10-minute (σ=548kW)

1-hour (σ=1.1MW)

±1σ 0.84862 0.85808 0.83122 0.78389 ±2σ 0.94342 0.95387 0.95466 0.94628 ±3σ 0.97876 0.98337 0.98447 0.98307 ±5σ 0.99727 0.99715 0.99677 0.99831 > ±5σ 0.00273 0.00285 0.00323 0.00169 138 turbines

1-second (σ=154kW)

1-minute(σ=849kW)

10-minute (σ=3.6MW)

1-hour (σ=9.3MW)

±1σ 0.83681 0.83178 0.83543 0.78813 ±2σ 0.93999 0.95598 0.95479 0.94462 ±3σ 0.97842 0.98666 0.98365 0.98339 ±5σ 0.99693 0.99795 0.99637 0.99823 > ±5σ 0.00307 0.00205 0.00363 0.00177 250+ turbines

1-second (σ=0.2MW)

1-minute(σ=1.5MW)

10-minute (σ=6.4MW)

1-hour (σ=16.7MW)

±1σ 0.84126 0.87624 0.84122 0.78629 ±2σ 0.96155 0.97424 0.95850 0.94601 ±3σ 0.99005 0.99253 0.98495 0.98419 ±5σ 0.99896 0.99816 0.99583 0.99803 > ±5σ 0.00104 0.00184 0.00417 0.00197

2.5.3 Wind Plant Ramping Behavior

Step change statistics define the outer boundary of the wind power fluctuations. The rate of change of wind power levels in a given time interval (e.g., 10 minutes or 1 hour) is another indicator of wind power plant behavior. Power levels from a wind power plant change continuously because the wind speed changes continuously as it moves through the wind power plant. For example, in a 1-second wind power data series, the average duration for wind power increases or decreases is only 2.2 seconds (with a standard deviation of 44 seconds). With 1-minute data series, the average duration of the increases and decreases is 2.4 minutes (with a standard deviation 8.4 minute).23 To analyze the ramping behavior of wind power, we calculate the apparent rate of changes of wind power over a fixed time interval. In this report, the apparent 23 Calculated from output power data of the 14-turbine string. For 138-turbine wind power plant, the average durations are 2.2 seconds (standard deviation 37 seconds) with 1-second data series and 2.8 minutes (standard deviation 7.3 minutes) with 1-minute data series.

56

57

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

0:0

0

1:0

0

2:0

0

3:0

0

4:0

0

5:0

0

6:0

0

7:0

0

8:0

0

9:0

0

10

:00

11

:00

12

:00

13

:00

14

:00

15

:00

16

:00

17

:00

18

:00

19

:00

20

:00

21

:00

22

:00

23

:00

(HH:MM)

Po

wer

(M

W)

- 93.6MW in 32 min.- 2.9 MW/min

91 MW in 44 min.2.1 MW/min

rate of change is the slope of the straight line that fits the wind power data points in a 10-minute interval. Table 2.5-4 lists the resulting ramping statistics.

Table 2.5-4. Wind Power 10-Minute Ramping Characteristics

Average Standard Deviation

Max (+) Max(–)

(kW/min) (%/min) (kW/min) (%/min) (MW/min) (MW/min) 14-turbine

47 0.45% 78 0.74% 1.2 –1.4

138-turbine

269 0.26% 465 0.45% 7.0 –12.2

250+ turbine

438 0.18% 811 0.34% 11.3 –29.0

The average rates of wind power changes in 10-minute intervals are very small (less than 0.5% of the wind power plant capacity per minute). The maximum ramping rates can be large, especially for the negative ramping (rapid wind power decreases). Figure 2.5-6 is a daily profile of 1-minute average wind power from the wind power plant with 138 turbines. The graph shows that output from a large wind power plant can ramp up and down quickly.

However, such big changes occur infrequently. To gain a better picture of wind power extreme ramping behavior, we calculated the distribution of all ramping rates as shown in Table 2.5-5.

Figure 2.5-6. Example of high wind power ramping rates

As shown in the table, high ramping events (defined as ramping faster that 5% of plant capacity per minute) decrease as the number of turbines in the wind installations increase. When the numbers of wind turbines increase (larger wind power plants), the ramping-rate distribution becomes more tightly bundled (i.e., smaller ramping values). The last column lists the relative frequencies and the actual number of occurrences (over one year) of large ramping rates that are over 5% of capacity per minute. Further examination of the data indicates that not all of those very large ramping rates are the result of wind speed changes. For the 14-turbine data series, 51 out of the 64 large ramping rate occurrences are wind-speed related. For the 138-turbine data series, only 6 out of the 21 occurrences are wind-speed related; 3 of them are up ramping and 3 are down ramping. For the 250+turbine wind data series, only 1 such occurrence (down ramping) can be related to wind speed change.

Table 2.5-5. Distribution of Wind Power 10-Minute Ramping Rates

< ±1σ (78 kW/min)

< ±2σ (156 kW/min)

< ±3σ (234 kW/min)

< ±4σ (312 kW/min)

< ±5σ (390 kW/min)

> 5%/min (525 kW/min)

14-turbine

0.82476 0.95146 0.98305 0.99299 0.99673 0.00122 (64)

< ±1σ

(0.5 MW/min)

< ±2σ (1.0 MW/min)

< ±3σ (1.5 MW/min)

< ±4σ (2.0 MW/min)

< ±5σ (2.5 MW/min)

> 5%/min (5.2 MW/min)

138-turbine

0.84047 0.95742 0.98450 0.99304 0.99619 0.00040 (21)

< ±1σ

(0.8 MW/min)

< ±2σ (1.6 MW/min)

< ±3σ (2.4 MW/min)

< ±4σ (3.2 MW/min)

< ±5σ (4.0 MW/min)

> 5%/min (12.1 MW/min)

250+-turbine

0.86471 0.96779 0.98856 0.99438 0.99649 0.00044 (23)

2.5.4 Rare Situations

The analyses of 1-second, 1-minute, and hourly wind power series show that the majority of wind power fluctuations are limited in narrow ranges and their average values are relatively small. However, with strong winds sweeping across the plant, the output power can increase quickly as indicated by the extreme values of wind power step changes and ramping rates. During such an event, if the ramping up rate needs to be controlled, the plant operators can change it by temporarily stopping some of the turbines.

Operators are also concerned about sudden losses of power from large wind plants. During certain weather events the wind speed will keep increasing and reach the turbine cut-off speed eventually. It has been surmised that this condition will cause all the turbines to shut down after they have been operating at full capacity. Such an event may burden the power system considerably as it tries to bring up generation reserves to compensate for the sudden loss of a large amount of wind power. Although grid disturbances or equipment malfunctions may cause the entire wind power plant to trip off line in a very short time, the data collected to date show no evidence of an event where high wind caused all the turbines within a plant to reach cut-off state at the same time. High wind will cause individual turbines to shut down, but not the entire wind power plant.

58

59

Figure 2.5-7 is an example of strong winds that exceeded turbine cut-off speed and caused output to drop. In the figure, 1-minute average powers from four turbine strings24 and the total wind power plant are plotted along with the reference wind speed at the site25 for a 24-hour period. It shows that the plant had been generating at full capacity (about 100 MW at the time) since 3:00 a.m. with continuously increasing wind. Shortly after 12:00 p.m., the output power began to drop, and at about 15:00, when wind speed appeared to be at peak, the output power dropped to the lowest point for the day. After 15:00, the wind speed began to decrease, but the wind plant’s output began to increase (with some fluctuations closely related to the recorded wind speed). At about 17:00, the plant finally generated full power and remained at that level for the rest of the day.

30-turbine

24 These four turbine strings of 14, 30, 39 and 5 turbines make up the wind power plant. Output data from each string are recorded separately. 25 The wind speeds were recorded by an anemometer mounted on a pole about 4 m above ground inside a fenced area. Although not hub-height wind speeds, they nevertheless give good indications of the wind conditions at the site.

0

10

20

30

40

50

60

70

80

90

100

0:00

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

0

11:0

0

12:0

0

13:0

0

14:0

0

15:0

0

16:0

0

17:0

0

18:0

0

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

Time (hh:mm)

Po

wer

(M

W)

0

2

4

6

8

10

12

14

16

18

20

Win

d S

pee

d (

m/s

ec)

39-turbine 14-turbine

55-turbine Total Wind Speed

Figure 2.5-7. Example of wind power during high wind period

60

Figure 2.5-8 Detail of turbine cut-off with 1-second power data

series

0

5

10

15

20

25

30

35

40

4512

:00

12:1

0

12:2

0

12:3

0

12:4

0

12:5

0

13:0

0

13:1

0

13:2

0

13:3

0

13:4

0

13:5

0

14:0

0

14:1

0

14:2

0

14:3

0

14:4

0

14:5

0

Time (hh:mm)

Ou

tpu

t P

ow

er

(MW

)



The resolution in Figure 2.5-7 does not provide enough information about plant operations during a high wind period. Figure 2.5-8 plots 1-second output power from the four turbine groups during the 3-hour period from 12:00 p.m. to 15:00 p.m. The step changes in the groups’ output have an average magnitude of 750 kW and they occur within 2 seconds (2 data points). The fact that there are down-steps and up-steps during this period provides further evidence of wind diversities within the plant. These step changes show that individual turbines shut down during this high wind period. They also show that although some turbines shut down in rapid succession, not all turbines shut down at the same instant. The largest 1-second power drop during this 3-hour period was 1.4 MW (two turbines shut down at the same time), but the average ramping was only 293 kW/s (less than 0.3% of total capacity per second). Using 1-minute power data for this period, the average ramping rate was only 586 kW/min and the maximum ramping was 2.2 MW/min. This example offers clear evidence that output from a large plant does not drop from full power to zero rapidly because of a very strong wind that exceeds turbine cut-off wind speed. Physical separation and local terrain will cause variations in wind speed at individual wind turbines, and therefore, all turbines will not be at the same operating status.

2.5.5 Summary and Conclusions

The analysis of actual wind power data collected by NREL from operating plants has provided several important insights about wind power plant behaviors. The wind power is stochastic but not totally random. Because of wind persistency, wind power changes from one instant to the next are mostly confined in a narrow range, especially for 1 second and 1-minute periods. The data show that short-term fluctuations of a wind power plant are very small: for 1-second data series the average step changes are less than 0.5% of plant capacity, and about 1% for 1-minute data series. More than 99.8% of the 1-minute step changes are within ±10% (±5σ) of plant capacity for all wind power plants. None of the step changes with a magnitude greater than 10% of the plant capacity is related to wind speed changes.

Although wind power can experience large changes during longer time frames, large changes are infrequent. More than 98% of all 10-minute step changes are within the range of ±3σ (approximately ±8% of the plant capacity for larger plants and ±16% of plant capacity for smaller plants). Ninety-eight percent of the hourly step changes are also within ±3σ, which represents ±24% of plant capacity for large plants and ±31% of plant capacity for small plants.

Wind power ramping rates are also small. The average ramping rate for a wind power plant is less than 0.5% of the plant capacity per minute. More than 99.6% of wind plant ramping (in 10 minutes) is less than 4% of plant capacity per minute. It is extremely rare for the ramping rates of a large wind power plant to be more than 10% of plant capacity per minute.

The data also suggest that output fluctuations are influenced mainly by the size of the wind power plants. Differences in turbine types and plant locations have less influence in the step changes and ramping of wind power. Step change statistics for two wind power plants with the same installed capacity will be similar.

The actual wind power data have shown that simply scaling the output of a small wind plant to match the expected output (either power or total energy) of a large wind power plant will exacerbate the fluctuation characteristics of wind power. Scaling down the output of a large wind power plant to match the expected output of a smaller wind power plant will have the opposite effect. Simple extrapolation from a single anemometer reading to the output of a wind power plant will produce an even worse effect. The wind power profiles that result from either approach will not show the true behavior of the intended wind power plant.

2.5.6 Diversity between Wind Generation and Load

The principle of diversity applies separately to wind, load, and to wind and load together. Over different time scales the degree of correlation between wind and load will vary. In this section we examine the impact of this diversity on power system operational requirements, based on extensive studies that have been performed throughout the United States.26

Generally, power system operations tasks can be divided by the time scale that is involved. Figure 2.5-7 illustrates this graphically. The shortest operational time scale is often referred to as the regulation time scale, and consists of seconds to minutes (there is not universal agreement on

26 Smith, Milligan, DeMeo, Parsons, Utility Wind Integration and Operating Impact State of the Art, IEEE Transactions On Power Systems, Vol. 22, No. 3, August 2007. Available at http://www.nrel.gov/docs/fy07osti/41329.pdf (posted by permission).

61

http://www.nrel.gov/docs/fy07osti/41329.pdf

the precise boundaries of the time scales in this discussion, but the principles are broadly accepted). In the regulation time scale, generation is typically controlled by AGC to counter small increases or decreases in load. The resulting regulation service is a capacity service that has no energy component.

Table 2.5-6. Regulation and load following differ

REGULATION LOAD FOLLOWING

Patterns Random, uncorrelated Largely correlated Generator control AGC Manual Maximum swing

(MW) Small 10-20 times more

Ramp rate (MW/minute)

Many times more Slow

Sign changes 20-50 times more Few

The load following time frame is longer than the regulation time frame, and is generally from several minutes to one or more hours. During the load following time frame the generation must follow trends in the load, such as the morning load pickup. Generators are economically dispatched to follow these trends, while the regulating units pick up the fast variations in load and return to the midpoint of their operating range so that they are positioned to regulate load over the next several minutes. The different properties of regulation and load following are illustrated in Table 2.5-6.

The longer term scheduling, or unit commitment time frame encompasses several hours to days. This is the period during which slow-start thermal units must be brought up to operating temperature before generating power. Once these units have been started, they have minimum up-times, and when they are cycled off, they have minimum down times. Accurate wind forecasts can be very helpful for the unit commitment process, because un-necessary committed capacity is expensive, and insufficient committed capacity can result in higher costs or insufficient online capacity.

62

Figure 2.5-9. Time periods for system operation

The impact of wind on the regulation time scale is generally quite small. This is because wind and load are uncorrelated on this short time scale. The result is that the regulation needs add geometrically. As one example, the Minnesota 20% Wind Integration Study27 utilized to determine the impact of wind on the regulating requirement of the grid operator. In this case, k is a scaling factor that relates the standard deviation of the load to the regulating reserve that would be held for the system without wind. N is the rated capacity of the wind on the system divided by 100, and σW100 is the standard deviation of wind per 100 MW.

Table 2.5-7. Regulating requirement of wind increases less than linearly with wind capacity 27 Final Report - 2006 Minnesota Wind Integration Study, Volume I, available at http://www.uwig.org/windrpt_vol%201.pdf

63

Scenario

Regulating Requirement (MW)

No wind 137 15% wind energy 149 20% wind energy 153 25% wind energy 157

The impact that wind has on power system operations during the load following and scheduling time frame is more significant. Because wind has a near-zero marginal cost, it is typically the most economic generator and will therefore nearly always be used if it is available.28 It is common to examine the load after subtracting wind generation for each time period to determine the combined actions that are required from the balance of the generation fleet to balance loads and resources. In the discussion that follows, this “net load” represents the remaining load that must be served by non-wind resources after accounting for the load that can be served by wind. The examples discussed here are taken from the Minnesota 20% Wind Integration Study and the supporting data set. Although the quantitative impacts of wind would differ across the North American grid, the qualitative results are still valid. This can be confirmed by examining the many wind integration studies that have been done to date, many of which are summarized by Smith, et al (2007).

Using data from the 25% wind penetration case (based on wind energy relative to total energy demand) Figure 2.5-10 shows the hourly load and the hourly net load for a one-year period. This penetration rate of wind is larger than any that currently exists within NERC, but West Denmark has a significantly higher wind penetration rate.29 As indicated in the graph, the loads were scaled to represent demand in 2020. At this large wind penetration, it is clear from the graph that there is a need for increased turn-down capacity when the wind is high and the load is low. Because so much detail is lost by showing the full year, Figure 2.5-11 illustrates a typical week from the one-year data. The daily load cycles can be easily discerned. The net load appears in front. The wind generation during this week effectively changes the shape of the load that must be met by conventional generation, although the extent to which the net load shape changes is not the same from day to day.

It is useful to examine some of the impacts that are shown in the graph. The first observation is that the minimum net load at night is significantly lower than load alone. This implies that the generator fleet must achieve a lower turn-down level during these times. The required turn-down level is not constant from one night to the next.

28 If wind resources were to bid into the market LMP constraints could result in other generation running and wind could be curtailed to relieve the constraint. However, to focus on the potential implication that wind has on the balance of generation, we ignore this possibility for the present discussion. 29 Holttinen, et. al (2007), Design and operation of power systems with large amounts of wind power State-of-the-art report, International Energy Agency Task 25 Report. Available at http://www.vtt.fi/inf/pdf/workingpapers/2007/W82.pdf

64

65

16x103

14

12

10

8

6

4

2

0

MW

80006000400020000

Hour

Combined Load (2020) Combined Load and Wind

Figure 2.5-10. Hourly data for one year from the Minnesota integration study shows the impact of wind in the hourly time frame

During some of the morning load pickups, for example just after hour 50, there is a very large upward ramp requirement. This is caused by a reduction in wind generation at the same time that load is increasing. A similarly large ramp requirement occurs just past hour 140. During the other morning load rise periods, the ramp requirements are less severe, and even are similar to the no-wind case.

It is also clear that the amount of wind energy delivered during any given day differs from adjacent days. It seems clear that an accurate wind forecast would be valuable to the system operator to determine in advance how much generation should be committed to reliably supply the load.

66

16x103

14

12

10

8

6

4

2

0

MW

16014012010080604020


0

Hour

Figure 2.5-11. One week of selected hourly data from the Minnesota study shows an increased need for ramping capability and lower turn-down levels

The degree to which wind matches load differs widely in different geographic areas. The example above is based on data from Minnesota in the winter. During that time of the year, it is common to have large powerful weather fronts sweep thru the state, bringing high winds. During the summer these frontal passages are not as prevalent or as strong. This results in a different pattern of wind generation relative to load. Figure 2.5-12 shows a week when wind contributes to some of the daily peaks, but not all of them. It also shows days of relative calm, when there is little, if any, wind generation.

It is not possible to generalize the pattern of wind generation across the continent. Some wind regimes are driven by daily thermal cycles, whereas others are driven primarily by frontal passages. It is critical to ensure that wind data comes from the same time as load data whenever load and wind power are compared. Because weather is a common driver for load and wind, analysis should take into account the complex correlation between them. This issue is discussed more fully in Chapter 4 (Planning).

67

16x103

14

12

10

8

6

4

2

0

MW

478047604740472047004680466046404620Hour


Figure 2.5-12. Wind does not always contribute to the daily peak load

Another view of wind’s impact can be seen in the load duration curve in Figure 2.5-13. Wind causes a lower turn-down level during nearly 3,000 hours of the year, although in some cases the new level is not significantly different. However, it seems clear from the graph that there is a significant reduction in base load requirements throughout the year. The implication of these graphs is that a system with such a high wind penetration level will need additional flexibility, both in terms of ramping capability and low minimum run levels from the generation fleet.

16x103

14

12

10

8

6

4

2

0

MW

800 060004 00020 000Hours

202 0 Dura tion C urves L oad L oad and W in d

Figure 2.5-13. Load duration curve shows the impact of a large wind penetration on turn-down levels

We can also look at the impacts of wind on the 5-minute data. Figure 2.5-14, taken from a different week than Figure 2.5-11, still shows the increase in ramp requirements during several of the morning load pickups and evening load drop offs. One of the morning ramp requirements is more than 9 GW with wind, compared to a morning ramp is slightly more than 5 GW without wind

Although it is difficult to discern in the graph, the fast variations in the 5-minute time frame do not substantially change the shape of the non-wind generation requirements. Since the 5-minute data are on the boundary between load following and regulation, the correlation of the 5-minute wind power data with the 5-minute load data tends to be small (approximately -0.15 in this case). One significant implication of this lack of correlation is that it is not necessary to match each movement of wind power with corresponding movements of another generator. There are times that wind and load move in the same direction, reducing or eliminating the need for balancing action, and times that wind and load move in the opposite direction, which increases the need for control actions.

For system operations and planning it is important to have an idea of how often ramps occur and over what time scales. Figure 2.5-15 thru Figure 2.5-17 show the distribution of 5-minute ramps for three wind penetration levels, and are taken from the Minnesota study. The underlying data covers a three year period for each of the wind penetration scenarios.

16x103

14

12

10

8

6

4

2

MW

36.0x103

35.53 5.034.5

5-Minute Periods for 1 week

Load Combined wind and load

Figure 2.5-14. 5-minute wind and load data also show a need for increased ramping and lower turn-down levels over a selected week

68

69

Figure 2.5-15 shows the frequency distribution of 5-minute ramps for the 15% wind penetration case. The first distribution is for load alone, and there is a standard deviation of 50 MW. When wind is added to the mix, standard deviation of the net load increases from 50 MW to 55 MW. The graph shows a reduction in the mode, and an increase at each of the tails. This suggests that with wind there will be more, larger ramps, but these events are not frequent. The same general conclusion holds for the 20% and 25% penetration cases also, and is shown in Figure 2.5-16. It is also evident from the figures that the standard deviation of the net load increases with wind penetration, and this increase is not linear. The incremental increase in variability that arises as wind penetration goes from 15% to 20% is 2 MW, as measured by the standard deviation. However, when penetration increases from 20% to 25%, this incremental increase is 5 MW. This increase in variability is also captured by the graphs, which show an increase in the number of tail events at higher wind penetrations.

Figure 2.5-15. Distribution of 5-minute ramp requirements at a 15% wind energy penetration; Minnesota 20% Wind Integration Study

70



Although the quantitative impacts of wind on ramping cannot be generalized until more similar analysis is done, the qualitative impact can be generalized. As wind penetration increases on the power system, there is an increased need for ramping because the variability of the net load that must be served by the non-wind generation fleet is higher than it would be without wind. Many times the ramp requirements may not be significantly different than the no wind case. During other times, the wind may be moving in the opposite direction than load, which could have a significant impact on ramp requirements. This is true especially during the morning load pickup or evening load drop off. During those periods, it will be critical to ensure that sufficient ramping capability is available, which indicates the importance of wind forecasting. Although accurate forecasts are desirable, it will also be useful for the system operator to have some warning when there is the potential for a large wind event so that the system can be positioned appropriately.

It also seems clear that systems with significant wind penetrations will need lower turn-down points so that periods of high wind and low load do not cause reliability problems. One key question is how to ensure long-term development of a resource mix that can ensure reliable system operations at high penetration rates of variable resources. In regions that are part of an ISO or RTO, a system of transparent prices can encourage the development of the appropriate generation resources. For example, if market prices were consistently low or negative during the night (low load, high wind) then generation with high turn-down rates would be discouraged from building because they would lose revenue during low or negative price periods.

Additional information on wind diversity is presented in Chapter 5. In summary it should be noted that geographic smoothing can have a powerful impact on variability. This is true for load, wind energy, and load and wind together. Generally, geographically disperse wind generation of a given rated capacity will have less variability than a geographically concentrated wind plant of the same size. The same is true of load. When analyzing the impact of wind on power system operational needs, it is clear that larger balancing areas, larger markets, and more geographic dispersion of wind all play an important role in reducing variability on a per unit basis.

Although NERC is concerned about reliability and not markets, it is clear that markets can help foster behavior that contributes to system reliability. Robust energy markets that operate on short time frames can more easily access needed ramping capability that is required at moderate to high wind penetrations. Large balancing areas and markets help smooth loads and wind, and also provide a deep stack of flexible generation. The combination of this reduction in required ramping capability per unit, along with an increase in actual ramping capability, means that the balancing area operator is less likely to experience difficulty in maintaining system balance.

Diversity in wind forecast errors will also help. Over large areas, wind forecast errors tend to be uncorrelated, and the evidence suggests that the reduction in forecast error can be reduced up to 50% over large areas as compared to concentrated wind plants.

2.5.7 Diversity across Technology Families There are many different types of variable generation that depend on different types of fuel. These different fuel types have different characteristics based on when they are available. Generation that depends on wind as its fuel will usually have high availability of wind and therefore high generation output during colder months of the year as well as during the night and

71

72

in the early morning. Generation that depends on solar as its fuel on the other hand has higher outputs in the summer months and during the day. Therefore, in a system that has a similar penetration of wind power and solar power the total generation would look somewhat constant throughout the day and consistent throughout the year. Unfortunately, because of these differences in fuel availability it is often hard to find geographic areas that have good resources for both fuel types. However, systems like California can benefit from this diversity of technology based purely on the size of its system and its own geographic and climatic diversity. California has both good wind resources in places like the Tehachapi pass as well as good solar resources in the sunnier southern portion of the state. Figure 2.5-18, below, shows California averages of wind output, solar output, load, and net load (load – wind – solar) for July of 2003.30

Figure 2.5-18. California averages of wind output, solar output, load, and net load

Figure 2.5-18 shows a minimum of wind around the same time of day that there is a maximum of solar, and vice-versa. By comparing the total load and the net load it can be seen that the profile shapes are closely identical and that the net load is displaced by somewhat of a constant number throughout the entire day. This can benefit the system as a whole by having less variability from one hour to the next. The larger the range of the minimum and maximum net load is the more the need for flexible generation to be dispatched. With only the wind dipping down during the peak load time during the day the net load becomes more extreme. By having solar generation in the system portfolio it can shave off this peak to reduce the net load during the peak hours. Figure 2.5-19 shows this phenomenon in more detail.31

30 http://www.uwig.org/CEC-500-2007-081-APB.pdf (page 40) 31 http://www.uwig.org/CEC-500-2007-081-APB.pdf (page 24).

73

Figure 2.5-19. Net load mitigation due to impact of solar generation

The squares on this graph represent the delta from the hour prior to the current hour. At hour beginning 6 there is a maximum positive change in load. The load delta (represented by the blue square) shows about a 3,600 MW change (deltas are shown on the right side of the y-axis). Since the wind is decreasing this hour this enhances the delta of load minus wind (represented by the green square) to about 4,200 MW. However, since solar has actually increased this hour, the total net load (represented by the gold square) reduces to about 4,000 MW.

With the small levels of solar penetration that are on the power system today there is little experience on the true values of technological diversity. The effects of solar and its correlation with wind on net load is something that will continue to be researched. The Western Wind and Solar Integration Study is looking at levels of 10% wind/1% solar, 20% wind/3% solar, and 30%wind/ 5% solar. One of the results of this study and possible others should show more insight into the benefits of diversity between the two technologies.

2.6 Technical Power System Operation Challenges with Variable Generation and Current Operations Practices to Address Such Challenge

The operational issues created by variable generation result from the uncertainty created by the variable output and the characteristics of the generators themselves, such as the inertial response and dynamic response during fault conditions. The impacts are also affected by factors specific to the particular variable generation site, its interconnection to the power system, the

characteristics of the conventional generators within the system being operated, and the rules and tools used by the particular system operator.

The operational issues created by variable generation can be considered in terms of various time frames: seconds to minutes, minutes to hours, hours to day, day to week, and week to year and beyond.

The operational experience of most utilities in the United States and Canada with variable generation has been limited to a small portion of the total generation within a power system or control area. Historically, most generation on the system could be scheduled and the output controlled by the system operator.32 Historically, for most utilities the most uncertain element in the power balancing was the load forecast; and the larger imbalances were created by contingency events. However, this situation is changing rapidly and in the next few years, variable generation and wind plants in particular are or will be significant producers of the energy portfolio in many North American regions. The integration of large amounts of variable generation in control areas and power systems poses new challenges to the power system operation. The operational practices to address them are at this time a work in progress, being developed and adjusted as more experience is gained.

2.6.1 Seconds to Minute Time Frame

2.6.1.1 Automatic Generation Control (AGC), Area Control Error (ACE), and Frequency Regulation:

The changing output from variable generation creates an imbalance between production and demand. This imbalance will create area control error, which represents the error in the amount of actual interchange versus scheduled interchange (for an interconnected area) and error in the amount of frequency error, in terms of MW. The immediate response to the imbalance (in between AGC cycles) will be the inadvertent interchange. The secondary response is by the control actions initiated by AGC to correct ACE. The AGC response occurs on a processing cycle of a few to several seconds.

If variable generation is a small portion of the total amount of generation in the control area, the impact is not significant and can be managed by the same means in place for uncertainties in load forecast. When the amount of variable generation is significant, the variability can exceed the uncertainty and variability in the load and the variable generation can become the dominant factor in the area control error. The amount of variable generation required to impact frequency control and area control performance will be dependent upon several factors. For many regions and control areas, studies have been conducted to predict the impact significant amounts of variability would have for this time frame, but at present levels in most cases no change has yet been necessary to operational practices. It is anticipated that as the amount of variable generation increases, there will be a larger average ACE and a greater number of excursions into emergency control regions. It will become more challenging for the system operators to meet

32 Nuclear plants provide a significant exception though their output is not variable. Run-of-river hydro, and increasingly fish-constrained hydro are also exceptions.

74

NERC’s Control Performance Standards. Although at present levels of wind and solar production most systems in the North America have been able to meet the Control Performance Standards, it has been demonstrated for autonomous grid systems with relatively large wind plants that frequency error increases with the addition of wind generation (ERCOT, EIRgrid, Hawaii Electric Light Co). It has also been found that a high percentage of variable generation results in a larger number of control actions and greater range of control for regulating generation under AGC control, which may eventually lead to increased operational costs.

The degree of impact in this time frame is dependent upon several factors:

Degree to which the variable resources are correlated within the control area: Diversity will lessen the impact of the power fluctuations in the control area. Diversity does help smooth the variability in the seconds-to-minute time frame.

Flexibility of the regulating generation for the control area in responding to frequency error through droop and/or AGC control, and amount of available reserves: The impact will be reduced if the regulating/responsive generators have fast control capability with sufficient reserve (up and down) to manage the imbalances.

Frequency bias of the control area: A system with a smaller frequency bias will experience a greater amount of area control error (frequency error and/or interchange error) for a given power imbalance created by variable generation.

Constraints in the interconnection: A system with constraints in the transmission interconnection may not be able to tolerate a large amount of inadvertent interchange on the tie lines created by variable generation as a strongly interconnected system.

Lack of interconnection: A system operating as an electrical island will experience all power imbalance created by variable generation changes as frequency error.

The mechanisms available to the operator to procure and manage regulating reserves in the short term: The term used for this function or category of reserve will vary for the specific market or system, however the variable generation will require greater use of the reserve generation which is responsive in the short term and available for frequency control. A flexible mechanism which allows the system operator to add or reduce the regulating reserves can adjust the amount of reserve available depending upon the actual conditions in real-time.

It is necessary to optimize the response of regulating units to AGC when there is a large amount of variable generation. Regulating generators also have to respond to a wider range of operation, and it may be necessary to modify the area control parameters to address the latency inherent to the AGC control. AGC operates on a cycle of a few to several seconds, and thus does not reflect changes in the variable generation (and ACE) that occurred during the AGC processing and control response. NERC Control Performance Standards measure imbalances of one minute and longer. Faster control is required on the interconnected system only to the extent that it helps better meet the one minute balancing requirement.

A review of present control area practices finds that most system operators in the North American regions have not established a firm policy of increasing or modifying the regulating reserves by a set amount to accommodate the present levels of variable generation, although studies have reviewed the need for reserves to increase with added wind and there is work in progress for several systems to address this in the future. Most studies and system operators

75

agree that some additional reserves for frequency control will be necessary as the amounts of wind on the system are increased. There is less information available regarding the impact of other types of variable generation on frequency control.

The mechanisms to procure additional reserves for frequency control are specific to the particular system or market and may affect ancillary service markets, regulating reserve requirements, and so on. At this time several systems permit the operator discretion to increase the reserve levels in order to meet the control performance criteria (or manage frequency for autonomous grids) as necessary when variable energy is on the system.

Analysis of frequency data from systems with significant wind energy on autonomous systems have verified that the increased frequency error is linked to the magnitude of the energy changes on a second-to-second time frame, rather than the total amount of variable energy being produced. This variability is more pronounced when wind plants operate in the mid-range of the turbine power curve.

As a means of addressing the problem, some interconnection requirements or grid codes require the wind plants to mitigate sudden changes in output resulting from plant startup, shutdown, or wind changes. There is often an “out” clause for mitigation during wind reduction. System operators on autonomous grids have employed curtailment of the wind plant as a means of reducing the variability under cases of extreme volatility, in response to the difficulties it created in managing frequency and the stresses imposed on those units responding to correct frequency. There have been a few subsidized demonstration projects illustrating that storage devices can be used to mitigate the second to second power fluctuations, acting to smooth the power exported to the grid. It is best to use storage as a system balancing resource and to first aggregate wind variability with load variability. This reduces the total amount of storage required to reduce system variability. However at this time storage solutions are not in wide-spread use or consideration, due to cost.

If the amount of wind production at night, coupled with the minimum generation limits of conventional generators, exceeds the ability of the system to absorb the energy, an excess energy condition exists. This is discussed further in a later section titled “Excess Energy”. Under such conditions, the regulating generators often operate closer to their technical minimums than historically the case. If the regulating generators become constrained towards minimums, it can be difficult for the system to respond to over-frequency and the control area performance drifts up. It may be necessary for the system operator to re-examine the criteria for maintaining sufficient regulating reserve down to manage credible contingency events involving loss of load, to ensure that regulating units will not be driven below their technical minimums.

2.6.1.2 Voltage/Reactive Power Control

As wind plants become a larger amount of the total generation on a system, and conventional generation is displaced, it is now widely recognized that wind plants need to provide the voltage regulation and reactive power control capabilities presently met by the conventional generation. FERC and AWEA recognized this need and incorporated voltage-ride-through and reactive power support requirements for wind plants in the Large Generator Interconnection Agreement as part of FERC Order 661-A. Wind plants must provide dynamic and static reactive power

76

support and voltage control if system interconnection studies show that it is required for power system reliability. In some cases, voltage regulation is required even if the plant is not producing real power; this is in order to avoid having to operate conventional generation solely for voltage support. The requirement for active voltage control is viewed by many systems as necessary to maximize the percentage of wind production on a system. Distribution-connected wind plants typically do not provide voltage regulation. The reactive capabilities of wind turbines vary by manufacturer and model, and in some cases a supplemental reactive device is required to meet the voltage regulation needs.

Some systems have found voltage stability constraints. ERCOT performs voltage stability studies in the day-ahead time frame and issues operating limits for a subset of the wind farms based on those studies.

2.6.1.3 Contingency Reserves

Contingency reserves provide very short-term reserves to manage unplanned outages and equipment failures.

One consideration in evaluating contingency reserves is whether the contribution of variable generation is included in the available contingency reserves. There is no consistent policy.

Variable generation resources are unique in that a sudden loss of generation can occur in the absence of equipment malfunction, due to loss of the original energy source. If the potential loss of variable generation is comparable to the existing defining contingency, there is no change required for contingency reserve to manage loss of the variable generation resource. If there are events that could result in a coincident near-instantaneous reduction from multiple variable generation sources, the aggregate loss could become the new defining contingency. Solar has the potential to ramp very quickly and could theoretically have a similar impact on the system as a generator trip. For wind plants, usually a sustained loss or increase in power production results in a ramp event which occurs over a longer time period. However, high wind-speed cut-out can result in a near instantaneous loss of power production from a wind plant which could require contingency reserve to cover. Geographic aggregation mitigates the speed of the wind generation reduction for plants with tens or hundreds of MW of output. Large amounts of wind generation are unable to simultaneously experience a high speed cutout event and instead experience large hour or longer ramp events. The ability to fore-warn the system operator of impending high wind speed cut-out is under development but does not seem to be available to the operators in the control centers today. At this time system operators in the North American regions have not had to acquire additional contingency reserves to manage the variable generation.

The inability of wind plants to remain connected following a fault on the system could result in loss of multiple wind plants for a single incident, and thus be the cause of correlated outages. Rather than being addressed by contingency reserves, this is being addressed through low voltage ride through requirements, discussed in the System Security section.

2.6.1.4 System Stability and Dynamic Response Characteristics

Wind and solar generation have different inertial and dynamic response characteristics compared to conventional generators. The impact depends on the operating policy to some extent. The

77

impact increases when variable generation such as wind plants displace conventional generation completely, such that a certain amount of conventional generation that would have been connected is now left online. The system inertia is reduced, a change which results in a higher rate of change of frequency and smaller frequency bias. Short circuit levels can also be affected. These issues are studied in the planning process, and may require modification of protection schemes, load shed schemes, and operator action levels. Some European and island grid codes are requiring an inertial response and frequency (droop) response from wind plants in order to permit the largest possible aggregate amount of wind, measures which can mitigate the effect of loss of inertial response.

The impact of these potential system changes on system operations may include change in procedures and modification of frequency control parameters (through AGC) to reflect the change in system bias. In autonomous systems studies have defined the minimum amount of conventional plant that must be kept online to provide system stability.

Depending upon the particular characteristics of the wind or solar facility, reactive power control issues can also arise. As mentioned above, some wind plants are designed to provide reactive control. However there have been some operational constraints identified at ERCOT which require operating limits for a subset of wind farms based on voltage stability studies.

2.6.1.5 System Restoration

During System Operations procedures, the system is in an abnormal operating condition. If the system has been islanded, frequency regulation is difficult and at times is managed through a different mechanism than under normal operating conditions. Under these emergency conditions, the additional uncertainty and imbalance created by variable generation would add to the challenges. At this time, wind plants and other variable generation do not participate in system restoration. Some systems are studying how wind plants could potentially contribute to system restoration in the future. Planners will have to examine the potential changes to existing restoration plans to consider the impact of large amounts of wind plants especially if a region of concentrated wind energy has the potential to be islanded.

2.6.2 Minute to Hours Time Frame

For this time frame, the system operator is responsible for monitoring the system for system problems such as thermal overloads, and taking necessary control actions; and ensuring the balance of power production with load demand to maintain interchange and system frequency. The system operator also maintains generation reserves in various time frames of response, to manage uncertainties and contingencies.

Historically, with limited amounts of variable generation, the major uncertainty for system operators came from the load forecast. Over time tools have been developed for load forecasting and most system operators have a fairly accurate load forecast. Perhaps more importantly, system operators have a very good understanding of the uncertainty in this load forecast which is included in the planning for reserves. Operator corrective actions were typically required for contingency events, such as loss of transmission or generation. Tools are in place to manage such

78

events effectively and generally result in short-term excursions, with the system rapidly being returned to normal parameters.

The output of variable generation introduces a new uncertainty factor for the system operator. The variable generation output affects the generation/load balance and can also create transmission system issues such as voltage problems or thermal overloads. The key challenge at present is the uncertainty in wind plant forecasts. When variable generation comprises a significant amount of the system’s energy production, the impact on the system operator from the uncertainty and forecast errors in the power production from those sources exceeds the uncertainty of load forecast.

Perhaps the greatest challenge for the system operator in this time frame comes from sustained ramping up or down from wind plants, which can be avoided in with proper commercially available forecasting tools. Sustained down-ramps of large amounts of wind power have created large frequency errors on several systems to the point of requiring load-shed. The ramps can occur over a period of several minutes to hours.

Another issue facing system operators on several systems is managing transmission constraints or transfer limits. In several systems the amount of possible wind production can create overloads of tie corridors or transmission systems.

These issues could be mitigated by the system operator having accurate forecasts in the near term and hour ahead for wind, and incorporating these forecasts into dispatch decisions. Very accurate forecasts in the short term are not presently available, and there is some question as to whether intra-hour events can be captured. Accurate forecasting of wind energy for hour ahead and day ahead time frames is an area of active research and development. There is not a consistent handling of wind forecasts that are available in the either the planning or operations. System operators are developing the tools and working to better integrate the available information into dispatch decisions, reserve planning, identifying constraints, and operations plans.

2.6.2.1 Near-Term Reserves

The term used for the reserves supplying near-term load balancing and frequency regulation is dependent upon the particular operational and market of the system operator. In addition, the time period of response covered by near-term reserves is specific to the system, and must consider the time period to obtain supplemental reserves. For example, at CAISO, the “Regulating Reserve” used for frequency regulation and control area performance, must be adequate to meet deviations within a five-minute interval. ERCOT utilizes “Responsive Reserve” to restore frequency within the first few minutes of a deviation from the standard frequency. AESO manages near-term imbalance with the Ancillary Services market. Historically, this type of reserve responsive within a minute primarily managed error in load forecast and variation in load demand, or provided contingency reserve.

Variable generation creates additional uncertainty, which results in a greater use of reserves. Near-term reserves must manage the short-term variability as well as the maximum potential ramping events than occur within the near-term time frame. The factors influencing the impact on short-term reserves are the same as those described in the section titled “Automatic Generation Control (AGC), Area Control Error (ACE), and Frequency Regulation.” For most

79

systems, the amount of near-term reserves can be adjusted by the system operator in response to observed system conditions, including the impact from wind plants or other variable resources. There is not a general guideline in place to establish a set reserve increase in this time frame based on wind or other variable generation amounts.

2.6.2.2 Load Balancing / Energy Balancing, and Sustained Ramp Events

One of the most significant impacts of increasing amounts of wind plant has been the result of wind ramps. Wind ramps can occur in any hour of any day, and in any month. Wind ramps can occur due to loss or increase of wind, or as a result of the rapid shutdown of wind plants in response to very high speed wind conditions.

The impact of sustained ramps on a system depends on several factors, many of which are also factors for sub-minute impacts:

Degree to which the variable resources are correlated within the control area. It has been shown that ramp events can affect wind production over a relatively large geographic area as a result of weather fronts.

Availability of wind production forecasting to the operator. If a system operator can obtain advance awareness that a wind ramp event may occur, additional reserves can be procured ahead of time (up or down).

Flexibility of the regulating generation for the control area in responding to AGC control, and amount of available reserves. The impact will be reduced if the regulating/responsive generators have fast ramping capability with sufficient reserve (up and down) to manage the imbalances.

Frequency bias of the control area. A system with a smaller frequency bias will experience a greater amount of area control error (frequency error and/or interchange error) for a given power imbalance created by variable generation.



The mechanisms available to the operator to procure additional generation reserves or commit additional generation in real-time. A flexible mechanism which allows the system operator to add generation/energy resources to the system gives the operator more options to adjust the balance of supply and demand to compensate the for an unforeseen and large change in wind production.

The availability of generation in the sub-hourly energy market: Systems with five, ten, or fifteen minute energy markets are better equipped to accommodate variable renewable generation with energy market response. Policies which increase the depth of the available generation stack in the sub-hourly market further reduce the cost of integrating variable renewable generation.

80

Availability of dispatchable or curtailable loads can increase the options available to the system operator to manage a ramp event.

Systems with existing wind installations have incurred an increase in ramp rate requirements. For systems with moderate levels of wind energy the existing capabilities have been sufficient to manage the increase. However, as wind energy increases, the ramping capabilities have at times exceeded the ability of the system to respond. This issue has become apparent first on autonomous systems with high wind penetration, such as ERCOT, Maui Electric Co., and Hawaii Electric Light Co. It is widely recognized that wind energy can become a larger percentage of total energy on a system with flexible plants. Good characteristics for conventional plants from the perspective of helping mitigate wind energy impacts include a wide operating range, good ramping ability, cycling capability, fast start-up capability, and short minimum down time. Generators that have these capabilities, such as simple cycle gas turbines and reciprocating engine plants, are not necessarily those that have historically had the lowest cost in comparison to less flexible plants (such as coal plants or combined cycle turbines). There is question as to how to encourage the construction of flexible plants in the future for market systems.

The following mechanisms have been used by system operators today, in response to wind ramp events, to balance frequency:

1) Increase available regulating and near-term reserves to manage wind energy

2) Operate sub-hourly energy markets

3) Reallocate reserve to maximize available ramping capability

4) Provide visual indication, through a trending display, to the system operator in the control room of the aggregate wind power output

5) Provide ramp-rate alarm indication to the system operator via the EMS system

6) Call upon offline (fast starting) generating reserves when ACE approaches action levels

7) Impose ramp rate limitations upon wind plants as an interconnection requirement (For Maui Electric Co and Hawaii Electric Light Co, an out-clause for downward ramping has resulted in upsets for down-ramp events but helped mitigate problems for up-ramp events)

8) Arrange for transfer from neighboring utility to provide balancing (where available)

9) Remotely curtail wind production (for up-ramp events)

10) Utilize curtailable loads to reduce demand during large ramp-down events

11) Under-frequency load-shed

Accurate forecasting of wind production is a key to avoiding large imbalances caused by unforeseen large-scale changes in wind. At present, the forecasting tools available to the system operator are not adequate to avoid significant system upsets as a result of sustained ramps for large amounts of wind energy. For system operators with significant existing wind, most control rooms provide an indication to the system operator of the total real-time amount of wind production, at least at the transmission level. However, accurate wind production forecasts are not yet readily available to the system operator. There is not a consistent mechanism to obtain forecasting, nor is there consensus across all areas as to who should provide it.

81

At present, some system operators have no wind energy forecasting is available. Other systems require the wind plants to supply forecasts either directly or through the market schedule. As the result of issues encountered to date, and because it is easier to forecast aggregate wind performance than it is to forecast individual wind plant performance, many system operators have procured or are developing an independent forecast without relying upon the wind plants themselves. There are some system operators using forecasts from third-party forecasters and others developing the forecast in-house.

Practical issues have come up that present a challenge to developing accurate forecasts. These include difficulty in obtaining wind turbine availability, difficulty obtaining accurate wind data, concerns over data confidentiality, and difficulty in developing an actual forecast strategy. It is becoming apparent that the analytical approach used with success in one region may not apply well to another. Another problem identified is failure to capture distributed wind plant output, as a separate item from load. In some systems wind production for plants connected on the distribution appeared to the operator as a load reduction, which further complicated forecasting efforts.

Even for system which has wind energy forecasting available, the process are not yet in place to incorporate the dispatch into control room decisions. For example, at this time ERCOT does not include wind forecast into the reserves planning, but presently incorporates the market participant schedules. AESO includes wind in the short-term adequacy, up to 80 MW. For both of these systems, work is in progress to incorporate wind forecasts in the dispatch decisions and reserve calculations, as well as provide forecasting information to the system operators in the control room.

2.6.2.3 Transmission Constraints and Transfer Limits

Often, very good resources for wind plants are located far from load centers. In several systems the installed capacity of wind plants exceeds the transmission capacity available to transfer the power. When an overload occurs in real-time, it is standard practice for the system operators reduce the output of the variable generation through dispatch controls (also called curtailment) to eliminate the overload, which requires a commensurate increase in conventional generation to compensate. Operators have had varying experiences with dispatch control of wind plants. The ERCOT system operator has found that the response to the dispatch signals is at times delayed, or that the response is a step function rather than a ramp down to the dispatch level. Hawaii Electric Light Co has had issues with failure of the remotely controlled curtailment system at one wind pant, which has required the system operator to call the wind plant operator to effect the curtailment. It appears technically feasible to implement a smooth ramp to the dispatch level, based on experiences in some locations. This is apparently easier to implement for some turbines than for others and is dependent on the degree of power limiting that can be implemented before the particular turbine must be taken offline. If turbines can reduce power to only 50% before being taken offline, for example, a smooth curtailment is more difficult than if the turbines can be curtailed to very low percentages.

For operational planning, wind power forecasts are being used to varying degree in the calculation of transmission constraints. In general, the wind forecast is not used for day-ahead planning but actual wind production is considered in the real-time contingency analysis.

82

2.6.2.4 Excess Energy (Over-Generation Condition)

In some systems, the output of variable generation resources such as wind is not well correlated to system demand. At times there is high production of wind energy during low-load conditions when fewer generators are online. For systems with a large amount of wind energy relative to the system demand, the system may not be able to absorb the amount of wind energy being produced during such conditions. This excess energy condition exists on island systems for up to eight hours of a day, depending on the production from variable resources (run of river hydro and wind). With the increase in wind plants expected for many other systems, it is expected it will be necessary to curtail production during lower load conditions for larger systems.

An important factor relating to this issue is the flexibility of the existing conventional plant. In order to accommodate wind and avoid excess energy, it is desirable for plants to have low technical minimums and plants which can be taken offline. Some systems have found it desirable to ban the addition of new base load conventional power plants to avoid their aggravating the minimum load situation. However in some systems; existing plants must operate continuously and have relatively high technical minimums. It is also necessary for the system operator to keep a minimum number of conventional plants online to provide system stability, adequacy of supply, regulating reserves, and address system constraints.

Autonomous systems with significant wind have found that at times, additional reserves must be maintained to manage wind ramping. In order to maintain this level of reserve, some plants must be kept online that might otherwise be taken offline. This leads to an increased level of minimum load. From the perspective of the wind energy producer, this is a drawback to using increased reserves as a strategy to manage wind energy changes.

For systems experiencing excess wind production, the standard practice is for the system operator to dispatch the wind suppliers to curtail the output to the level the system supports. There is not a standardize rule used for the allocating or prioritizing the curtailment. It can be based on contractual agreements (as in the island systems of Maui Electric Co and Hawaii Electric Light Co) or by pricing in Market systems. Clear rules need to be established and it is important that the administration of the rules not become overly complex. It is important for the system operator to have guidelines for regulating and near-term reserves down as the system will be operating constrained towards minimums during excess energy periods.

Practical issues encountered for dispatch (curtailment) signals include:

1) Dispatch being implemented as a sudden change (step function) as opposed to ramping to the desired level

2) Changes in wind production occurring after the curtailment occurs

3) Administrative burden to system operator to manage energy curtailments

4) Remote curtailment interface becoming inoperable so that the wind plant does not respond to the dispatch signal

5) System operator uncertain when to invoke the curtailment. Delay in curtailment resulting in over-frequency.

83

6) System operator uncertain when to remove the curtailment, uncertain as to the level wind output will be at once curtailment is lifted.

2.6.3 Longer Time Frames

Most system operators do not incorporate wind and other variable generation forecasting in the longer-term planning activities such as Outage Scheduling, Security Analysis, and Available Transfer Capabilities. Several systems are studying what amount of wind capacity might be reasonable to include in longer time plans. Data is being collected to evaluate monthly trends which could be considered in the timing of maintenance outages for conventional generation resources. Most systems expecting a significant amount of wind energy within the next year to several years are looking to incorporate wind forecasts in near-term analysis for security constraints, transfer capabilities, and voltage constraints. System operators are in the processes of evaluating how large amounts of wind plants affect existing emergency procedures and restoration plans, and how they may participate in the future, but at this time systems with significant wind do not expect participation of wind in restoration plans. For island systems, the uncertainties in wind have made fuel delivery schedules more difficult to plan accurately.

2.7 Technical Power System Planning Challenges with Variable Generation

A Transmission Planner faces many challenges when dealing with variable generation, including:

Interconnection queue logjams,

Pressure to use the existing grid more efficiently,

Transmission cost recovery mechanisms,

Transmission siting,

Operational Issues,

Distributed generation,

Grid Codes and Reliability Rules.

In many regions of the country, there are significant numbers of interconnection requests for wind plants resulting in queue logjams and in some cases planning paralysis. Interconnection requests are processed in queue order and are sometimes combined into group studies. As requests fall out of the queue or go into suspension, the ones that remain need to be restudied. Many regions are investigating options or implementing changes to help speed up the process.

Given that wind generation has a low capacity factor and can usually receive environmental approvals and be constructed faster than transmission, there is pressure for the Transmission Planner to be creative and consider other non-traditional options. For example, can transmission lines be dynamically rated during periods when the wind is blowing? Can other generators be redispatched to avoid congestion? Should storage options be considered to improve the transmission usage?

Once a transmission plan is developed, then the difficult question is asked, “Who pays?”. In the past, it was relatively easy for conventional, high capacity factor generation (coal, nuclear, hydro), to pay for the needed infrastructure to have a firm connection. This is more difficult for

84

variable generation. Policies are in place or under development in many areas of North America to help enable the connection of renewable generation. In Texas, Competitive Renewable Energy Zones have been established allowing for transmission to be built serving these areas. These new transmission projects don’t have to meet the same standard “used and useful” tests for load serving or reliability projects in order to get rate recovery.

In many areas of North America, it is difficult to site transmission. Lengthy environmental reviews make future planning difficult because of the uncertainty in being able to build projects. The U.S. Department of Energy is proposing National Interest Electric Transmission Corridors to help identify in advance important right of ways but this is being challenged. National interest wind studies are being conducted to determine possible transmission plans to accommodate high levels of wind (e.g. 20% by 2020 or 30% by 2030 visions). These studies are useful to help educate the public.

Normally, a Transmission Planner doesn’t have to consider operational issues when studying variable generation. The high degree of variability of this new technology may require consideration of a number of new issues. For example, does the variability lead to voltage flicker? Are there any power quality impacts due to harmonics? The variability is corrected to an extent by other generators operating on automatic generator control (AGC). Transmission planning studies should at minimum demonstrate that there is no congestion between the wind plant and the AGC plant. If there is insufficient load following capability then the NERC CPS1 and CPS2 metrics will be violated. It is a challenge for a Transmission Planner to make this determination as Balancing Areas are proposed to become larger and as ancillary service markets evolve it may be difficult to predict where AGC plants are located in the future. In addition, improved wind power forecasting or wind power management (curtailment and ramp limiting) are also evolving and will help limit the variability.

At present, the majority of wind plants currently connected or proposed to be connected are on the bulk transmission system (100 kV or higher). In some regions of the country (e.g. Ontario), feed-in tariffs are promoting significant numbers of connections on the distribution system. Ontario expects more than 1000 MW of distributed generation to be connected in the next 5-10 years. All of these installations are less than 10 MW in size. There are many technical challenges associated with a high penetration of DG. What is the reactive power exchange with the transmission network? Normally, voltage control and remote communication are not important for a handful of DG. Grid codes must evolve to ensure proper voltage coordination and to ensure the operators know the amount of DG connected for real time operation. In Europe, there are proposals to group large number of DG together to behave like a virtual power plant and to participate in ancillary services. High penetration of DG also requires careful coordination between the transmission and distribution systems for islanding protection, underfrequency and undervoltage load shedding.

Finally, grid codes and reliability rules are continually evolving. It is a challenge to both develop the rules as experience is gained and to follow the new rules. The next few sections describe this issue in more detail.

85

2.7.1 Facility Connection Requirements “Grid Code”

Variable generation must be able to meet a minimum set of technical requirements before the Transmission Planner begins technical interconnection studies. These requirements or grid codes have been evolving around the world as more experience is gained with high penetration levels of variable generation - especially wind. Some of the main items that should be considered in a Transmission Owner’s grid code include:

Voltage tolerance (both low voltage and high voltage)

Frequency tolerance

Reactive power capability

Voltage control

Turbine power control (both curtailment and ramp rate limiting)

Frequency response (optional)

Dynamic modeling data (including validation)

Special testing requirements

At present, wind production has no obligation for primary or secondary frequency control in Canada or the U.S., therefore requiring wind generators to be responsive to frequency is optional dependent on the location. In Quebec, some provisions for primary frequency control are being discussed with the manufacturers for future wind installations. Future wind plants will have to be equipped with a frequency control system capable of making an inertial contribution comparable to that of conventional generating stations during substantial drops in frequency. Hydro-Québec’s power system is unique because it is not synchronized with neighboring systems. In addition, the main hydroelectric generating stations are located to the north, 1,000 km from the load centers. As a result, during disturbances when the wind penetration is high and the load is low, Hydro-Québec’s network may have to deal with transient and dynamic instability as well as voltage and frequency instability.

The current NERC standard FAC-001-0 does not clearly require the Transmission Owner to document all of the facility connection and performance requirements needed for high penetration levels of variable generation. It is a challenge for a Transmission Owner to stay up date with the latest suggested revisions to grid codes. It’s also a challenge for manufacturer’s to design equipment that meets a wide variety of different requirements.

2.7.2 Variable Generation Models

The availability of accurate models of variable generation has been a major challenge in the past. However, manufacturers are catching up and providing detailed models for planning interconnection studies and simple models for incorporating in the NERC regional models. Technology is still developing rapidly and there are times when substitutes must be made or turbine vendors are switched at the last minute. Tariffs provide provisions for restudy in order to ensure the reliability of the system is maintained. Even if detailed models are available, they need to be configured with the correct input parameters (e.g. wind speed input, fault ride through scheme, voltage control enabled or not, etc.).

86

Figure 2.7-1 below demonstrates the high voltage response when a 100 MW wind plant is recovering from a slow-clearing single line-to-ground fault (SLGF) into a bus with a SCR (i.e. MW of wind plant/3-phase short circuit level at interconnection) of 11%. The wind plant is producing rated output. The various curves represent wind speed of 15 m/s to 20 m/s. The default parameter provided with the model is 15 m/s. The damping of the oscillations decreases as wind speed increases. The mechanical oscillation behavior is the same between 18 m/s and cutout speed of 25 m/s.

Wind Speed Impact to Damping

Time (sec) 0.0 1.0 2.0 3.0 4.0 5.0

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1.10

Voltage (pu)

15 m/s 16 m/s 17 m/s 18 m/s 19 m/s 20 m/s

18-20 m/s wind speed

15 m/s wind speed

Figure 2.7-1: 230 kV voltage. 100 MW wind plant. SLGF with breaker failure scenario

2.7.3 Facility Connection Studies

In most jurisdictions in North America, generators must follow the procedures given in the FERC Tariff. Interconnection and delivery service are treated separately.

Currently, in FERC's Large Generator Interconnection Tariff Order 2003, generators have the choice to take either Energy Resource Interconnection Service (ERIS) or Network Resource Interconnection Service (NRIS). If there is significant grid congestion in the area, most wind generators are opting for ERIS, where only local fault, thermal and stability issues are addressed by transmission additions in order to minimize interconnection costs. The definition of local impacts varies considerably. In some cases, local impacts are determined by scheduling to the nearest generator and in other cases by determining if the impact on a facility is significant (e.g. greater than 20%). The Transmission Planner must consider the NERC TPL standards when conducting generator interconnection evaluation and facility studies. It’s up to the Transmission Planner to determine the critical system conditions needing to be studied, which typically means peak load maximum generation cases or some off-peak load cases. There’s no requirement for a minimum load, high wind case to be studied, for example. Coordination of studies is covered in NERC standard FAC-002-0 as well as in procedures given in the tariff.

Appendix G of Order 2003 includes technical standards (derived from FERC Order 661) for wind generating plants such as low voltage ride through, power factor design criteria, and SCADA capability. There is a conflict given that not all generators are under FERC jurisdiction

87

and NERC standards are mandatory in North America. Perhaps there should be a transition to move technical standards out of the generator interconnection tariff and into the NERC standard process in order to be more open and transparent.

The generators that are designated energy resources apply for transmission service under the Transmission Provider's open access transmission tariff, where they would typically be eligible for non-firm point-point service, secondary service or they could request firm service as well. FERC Order 890 requires transmission providers to also offer conditional firm service. There could be reliability concerns if there is a large penetration of energy resources taking non-firm transmission service. FERC has stated that they don't want to create extra layers of reliability rules for their proposed conditional firm service and generator redispatch in Order 890A. The onus is on the Transmission Provider to manage the risks by making reasonable assumptions in their models and system studies. Conditional firm is not offered in areas covered by energy markets.

If the energy resource is counted on as supplying some firm capacity should this portion be backed up by firm delivery service?

It is a challenge for a Transmission Planner to keep track of the number of energy resources and transmission service arrangements in a given study area or model. New generator connection studies should ensure the base case includes all firm Network Resources and firm transmission obligations. This is a difficult task, especially in an energy market where the dispatch changes every 5 minutes and typically all variable generation is allowed to self schedule.

2.8 Variable Generation Forecasting Methods

The fundamental questions associated with variable generation need to deal with where, when, and in what quantity these resources are available. For example, one of the biggest challenges to integrating large amounts of wind energy has to do with the inherent uncertainties associated with availability of wind. The available generating capacity of variable generation at time of peak may be significantly less than their name plate capacity off peak, varying with location unless coupled with storage technologies or demand-side management options. Those entities responsible for bulk power system reliability must take these unique characteristics and attributes into account to ensure variable resources are reliably integrated into the system. The uncertainty in availability i.e. variability of wind can be better captured by better forecasting approaches.

2.8.1 Wind Generation Forecasting

Electricity generated by wind turbine generators varies fundamentally from that generated by other ‘traditional’ means in so far as the amount of that electricity depends on the available wind resource. Since the wind can neither be scheduled nor dispatched in the same manner as is the case with traditional (i.e. thermal, hydro and nuclear) generators; understanding the nature of the wind resource is critically important. For countries with a substantial share of wind power in the

88

generation mix, such as Germany, Denmark and Spain, such models are already an essential part of grid and system control33.

Wind energy is a variable resource, varying with wind speed, rapid rate of changes and with production only occurring when the wind blows. The available generating capacity of wind resources is considerably lower than its name plate capacity at the time of peak. Other factors such as weather, seasons, locations and geographic diversity can further impact availability and variability of wind resources. This poses reliability related planning and operating challenges in forecasting and performing resource adequacy assessments and modeling. Better forecasting can play an important role in capturing the wind uncertainty and is expected to be an important tool for managing the variable nature of wind power.

Typically Capacity Factor (CF) and Capacity Value (CV) are the terminologies that are used to determining the capacity of wind. The classical definition of capacity factor is the average power output during all the hours over a defined period of time divided by the nameplate rating of the generation resource. Capacity value is a measure of the generation resource output during risk or critical (peak) periods throughout the year. These peak periods can also be sub categorized such as, when the load is within 10% of its peak or top 5 or 10 load hours (for which the system reliability risk is highest) etc. An intermittent generator such as wind that never delivers during high-risk (peak) periods should have a low or zero capacity value, whereas an intermittent generator that consistently delivers on peak should have a relatively high capacity value. A generator that sometimes delivers during peak periods should have a capacity value somewhere in between these extreme values. In this case, a “sometimes-available” generator cannot be counted on with a high degree of certainty, but it does reduce the risk of insufficient generation when it is available. It is therefore important to elaborate on capacity value analysis/approaches.

2.8.1.1 Capacity Value Analysis

The true capacity value of wind generators is often a source of great debate and concern among system operators. Wind generators are non-dispatchable resources because their power output varies according to the wind conditions at any given instant in time, and is difficult to predict or forecast with any degree of accuracy. The unpredictability of the wind is a reasonable concern that certainly has an impact on the overall value of the resource. Calculating the value of the wind during the highest risk periods throughout the year provides an assessment the value of the wind resource.

A number of ISOs define capacity value as the capacity factor during those hours of the day when the peak load is likely to occur in the peak months of June, July, and August. For example, the capacity value can be defined as the average hourly wind power output during the periods when load is within 10% of its peak or an hourly wind power output during the top 5 load hour periods (when system reliability risk is highest). This is per unit of the wind power output nameplate rating. Unlike firm, dispatchable generation resources, wind generator output varies

33 Bernhard Ernst, Brett Oakleaf, Mark L. Ahlstrom, Matthias Lange, Corinna Moehrlen, Bernhard Lange, Ulrich Focken and Kurt Rohrig. ‘Predicting the wind’. IEEE Power and Engineering Magazine. Wind Integration - Driving Technology, Policy and Economics. Nov./Dec. 2007

89

on a continuous basis. Due to this variation, the value of the wind generator is a smaller percentage of the installed nameplate value when compared to fully dispatchable generators.

Various capacity margin, forecasting and resource adequacy techniques are used among different regions/sub-regions as per their specific planning needs and reliability related requirements. These techniques may also differ for various planning time frames such as near term, mid and long term needs. Wind can be represented in reliability models using several approaches including probabilistic and deterministic approaches. Similarly, wind capacity contributions can also be determined via deterministic or probabilistic approaches. The capacity value analysis may highly rely on historical performance of a wind generator or generators to analyze the corresponding impact on forecasting and reliability.

The probabilistic approaches i.e. the loss of load expectation [LOLE] criteria is typically used for conventional resource adequacy models. The appropriate modeling of variable generation resources may also be suited to probabilistic methods, however approximations may be used by utilizing deterministic techniques based on specific planning needs and reliability related requirements within different regions/sub-regions.

2.8.1.2 Near Term and Real Time Wind Forecasting:

Reliable power system operation requires precise balancing of supply and demand in accordance with system operating criteria. Typically the system operators manage supply and demand balance on a minute to minute basis considering load and other forecasts, operating uncertainties and using available resources and the specific rules.

Wind forecasting for the operational near term, day ahead, hour ahead and in the real time poses further complexity and challenges to the system operators, wind farm and/or the wind schedulers. Wind forecasting can be performed during these time frames using several methods such as an aggregated weather-based forecasting approach (e.g. mesoscale Numerical Weather Prediction (NWP) model techniques), use of persistence forecasting, historical analysis etc. For example the day ahead prediction systems commonly depend upon and uses numerical weather prediction model (NWP).

It is envisioned that all wind power facilities will forecast their power output (or provide the data necessary for forecasting) for next day as well as two hours prior to the start of the delivery hour, and possibly other timeframes determined to be useful and reasonable.

Work is continuing on further understanding various achievable and reasonable forecasting techniques as wind evolves.

More specifics on wind forecasting are elaborated in Chapter -------- of this report.

2.8.2 Solar generation forecasting

As with all weather-driven renewable fuels, solar energy is variable, and the important question is raised as to where, when, and in what quantity of energy from sun is available. Forecasting of this technology will become more important when the amount of installed solar energy increases,

90

whether it be from large-scale solar thermal facilities, local scale PV systems, or mixed solar hybrid projects.

One beneficial aspect of solar generation is that the output of a solar plant has some degree of positive correlation with the system peak on a daily and seasonal basis. From a day-ahead perspective, all forms of solar are captive to meteorological forecasting of ground conditions. From the perspective of hour-ahead and minute-to-minute forecasts, solar technologies benefit from the general precision in identifying the size and location of cloud fronts, which are the primary driver of variations in solar system output through the day. However, the implications of intermittent cloud cover for causing variations in system output varies dramatically for different types of solar technology.

Techniques and procedures for solar plant output forecasts are currently under development by utilities and ISOs. It is likely that different forecasting techniques will need to be applied for individual technologies that are collectively classified as “solar”. A description of the various technologies and subsequent need for distinguished forecasting tools is explored below.

2.8.2.1 Forecasting by Technology

Solar plants that ultimately drive a conventional steam turbine (power tower, linear Fresnel, trough), have considerable thermal inertia in the form of a working fluid that flows through the solar field34. Plants of this type can draw on this inherent energy storage during periods of passing or intermittent clouds to maintain stable electrical output. As a baseline, a plant not having a dedicated energy storage element (such as molten salt tanks) will typically have 15 to 60 minutes of stored energy in the heat transfer fluid, associated piping and vessels. The result is that the a plant of this type can forecast electrical output 15 to 60 minutes in advance with the same degree of certainty as a fossil plant.

For a plant of this type, with thermal inertia in the working fluid (and no dedicated thermal storage, such as molten salt), the variability in output will be of limited magnitude and will be largely borne in terms of the ramp-down time and slope that occurs at sunset. Further, the end-of-day ramp will be a function of season as solar collection fields are generally sized to produce more than the turbine-generator output during summer months. In this way, during the summer months, the solar plant can maintain electrical output during partial cloud cover by drawing on the thermal inertia and make up the lost inertia by utilizing the entire capacity of the solar collectors that would otherwise be intentionally de-focused. A solar plant will have a higher degree of uncertainty in output during the winter months, when the solar resource is lower and thus the ratio of solar field rating to turbine-generator rating is closer to one. During the winter, the plant will have less of an opportunity to make up thermal inertia that is lost during partial cloud cover.

Solar plants driving steam turbines can also extend the time frame of forecasting certainty by incorporating large scale thermal energy storage (molten salt tanks) and/or natural gas burners. A 280 MW plant on the Arizona Public Service (APS) system has been announced with 6 hours of molten salt energy storage. APS has characterized this plant as both a capacity resource and 34 Power towers will have substantially less thermal inertia than comparable trough and Fresnel systems as the heat transfer fluid flow is limited to the tower and power block, it does not include the entire collector field.

91

an energy resource. APS will be ascribing very high capacity value to the plant in resource adequacy analyses.35 APS has offered the following details about the forecasting process for this plant:

There are several levels of production forecasting and dispatch planning which will occur at the APS plant. While forecasting is important, building a dispatch model which consumes the forecasts and turns them into a plant output is as important when considering the effective use of storage. This is compounded by efficiency losses when storage is used. In addition, the storage must be managed in a way which limits the risk of wasting energy when the storage is full and the sun is shining. In the summer, all of these constraints force operators and dispatchers to make sure the storage is used before the next sunny day.

The process that APS plans to follow is as follows:

1 Month Operating Forecasts. On a weekly basis, the plant operator will provide a thirty (30) day schedule indicating typical production profiles and production flexibility based on the typical seasonal solar conditions. This forecast will be used by APS and the operator to develop preferred dispatch of the storage.

Short-term Operational Forecasting - Three (3) day production forecast based on weather forecasts and seasonal solar conditions will be provided daily and include a default operation plan as well as two optional operating modes. As the days approach, the accuracy of these forecast become more and more accurate. These forecast and operational plans are used to develop the daily dispatch plans. On a daily basis APS will provide a Dispatch Plan based on the forecast. As time passes throughout the day (interval to be determined by operating committee), The operators will update APS on the operations for the remainder of the day and identify the amount of storage and the expected hours of operation.

The initial work on the forecasting tool has looked at a system based on NREL production models and commercially available weather forecasts. For forecasts beyond current day, the method is based on traditional meteorological approaches giving a % cloud cover. This is correlated to a solar intensity or flux. The forecast accuracy declines as the forecast timeframe increases. These forecasts are generally 3 hour blocks. There will not be hour to hour updates. For the same day forecast adjustments, the commercially available hourly forecasts will be correlated with weather satellite data to give shorter interval data with higher accuracy. These near-term forecasts will have a pretty high degree of accuracy.36

Other forms of solar that do not utilize steam turbines as the prime mover include the dish-Stirling and photovoltaics (PV). While these technologies are considerably different in the way they convert electro-magnetic radiation into electric power, they share the property of essentially having zero thermal inertia. As a result, forecasts of plant output on hour-ahead and minute-to-minute basis will be considerably less certain than solar plants utilizing steam turbines.

At this time, techniques are being developed for forecasting PV installations that are net metered. Initial reporting from utilities with large PV installations (10+ MW) is that the ramp-rate for PV

35 Source: Arizona Public Service, Resource Planning 36 Source: APS

92

93

installations is on the order of seconds following partial cloud cover.37 The implication is that the techniques for forecasting PV plant output will be substantially different than those for forecasting output from solar plants with thermal inertia and/or additional thermal energy storage.

[Awaiting input from Stirling Energy Systems and a couple of PV guys to obtain some additional narrative on this subject; Also may consider input from SDG&E, SCE Resource planners and CAISO].

2.8.3 Other technologies

Although hydro as a variable generation resource is dependent on the weather for their fuel source, somewhat similar to wind and solar, however, hydropower generation has the advantage that the fuel (water) can typically be stored in reservoirs. Hydro is also typically storable and is considered more reliable. The capacity factor for hydro is high, often more than 95 percent, far more than capacity factors that can be achieved for industrial wind energy.

Hydro can also be represented in reliability models using several approaches including probabilistic and deterministic approaches. Various capacity margin, forecasting and resource adequacy techniques are used among different regions/sub-regions as per their specific planning needs and reliability related requirements. These techniques may also differ for various planning time frames such as mid and long term needs.

2.8.4 Challenges Ahead

The conventional resource adequacy planning processes ensure that the generation contributes to system reliability and that the reliability of the bulk power system is maintained. Variable generation, including wind, hydro, solar can provide a significant share in the total generation mix as a capacity and energy resource. As variable generation continues to grow, more emphasis is needed to understand its unique characteristics and to and to integrate these resources into planning processes while maintaining reliability of the bulk power system.

Better forecasting of variable generation may require new tools, software, business processes and approaches to ensure their characteristics are considered while ensuring the continued reliability of the bulk power system. Study of forecasting and related requirements can help planners and operators to reduce uncertainty of availability of variable resources and to reliability integrate these into the system.

37 Source: Nevada Power

Planning and Technical Impacts

TTrraannssmmiissssiioonn aanndd GGeenneerraattiioonn PPllaannnniinngg IImmppaaccttss

PLEASE NOTE: This is a rough first draft that requires much editing and additional material to fill remaining gaps. This draft has not yet been reviewed by even the full Chapter 3 team and many of the points included have not yet been agreed upon within the sub-groups drafting individual sections of the chapter. PLEASE be aware that nothing presented here is final and all is subject to discussion and change as deemed appropriate by the group. Once agreement is reached as to the content, the lead editors of this section will work to shorten some of the material to a more concise format and move some of the material into appropriate appendices of the broader document, but in this first draft it was felt to keep everything to facilitate discussion.

3.1 Introduction (DB)

The goal of bulk power system planning is to ensure that sufficient energy resources and delivery capacity exists to meet demand requirements in a reliable and economic manner. System planners utilize forecasts of future demand and generating technology costs and availability to specify the resources (supply and demand) and delivery infrastructure required to meet stated reliability targets or to ensure economic availability of energy. In addition to ensuring sufficient resources and capacity to meet demand under normal operating conditions, planners must also ensure adequate reserves exist to reliably serve demand under credible contingencies such as the loss of a generating unit or transmission line.38 Traditionally, bulk system planning included generation planning and transmission planning. Generation planning is now more appropriately referred to as resource adequacy planning acknowledging the increased role of demand-side resources. Resource planning and transmission planning are inter-related as there must be adequate transmission to deliver supply to demand. As noted further in this chapter, planning a reliable bulk power system with high penetrations of variable generation may require a more iterative approach between generating resource and transmission planning. Generation adequacy would no longer be a function of resource mix alone. The transmission system may change the capacity credit on wind generation. The power transfer capacity of transmission associated with the Energy Markets integrated with wind generation may change the planning reserve levels at peak conditions. Bulk system planning is conducted for both long-term and near-term time frames. Long-term resource adequacy planning ensures that sufficient supply with the required capabilities exists to meet demand, planning reserves, and operating reserves many years in the future. Transmission planning is conducted for long and intermediate time frames – planning must be performed to ensure that the long-term resource adequacy plan is achievable, but in the intermediate term, planning is performed to ensure that system reliability is maintained as new generating or transmission facilities are sequentially added to the system. The long-term and intermediate-term resource and transmission planning impacts of high variable generation are addressed in this chapter. The impacts of VG on near-term operational planning -- ensuring sufficient

38 NERC, “Reliability Concepts, Version 1.0.2” Dec 2007.

94


resources are available for supplying energy and operation reserve requirements hours to weeks ahead -- are covered in Chapter 4.

3.2 Representation of Wind Generator Output for Planning Studies (MM, DO)

Planning studies that evaluate the impact of future potential wind plants on the power system need extensive data sets that provide a realistic estimate of wind power output over several time periods. Large wind power plants consist of many individual wind turbines, which do not necessarily experience the same wind speed at the same time. This means that the output profile of a wind plant will differ substantially than the output profile of an individual wind turbine. For this reason, it is critical that the wind power is accurately represented so that the results of the study are credible. This issue is discussed in more detail in Chapter 2 -- Characteristics of Variable Generation. 3.2.1 Some Data Requirements are Common for Transmission Planning and Wind

Integration Studies

Hourly wind generation MW data for wind generation that is time synchronized with load across the interconnected area with wind generation is necessary to ensure a reasonable characterization of both wind and load. Because wind and load both respond to the weather, time-synchronized data is critical. During hot spells, electric loads typically are high, resulting from air conditioning loads and other weather-sensitive electricity demand. During such times, it would be important to capture the wind generation’s behavior. Using load and wind data from different time periods will not capture the underlying correlation (which is likely complex and nonlinear) between wind and load. The wind data should be specific to the geographic area that the generator is located in models. Presently, this data does not exist in the U. S. except for a few states. State to state data is not time synchronized. If available, it would be desirable to use actual wind plant production data for planning studies. However, this data may not be available, or if it is, may not cover the same time period as the loads. To overcome this problem, large-scale numerical weather prediction (NWP) models can be used to numerically “re-create” the weather at any time and space. These models are based on the same models that are used for weather prediction, and typically cover large parts of the country. To create a data set for system planning studies, NWP models are initialized with large-scale weather data, and run for the year(s) that correspond to the load-year(s) for the study. The models can typically be run at time steps of 10 minutes, and are typically not very accurate at smaller time steps. Wind speed is extracted from the model at appropriate geographical locations and at the hub-height of the prospective wind turbine. To focus on a particular geographic footprint for regional or subregional studies, the NWP models are often run at meso scales, which represent the region. At present, each geographical data point is used to represent no more than about 30 MW of wind generating capacity (nameplate). To simulate larger wind plants, multiple locations must be extracted from the NWP. Once this process has been complete, power output is calculated for each location that was extracted from the NWP, and statistical adjustments are made at several steps of the calculation. The result is one or more years of chronological 10-minute wind power production data, simulated to represent alternative scenarios and locations of potential wind plants. This data set informs the

95


96

power system models that are used in various planning studies. Figure 1 illustrates the geographic footprint that was modeled for the Minnesota 20% Wind Integration Study. The upper panel shows that a large portion of the U.S. and Canada must be modeled so that the weather patterns and flows can be accurate represented as they move into Minnesota. The lower panel shows the more detailed modeling that is necessary to represent the wind speed at plausible geographic locations and at hub heights for wind turbines.

Figure 1. Meso-scale modeling footprint (upper panel) to capture the weather and wind in Minnesota. The lower panel zooms in to show the individual extraction points that represent clusters of 30MW of wind capacity (rated)

By the end of June, 2008, the National Renewable Energy Laboratory is expected to release data for the year 2004 for 80 meter and 100 meter modeled tower heights for 2 km square resolution across the geographic area for the U.S. portion of the Eastern Interconnection of the Joint Coordinated System Plan area. This represents about 90% of the land based wind potential in the U. S. portion of the Eastern Interconnection. The data based on the historical years 2005 and 2006 will be produced later in the year. Hourly and 10 minute data will be produced for about 500,000 MW of potential wind generation. Meso-scale weather modeling will be used as the technique to produce the models. Additional wind data simulation for 2004-2006 is currently underway (as of April, 2008) for the U.S. portion of the WECC. Wind data simulations based on NWP modeling is not perfect. However, it is at present the state-of-the-art for creating a wind power data set that can stand up to scrutiny and provide a reasonable estimate of wind power output over time scales of minutes to years. The use of a single anemometer or wind turbine to represent the power output profile of large wind plants is simply not credible, and will seriously compromise the results of any study. As wind turbine populations grow, the meso-scale models may be replaced with actual wind generator output of the same tower height and turbine design. Presently, there are too few turbine sites of similar type and tower height to provide an actual output data base. Any meso-scale model needs to be verified by using wind tower data. NREL is working on approaches to validate meso-scale model output. To expand on these activities will require additional meteorological towers with anemometers to collect data that can be compared to the NWP-simulated wind power data. Additionally, NREL currently collects actual wind plant data from more than 1,000 MW of operating wind plants in the U.S. Efforts to validate the NWP modeling for wind is expected to continue and evolve over the next several years.


Several years of hourly wind data is required as there is a considerable variation in the total wind energy produced by year as well as events in wind and load output that would produce examples to test the power system. Running multiple years of data through a model provides a reasonable evaluation of the variation that may occur from weather. Having NREL manage the preparation of the missing historical years of 2006 and 2007 and adding the current year with the same quality of the NREL-JCSP study would be a great service to the utility and wind industries. Adding promising off-shore wind sites would add time diversity as the water based turbines have a generation pattern that more closely matches load. Even though the cost of the sea based turbines appear to be expensive, without the data the value of sea based turbines cannot be used with the data bases that are currently under development. While more wind data makes a better wind data base, load data over a long period does not scale well to future years due to the change in the load factors associated with changing usage factors over time. Customer’s use patterns change over time with the advent of air conditioning, larger houses, larger electronics, etc. Based on experience, an estimated load-wind generation output data set of about the last seven years would probably be about all that could be used. Statistical analysis may be used to verify the seven year number. Adding data wind data from Canada that is produced to coordinate with the U.S. data would be helpful for complete system models that NERC uses. Discussions between NREL and the Wind Energy Institute of Canada (WEICan) have been initiated to work on the situation. Funding issues between the U.S. and Canada are the main issue. The significance of including the Canadian model is one of geographic diversity. The weather professionals indicate that the jet stream moves from south to north in the summer and those parts of Canada have a better summer wind pattern than the U.S. Canada has some promising wind sites that are being developed. This approach utilizes wind data from the same year as the load data so that any underlying systematic relationship between wind and load is captured implicitly in the data. More data is better than less data; we recommend a minimum of a 3-year data set if possible. In Milligan and Porter (2005) we discussed probabilistic methods of representing wind in reliability models. Because of the evolution of NWP modeling techniques over the past few years, we do not recommend probabilistic approaches unless they have been successfully benchmarked against several years of actual wind data. We also caution that the built-in Monte Carlo process that is built in to many reliability and production simulation models is likely inadequate for representing wind. This is because the Monte Carlo algorithms are typically based on probability distributions that do not adequately represent wind power and its changing distributions through the year.

97


3.2.2 Hourly Wind Power Simulated or Actual Data Can be Used for Wind Integration Studies and Long-term Transmission Planning Studies.

Transmission studies such as those conducted by MISO beginning with MTEP-03 need hourly wind production data, either simulated or actual, to assess line loadings and congestion and to evaluate alternative transmission expansion scenarios. In addition, wind integration studies that focus on the impact wind has (or will have) on power systems operations or planning also need hourly data. The hourly data makes it possible to run market simulations or economic dispatch studies for the time period of the study. Multiple years of wind and load data contribute to more robust results, subject to the limitations discussed above. 3.2.3 Wind Integration Studies Also Need Sub-Hourly Data

Because there can be substantial variability in wind production within the hour, wind integration studies must use sub-hourly wind production data. NWP models generally don’t do a good job of accurately simulating the weather on time scales shorter than about 10 minutes, so alternative data sets must be used on these faster time scales. Several of the integration studies done to date have used either actual wind production data or utilized NREL’s wind production data base on these fast time scales. Although this is imperfect, the lack of correlation between individual wind turbines and between wind and load in the minute-to-minute time scale allows a reasonable statistical comparison to be carried out, in spite of the lack of time- and geographic synchronizations. As more wind plants go into service, however, we expect that more actual wind production data from within the study footprint and time period can be used. 3.2.4 Caution Should Be Used With Existing Wind Plant Data

Wind technology has changed significantly over the past decade. Some older operating wind plants use older wind turbine technology, and often have hub-heights lower than 80 meters, a common hub height today in the U.S. These older wind plants may have data that is not representative of future wind plants that will use modern (or yet-to-be-developed) wind turbine technology. In addition, the wind is typically stronger at increasing heights from the earth’s surface. Therefore, turbines on shorter towers will likely under-represent energy capture, and can potentially have different power profiles. Existing wind plant data is often captured by PI systems that are tuned to optimize the large quantity of data that must be stored. Compression algorithms that use banding approaches for data storage may not capture the fast variations in wind plant output; those occurring over a few to several seconds. Care should be used when using this fast data for integration studies or other analyses of actual wind production data over these time scales.

98


3.3 Long-Term Resource Adequacy Planning (MM, DO, KP, DB)

NOTE: Much of the content in this section is copied verbatim from Michael’s and Kevin’s preliminary Draft Paper that is intended for presentation at the 2008 WindPower Conference, June 2008. It can be used by the NERC IVGTF internally, but is not for Citation or Distribution until after the conference, at which point it can be used freely by the IVGTF 3.3.1 Current Supply Adequacy Planning Approaches

Short intro needed here to briefly cover items that set up how VG impacts process such as: Definition of resource adequacy and purpose/scope relative to this document – reliability

vs. economic energy o More clarification as to inter-relation w/transmission planning o Scope of planning/ regional interaction with other organizations

Brief summary of current LT adequacy planning approaches that are not necessarily specifically designed for high penetrations of VG

o Metrics (capacity, energy, other) o Targets (1 day in 10 yrs, etc.) o Tools/Analytic Framework (probabilistic, deterministic, etc.)

Relationship of capacity credit of a particular plant and the contribution to supply adequacy evaluation

Load-serving entities (LSEs) such as electric utilities generally maintain some percentage reserve margin of capacity over and above their load requirements to maintain reliable electric service. State regulators, and even state statutes, may also require LSEs to maintain a certain reserve margin. NERC also requires regional reliability councils to meet certain reliability standards, and as part of that, regional reliability councils will require LSEs to have reserve capacity (generators that can respond quickly) and planning reserve capacity (generators that do not have to respond quickly). The regional capacity standards are voluntary, differ by region, and depend in part on how each region determines the capacity value of a generator. Regional transmission organizations also may require LSEs to have a capacity reserve margin. Although the source of the capacity reserve requirements may differ, common elements are present in all of them. For instance, the nameplate capacity of a generating plant is discounted to reflect the probability of the plant going off-line for scheduled or unscheduled maintenance. A time differentiation for capacity may also be applied, as it is generally (but not always) recognized that available capacity is more valuable at times of peak electric demand than at other times. An important trade-off is involved with capacity reserve requirements. Almost by definition, capacity reserve requirements involve making financial investments in generating capacity that will not be used or not used often. One could “overpay” for reliability by having too much reserve capacity. Therefore, the trade-off is enough reserve capacity to ensure reliable electric service while minimizing the costs of having reserve capacity.

99


3.3.2 Wind Plants and Supply Adequacy Evaluations

With nearly 17 GW of installed wind capacity in the United States at the end of 2007 and another 3.6 GW under construction (April, 2008; www.awea.org), the question of wind’s capacity value (sometimes called capacity credit) is gaining more attention. Wind’s low cost and environmental benefits, and the higher cost of competing fuels such as natural gas, mean that system planners will need to grapple with how to determine the capacity value of wind energy. It does seem clear that wind’s primary value is as an energy resource, but to the extent to which it contributes towards system adequacy is an important question. Wind generators occupy a unique place in the determination of capacity and effective load carrying capability (ELCC). Wind generators have typically very high mechanical availability, exceeding 95% in many instances i.e., the forced outage rate is often below 5%. However, because wind generators only generate electricity when the wind is blowing, a wind generator arguably has a forced outage when the wind does not blow. Therefore, the effective forced outage rate for wind generators may be much higher, from 50% to 80%, when recognizing the variable availability of wind. In addition, wind’s value to the electric system may also vary. The output from some wind generators may have a high correlation with load and thereby can be seen as supplying capacity when it is most needed. In this situation, a wind generating plant should have a relatively high capacity credit. The output from other wind generating plants may not be as highly correlated with system load, and therefore would have a lower value to the electric system and should receive a lower capacity credit (Milligan and Parsons 1999). The correlation of wind generation with system load, along with the wind generator’s outage rate, will determine how much capacity credit a wind generator will receive. Ensure that relationship between capacity credit of a particular plant and the contribution to supply adequacy evaluation is clear 3.3.3 Methods for Calculating the Capacity Credit Assigned to Wind Plants

To begin, it is useful to discuss the properties that a desirable capacity credit metric should possess. A capacity metric should be capable of assessing all types of generators, whether baseload, conventional, or variable. Because all generators have some probability of failure during critical times, the metric should recognize this potential, and plants that experience more outages should have a capacity value less than a plant that experiences fewer outages. A variable (or conventional) generator that never delivers during high-risk (peak) periods should have a low or zero capacity value, whereas a variable (or conventional) generator that consistently delivers on peak should have a relatively high capacity value. A generator that sometimes delivers during peak periods should have a capacity value somewhere in between these extreme values. In this case, a “sometimes-available” generator cannot be counted on with a high degree of certainty, but it does reduce the risk of insufficient generation when it is available. The capacity value must therefore be a probabilistic-based metric that can take these generator characteristics into account and rank the relative contribution of the generator fleet to the reduction of system risk. To properly value generators with different capacity contributions, it is useful to adopt the principle of vertical consistency. This concept, borrowed and adapted from public economics, says that power plants with a high capacity contribution should be ranked above power plants that have a lower capacity contribution. Plants that are equivalent in their reliability (and

100


delivery) should receive the same ranking. This latter approach is horizontal consistency. These fundamental principles, while very simple, can help guide us toward a suitable metric. Appendix XX provides a summary of the methods currently utilized by numerous vertically integrated utilities, ISOs/RTOs, and state PUCs to assign capacity credit to wind plants. 3.3.3.1 Effective Load Carrying Capability (ELCC)

Section needs to be consolidated Fortunately, such a metric does exist that possesses these properties, and has been used for several decades. It is based on well-established reliability theory and practice, and can be applied to all generators. This metric is based on one of several reliability metrics, such as loss of load probability (LOLP), loss of load expectation (LOLE), or expected unserved energy (EUE). The metric itself is ELCC, and can be carried out with a power system reliability model, with appropriate tweaking to properly account for the stochastic and variable nature of wind generation. ELCC can discriminate among generators with differing levels of reliability, size, and on-peak vs. off-peak delivery. It effectively rewards plants that are consistently able to deliver during periods of high demand, and ranks less reliable plants by calculating a lower capacity credit. For variable generators such as wind, the method can discriminate between wind regimes that consistently deliver during high-risk periods, sometimes deliver during high-risk periods, or never deliver during high-risk periods. In fact, ELCC can provide for a continuum of capacity values over these potential outcomes. With modern interconnected power systems, the LOLP does not necessarily measure the probability that load will be shed. The LOLP metric measures the risk that generation cannot meet the peak demand unless capacity is exported. When performing a reliability analysis it is necessary to select a risk-target. This is often chosen to be a loss of load expectation (LOLE) of 1 day per 10 years. This roughly corresponds to a 0.9997 probability that generation will be sufficient to cover load without unexpected imports. Because there is always a non-zero probability that any generator can fail at any time, methods for determining system adequacy must take these probabilities into account. One often-used approach is based on a fixed planning reserve margin, usually 15%. However, unless the reserve margin percentage has been determined by a reliability assessment, fundamental question, how adequate is the generation supply, is not addressed. When a reliability-based approach is used to determine system adequacy, an often used target is a loss of load expectation (LOLE) of one day per 10 years. This roughly translates into a probability of 0.9997 that there is sufficient generation to serve load. Other reliability targets or metrics, such as loss of load probability (LOLP) can be chosen if desired by the reliability organization. Figure 2 shows results from an example system that has a fleet of 100 MW generators that supplement large base load units. As the forced outage rate of the 100 MW units increases, the LOLE of the system increases. To bring the LOLE back to the 1d/10y target, additional generation must be added. The graph shows configuration A the 100 MW units have a 0.10 forced outage rate. As that rate increases up to 0.60 (this generation comprises about 10% of the system capacity), the required planning reserve percentage increases, as shown in the figure. If

101


102

System D had a 14.6% planning reserve margin, matching System A, in the absence of an LOLE calculation it would appear that they have the same level of system adequacy. However, because some of the generation in System D have higher forced outage rates, the level of system adequacy would be significantly less than System A. For system D to match System A, it would need an additional 6% planning reserve margin. From a system adequacy perspective, a probabilistic approach appears to be a more robust approach. In this context, the contribution that is made towards system adequacy, as measured by LOLE or a similar reliability metric, can be calculated with a reliability model, and is the effective load carrying capability (ELCC) of the generator. This approach has been used for decades, and an approximation method was first introduced by Garver (1966).

14.616.1

20.6

24.3

29.7

18.0

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

A B C D E F

Reserv

e M

arg

in (

% P

eak)

Figure 2. Otherwise identical systems that have different forced outage rates need differing quantities of planning reserves to achieve a 1d/10y level of system adequacy.

ELCC decomposes the individual generator’s contribution to system reliability. It can discriminate among generators with differing levels of reliability, size, and on-peak vs. off-peak delivery. Plants that are consistently able to deliver during periods of high demand have a high ELCC, and less reliable plants have a lower ELCC. For variable generators such as wind, the method can discriminate between wind regimes that consistently deliver during high-risk periods, sometimes deliver during high-risk periods, or never deliver during high-risk periods. In fact, ELCC can provide for a continuum of capacity values over these potential outcomes. With modern interconnected power systems, the LOLE does not necessarily measure the likelihood that load will be shed. The LOLE metric measures the risk that generation cannot meet the peak demand unless capacity is imported.


103

Reserve Margins as a Percentage of Peak Don’t Necessarily Address Adequacy As shown in Figure 2, a fixed reserve margin does not directly address the level of system adequacy that will be delivered by a given level of installed capacity. For a system with approximately 10% of its capacity in 100 MW units with varying forced outage rates, the relationship between these forced outage rates and the required installed capacity that would deliver a 1d/10y level of system adequacy appears in Error! Reference source not found.. As the forced outage rate increases on the 100 MW units, the required planning reserve margin to maintain the 1d/10y system adequacy level increases. This clearly suggests that two systems with the same reserve margin percentage do not necessarily achieve the same level of system adequacy.

26

24

22

20

18

16

14

Based on adding 54x100MW units @ 10% FOR to meet 1d/10yIncreasing the FOR means the reserve margin must increase tomaintain reliability

12

Pla

nnin

g R

eser

ve M

argi

n (%

)

0.2 0.3 0.40.1Forced Outage Rate on 100 MW Units

0.5

Figure 3. As forced outage rates increase, larger planning reserve margins as percent of peak are required to maintain system adequacy at a 1d/10y level.

Because the LOLP in any given hour is a function of the load and available generation, it is subject to many influences that include the available online generation and their outage characteristics. When calculating the ELCC of a variable power plant such as wind, there are many hours with significant LOLP that drive the ELCC metric. The wind ELCC depends primarily on the relationship (correlation) between wind power and load, but other factors will


also influence ELCC. Generation that is on maintenance is not available so will increase LOLP during those times. Controllable hydro generation is typically used to shave peak and/or scheduled during periods of high prices. Peak periods are generally those periods with highest LOLP, but that is not always the case when hydro and maintenance schedules increase hourly LOLP during lower-load periods. Analysis that was undertaken for the California Energy Commission39 found that during an unusually late, hot summer period when many units were taken out of service for scheduled maintenance, that the hourly LOLP in late September was nearly as high as during the peak summer period. Situations like this can result in a lack of recognition of the exposure of the power supply to potentially high levels of risk that can be overlooked. Any wind generation that would occur during these times would contribute to lowering LOLP, perhaps significantly. Because LOLP and other reliability metrics are heavily influenced by the available generation, the transmission system plays a key role. In larger Balancing Areas the grid allows the pooling of generation that can lower overall risk as measured by LOLP. Therefore, when investigating alternative transmission build-out scenarios or configurations, it is important to perform ELCC or LOLP evaluations holding the transmission system configuration constant between the wind and no-wind cases. To calculate ELCC, a database is required that contains hourly load requirements and generator characteristics. For conventional generators, rated capacity, forced outage rates, and specific maintenance schedules are the primary requirements. For a variable resource such as wind, at least one year of hourly power output is required, but more data is always better. Over the decades that ELCC has been widely applied, it has been used with a number of different reference units. Some early work (for example Garver, 1966) measured the capacity value of a generator against a perfectly reliable unit. Because such a unit does not exist, we prefer the alternative of measuring capacity value relative to a benchmark unit. Although we would prefer a widely adopted benchmark value (for example a gas unit with a forced outage rate of 5%) to allow for easier comparison among studies, it is important that the benchmark unit is clearly identified, and all units in a given region, such as a balancing authority, should be measured against the same benchmark. Although there are some variations in the approach, ELCC is calculated in several steps. Most commonly, the system is modeled without the generator of interest. For this discussion, we assume that the generator of interest is a renewable generator, but this does not need to be the case. The loads are adjusted to achieve a given level of reliability. This reliability level is often equated to a loss of load expectation (LOLE) of 1 day per 10 years. This LOLE can be calculated by taking the LOLP (a probability is between zero and one and cannot by definition exceed 1) multiplied by the number of days in a year. Thus LOLE indicates an expected value, and can be expressed in hours/year, days/year, or other unit of time.

39 California Renewable Portfolio Integration Standard Integration Cost Final Report. Henry Shiu, California Wind Energy Collaborative; Michael Milligan, National Renewable Energy Laboratory; Brendan Kirby, Oak Ridge National Laboratory; Kevin Jackson, Dynamic Design Engineering, California Energy Commission, 2006. Available at http://www.energy.ca.gov/2006publications/CEC-500-2006-064/CEC-500-2006-064.PDF

104


105

Once the desired LOLE target is achieved, the renewable generator is added to the system and the model is re-run. The new, lower LOLE (higher reliability) is noted, and the generator is removed from the system. Then the benchmark unit is added to the system in small incremental capacities until the LOLE with the benchmark unit matches the LOLE that was achieved with the renewable generator. The capacity of the benchmark unit is then noted, and that becomes the ELCC of the renewable generator. It is important to note that the ELCC documents the capacity that achieves the same risk level as would be achieved without the renewable generator. A graphical representation of ELCC appears in Figure 4. Adding the additional generation reduces the LOLP and shifts the reliability curve to the right. At the target of 1 day in 10 years LOLE, the wind capacity value is 100 MW.

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

8800 9000 9200 9400 9600 9800 10000 10200 10400 10600

Load (MW)

LO

LE

(D

ays/

Yea

r)

No Wind

Wind

Target

Figure 4. Effective load carrying capability is 200 MW at LOLE of 1 day/10 years

One concern is what happens if the generator does not generate at the level that was estimated by a prior ELCC calculation. For example, a conventional base load unit may go out of service during the peak period. Although this can put stress on the grid (and the system operator), proper planning usually alleviates the problem because the system is planned, built, and operated to account for such risks. This is why planning reserves are often calculated to be 15% to 20% of projected peak load, allowing for the possibility that some units may not be available when needed. Planning processes in the U.S. often do not perform risk-based analyses of the system, but instead rely on deterministic approaches for capacity planning, such as adding the installed


capacity of all the individual generators and applying a planning reserve margin on top. Many of these analyses use rules-of-thumb that were originally derived from probabilistic methods, and in some regions there is a slow return to probabilistic methods. The potential difficulty of deterministic approaches is that two hypothetical systems that are identical in almost every way could face significantly different risks. This can happen because units with high forced outrage rates (FORs) impose a higher risk of not meeting load than otherwise identical units with low FORs. If one system is characterized by generating units with high FORs and the other by low FORs, the system LOLP/LOLE will be different. Clearly, the objective is to carefully plan for contingencies, and to quantify risks whenever possible. Using probabilistic approaches such as ELCC allows these risks to be quantified and calculated in a systematic, data-driven way. It is useful to examine how a conventional unit would fare under an ELCC evaluation. A generator’s ELCC is driven by several factors, the most important one being the plant’s capacity and forced outage rate. As part of the work assessing the costs of renewable energy integration for the California renewable portfolio standard, a hypothetical conventional unit was modeled at several alternative FORs. An ELCC value was calculated at each FOR so that the impact of the FOR could be seen on ELCC. The benchmark unit is a gas combined cycle with a FOR of 4% and maintenance outage rate of 7.6%. Figure X illustrates the results and shows that the ELCC of this unit declines as a function of the FOR. For example, a unit with a FOR of 60% would have approximately 40% ELCC relative to its rated capacity and the benchmark unit. For conventional generation, the ELCC value often tracks the unforced outage rate (1 – FOR). This is not always true, however, and depends in part on the level of system risk that is used for the base case.

ELCC as a Function of FOR

0

20

40

60

80

100

120

10 20 30 40 50 60 70 80 90

Forced Outage Rate (FOR) %

EL

CC

as

% R

ated

Cap

acit

y

Can be approximated with unforced capacity (1-FOR)*Cap

Figure X. Comparing a generic 100 MW conventional plant to a gas benchmark unit

106


To derive the ELCC of wind, one ideally would have access to several years of wind generation data, load data, and other generation data. The wind ELCC could then be calculated using multiple years of data, which would provide confidence that inter-annual variability has been captured. But because a long wind generation record often does not exist, it is reasonable to expect that wind’s capacity value could vary from year to year. As wind projects come on-line, wind generation data will become available and a database can be created and updated to calculate some type of moving average of the wind capacity value. Examples of how some power pools and regional transmission organizations (RTOs) handle multiple years of data is discussed later in this paper. One way to help solve the problem of the year-to-year variability of the capacity value for wind is to create wind generation scenarios using meso-scale meteorological models. Conceptually there are many variations on this approach. For the Minnesota Department of Commerce (MN/DOC), for instance, Enernex and WindLogics developed a 3-year wind data record by re-creating the actual weather and normalizing to the long-term trend. A variation of this approach may involve the re-creation of several additional years of weather data, then running the reliability model for each of these several years to capture a longer time period. Other approaches have been used that involve Sequential Monte Carlo simulation, discussed further below. Regardless of the method used to calculate wind ELCC, a number of factors can influence the results. The key influence is the interaction of the system LOLP curve, such as the one displayed in Figure 4, and the timing of the wind delivery. Wind that delivers significant capacity during the times of relatively high system risk achieves a high capacity value. Conversely, wind that generates little or no output during these high-risk periods will have a low or zero capacity value. Good siting practices, technology characteristics, and geographic dispersion of the wind plant can all affect the potential delivery and timing of wind generation to the grid, and therefore the ELCC of the wind project. The LOLP curve is subject to several influences. The mix of other generation units, their capacity, and forced outage rates can play a key role. The way that these parameters interact with the load shape has an important influence on the shape of the LOLP curve. In a system with significant hydro generation there can be two distinct influences on the LOLP curve. The first is from the non-controllable hydro (run of river) that has arbitrary influences on the LOLP curve. This will vary from year to year as a function of the hydro flow and changing load shape. Controllable hydro is generally operated so that it benefits the system in some optimal way. Generally, controllable hydro is used to mitigate high risk and therefore will lower LOLP during peak periods. This has the effect of altering the shape of the LOLP curve and can perhaps shift the highest risk hours to near-peak hours from peak hours. Off-system purchases can also influence the risk profile. Because system operators want to ensure sufficient resources during peak periods, it is not uncommon to schedule purchases during peak periods. Of course, that will influence the risk profile and the ELCC of wind.

107


Maintenance on generators is normally deferred to off-peak months in the spring or fall. This is done for obvious reasons: the system operator wants to ensure that all generation is available during the peak periods when the system is most constrained and at highest risk. However, it is not uncommon for the spring or fall maintenance periods to drive up the system risk to levels at or near those found during peak periods. This significantly alters the risk profile, and therefore can play a large role in determining the ELCC of a wind plant. Because the LOLP in any given hour is a function of the load and available generation, it is subject to many influences that include the available online generation and their outage characteristics. When calculating the ELCC of a variable power plant such as wind, there are many hours with significant LOLP that drive the ELCC metric. The wind ELCC depends primarily on the relationship (correlation) between wind power and load, but other factors will also influence ELCC. Generation that is on maintenance is not available so will increase LOLP during those times. Controllable hydro generation is typically used to shave peak and/or scheduled during periods of high prices. Peak periods are generally those periods with highest LOLP, but that is not always the case when hydro and maintenance schedules increase hourly LOLP during lower-load periods. Analysis that was undertaken for the California Energy Commission40 found that during an unusually late, hot summer period when many units were taken out of service for scheduled maintenance, that the hourly LOLP in late September was nearly as high as during the peak summer period. Situations like this can result in a lack of recognition of the exposure of the power supply to potentially high levels of risk that can be overlooked. Any wind generation that would occur during these times would contribute to lowering LOLP, perhaps significantly. Because LOLP and other reliability metrics are heavily influenced by the available generation, the transmission system plays a key role. In larger Balancing Areas the grid allows the pooling of generation that can lower overall risk as measured by LOLP. Therefore, when investigating alternative transmission build-out scenarios or configurations, it is important to perform ELCC or LOLP evaluations holding the transmission system configuration constant between the wind and no-wind cases. For a system with a reliability target of 2.4 hours/year LOLE (1 day per 10 years) the system risk identified by significant LOLP is generally confined to a relatively small number of hours. The number of hours will vary based on several system characteristics, the most important of which is the load profile. Figure XX is a LOLP duration graph that is based on work performed as part of the California RPS Integration Cost study. The graph shows LOLP on the vertical axis and the top risk hours on the horizontal axis. The hours in the graph are not necessarily contiguous, and generally consist of hours of high demand. For the California work, a policy decision was made to eliminate scheduled maintenance from the modeling so that renewable capacity values would be independent of these schedules. In reality, maintenance scheduling of conventional units can have a profound influence on hourly LOLP, and therefore on capacity value for renewable energy. In Figure 2 the area under the curve can be integrated, and is 2.4 hours/year. Any

40 California Renewable Portfolio Integration Standard Integration Cost Final Report. Henry Shiu, California Wind Energy Collaborative; Michael Milligan, National Renewable Energy Laboratory; Brendan Kirby, Oak Ridge National Laboratory; Kevin Jackson, Dynamic Design Engineering, California Energy Commission, 2006. Available at http://www.energy.ca.gov/2006publications/CEC-500-2006-064/CEC-500-2006-064.PDF

108


generation that is unable to deliver during these hours will not receive any capacity value. Conversely, a unit (or units) that are able to fill the LOLP curve will receive a perfect capacity value. In general, the ELCC calculation finds the area under this LOLP curve that is covered by the benchmark unit. Then the capacity value of the renewable generator is the fraction of the risk reduction achieved by the benchmark unit.

Sample Hourly LOLP

1.00E-10

1.00E-08

1.00E-06

1.00E-04

1.00E-02

1.00E+00

0 200 400 600 800

Top LOLP Hours

LO

LP

(L

og

Sca

le)

Figure XX. Loss of Load Probability by Top Risk Hours

This helps us see the impact of two otherwise identical wind plants with alternative chronological delivery profiles. Assume for our discussion that wind plant A averages 30% of rated output during high-risk periods (generally high load) and wind plant B averages 5% of rated during the same periods. The annual energy for the two plants is the same. It should be clear that plant B does little to alleviate the risk of insufficient generation, whereas A does reduce this risk. However, it is important to realize that the capacity value of A and B may not be the same as their output during system-critical periods, although in some cases we have found that the capacity factor during peak periods can do a passable job of estimating the capacity value of the wind plant. We discuss this further below. Visualizing the curve helps us understand why generators’ ELCC declines as the unit size increases. In the simplest case for illustration, suppose that the addition of a 500 MW unit were to reduce all the risk under the curve. At that point, LOLP would essentially be zero for all hours of the year. If the 500-MW unit were replaced by a 1000-MW unit, the capacity value of the 1000-MW unit would be approximately 500 MW because there would be no risk-reduction benefit of the second 500 MW. Because of the potential difficulty of assembling the appropriate database to use for the ELCC calculation, interest in simpler methods has emerged over the past several years. To evaluate the capacity value of a wind plant, it would be desirable to have the ability to carry out the calculation using only the relevant wind data and whatever minimal auxiliary data set. Although

109


several methods can be used to approximate ELCC, an unfortunate aspect of all of these methods is that they are indeed approximations. However, in cases that ELCC can’t be calculated because of data or other limitations, these methods can be useful. In this section we examine several techniques that we are familiar with. Other methods may exist or may be developed in the future. Broadly speaking, the approximation techniques fall into two categories: risk-based or time-period-based. Risk-based categories develop an approximation to the utility’s LOLP curve throughout the year. Time-period-based methods attempt to capture risk indirectly, by assuming a high correlation between hourly demand and LOLP. Although this relationship generally holds, it can be compromised by scheduled maintenance of other units and hydro conditions. A further limitation of time-period-based methods is that all hours considered by the method are generally weighted evenly, whereas ELCC and other risk-based methods place higher weight on high-risk hours, and less weight on low risk hours. However, time-period based-methods are much simpler, and are easy to explain in regulatory and other public proceedings. 3.3.3.2 Simplified Risk-Based Methods

Risk-based methods utilize hourly LOLP information either from an actual reliability model run or as an approximation. The method discussed here can be performed using only a small number of reliability model runs, followed by a spreadsheet exercise to calculate an approximation to ELCC. Garver’s 1966 paper is indeed a classic in the power system reliability literature. The Garver technique to estimating ELCC was applied to conventional generators and was developed to overcome the limited computational capabilities that were available at the time. The approach approximates the declining exponential risk function (LOLP in each hour, LOLE over a high-risk period). It requires a single reliability model run to collect data to estimate Garver’s constant, known as m. Once this is done, the relative risk for an hour is calculated by

R’ = Exp{-[(P-L)/m]} where P = annual peak load, L = load for the hour in question, R’ is the risk approximation (LOLP), measured in relative terms (peak hour risk = 1). A spreadsheet can be constructed that calculates R’ for the top loads. Then modify the values of L by subtracting the wind generation in that hour. Calculate LOLE approximation for (a) no-wind case and (b) wind case by summing the hours. Use all hours for which no-wind risk exceeds some tolerance – probably around 500 hours. Compare to gas plant or other benchmark, de-rated by its forced outage rate. 3.3.3.3 Time Period Methods

To avoid using a reliability model altogether, it is possible to collect only hourly load and wind data for at least 1 year and use these data to calculate an approximation to ELCC. This approach is appealing in its simplicity, but it does not capture the potential system risks that are part of the other methods discussed above. Milligan and Parsons (1999) compared the ELCC with a series of calculations for hypothetical wind generation to determine whether these simpler approaches

110


111

are useful. Although several alternative methods were compared, the most straightforward approach was to calculate the wind capacity factor (ratio of the mean to the maximum) over several times of high system demand. The calculations were carried out for the top 1% to 30% of loads, using an increment of 1%. Figure 5 is taken from that study. Although an ideal match was not achieved, the results show that at approximately 10% or more of the top load hours, the capacity factor is within a few percentage points of the ELCC.

Capacity Credit vs. Capacity Factor (Year 1)

0

0.1

0.2

0.3

0.4

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29

Top Loads (%)

Cap

acit

y F

acto

r/C

red

it

Capacity Factor Capacity Credit

Figure 5. Comparing Capacity Credit Versus Capacity Factor

3.3.4 Recommendations for reliability assessments needed to ensure variable/dispatchable

resources influence on reliability/adequacy is reasonably assessed.

TO Be added based on group consensus


3.4 Transmission Planning (A. Ellis, M. Fecteau, Y. Kazachkov, S. Paquette, P. Pourbeik, D. Schooley)

3.2 Transmission Planning (A. Ellis, M. Fecteau, Y. Kazachkov, S. Paquette, P. Pourbeik, D. Schooley)

This section presents an overview of some of the key issues related to transmission planning as it relates to systems that are seeing a large influx of wind generation. A review is presented on current planning approaches. Then some of the gaps of present approaches are discussed with regards to being able to accommodated large amounts of wind generation integration. The section summary provides recommendation and conclusions related to needed research and development in this area.

3.2.1 A Review of Present Approaches to Planning and Current Grid Codes (AE)

The goal of transmission planning is to identify system improvements required to maintain grid reliability while taking into account the growing demand and interconnection of new generation. This is done by comparing system performance against a set of performance standards established by NERC, Regional Reliability Organizations and other standards development organizations. Power flow, dynamic and short circuit analyses are used to simulate power system performance. Generally speaking, this approach is applied to reliability assessments as well as interconnection studies; although special requirements established by FERC also apply to interconnection studies.

Variable generation resources are playing an increasingly important role in the power system, and have the potential to significantly affect power system performance. Due to their rapid evolution and fundamental differences compared to conventional generations, there is limited industry experience with respect to variable generation operating characteristics and representation. In some cases, these differences also lead to legitimate questions about the applicability of existing standards.

Modeling standards Transmission planning studies are of a regional character and are conducted in a collaborative manner. To support this process, NERC modeling standards require generator owners to submit power flow and dynamic models to the Reliability Entity (RE) to perform regional planning studies. Reliability Entities are required to maintain power flow and dynamic model databases for system studies, and to develop guidelines that define the specific modeling requirements. There are also requirements for periodic model validation. These modeling standards are not being applied consistently to wind power plants. As discussed in section XX, this is due in part to limited industry experience and limited access to appropriate models.

Generator performance standards A fundamental principle in transmission planning states that equipment rating should not be exceeded and no firm transfers should be curtailed as a result of a single (N-1) contingency41. Such contingencies are most often initiated by transmission system faults. Commercial variable generation emerged initially as small-scale projects deployed at distribution feeders. In such applications, emphasis is placed on generators to trip off line for safety reasons. In this context,

41 Intentional, controlled tripping of generation may be permitted under some circumstances.

112


faults tolerance was not encouraged. Voltage and reactive power support expectations associated with synchronous generators were not generally applied to wind and solar generation in these distribution applications. These performance expectations have been adjusted for obvious reasons. In the last few years, standards for wind generators or “Grid Codes” have emerged to address the perceived gaps in the standards. FERC’s Order 661-A, a set of requirements applicable to interconnection of wind generation, addresses voltage ride through (VRT) or fault tolerance, and reactive power capability. NERC has not yet adopted standards that directly address fault tolerance. Applicability of existing reactive power standards originally designed for synchronous generators remains unclear. As a result, reliability organizations including Reliability Entities and various jurisdictions in Canada have adopted or are proposing to adopt standards that address these and other concerns such as frequency tolerance, frequency response and ramp rate limits. Unfortunately, the standards are not yet consistent or stable. Some entities such as WECC have developed standards that apply to all generators, including wind. There are concerns about unintended impacts to conventional power plants.

A comprehensive review of NERC standards is needed to clarify points where there is ambiguity, and to modify or develop new standards where required. This is of paramount importance given the mandatory nature of NERC standards.

3.2.2 Transmission Expansion Required for Integrating Wind (PP)

Many of the regions in North America that are highly suited to wind generation (i.e. high wind capacity factor) tend to be remote from both load and other generation sources. Some North American examples are the panhandle and western regions of Texas [1] and many regions in British Columbia, particularly the North Coast and Vancouver Island [2]. This presents a challenge for integrating such wind resources into the power grid, due to a lack of transmission infrastructure in these remote regions. For such regions transmission expansion is a key phase of integrating wind generation into the system. There are two alternatives: alternating current (ac) or direct current (dc) transmission.

Extra high voltage (EHV) ac transmission expansion has the key advantage of being more accessible to tapping as the grid evolves, and thus allowing easily for intermediate substations, in case of load or other generation growth in the area. However, for very long distances (wind sites that are hundreds of miles from load centers) dedicated high-voltage dc (HVDC) becomes a more economically viable solution. Also, for offshore applications there are technical challenges42 that preclude ac submarine cables beyond distances of roughly 40 km [3]. With the advent of voltage-source converter (VSC) technologies43, there are several other benefits with the use of this type of HVDC technology, in particular for offshore wind plants:

1. Due to the inherent reactive power capability of the voltage-source converter terminals, this type of HVDC is able to significantly improve power quality (i.e. minimize voltage flicker).

2. The voltage-source converter stations have a smaller footprint and weight than conventional HVDC, which makes it more attractive for application on offshore platforms.

42 The capacitive charging of long ac cables precludes their application for submarine applications longer than about 40 km. 43 www.abb.com and www.siemens.com

113

http://www.abb.com/

http://www.siemens.com/


3. HVDC essentially isolates the wind plant from the power grid, thus faults on the transmission system, if cleared normally (i.e. within a few cycles), are not seen directly by the wind plant and thus this provides addition fault ride through capability to the farm.

An example of a wind plant with dedicated transmission are the Gotland HVDC Light project, which is a 70 km dedicated VSC-HVDC line for a 50 MW wind plant in Sweden. Another example is the NORD E.ON 1 project, which is scheduled for commissioning in 2009. This will be a 400 MW offshore wind plant connected to the German power grid via a 400 MW, ±150 kV VSC-HVDC cable. The cable will run 128 km under water to the shore and then a further 75 km under ground to a major 380 kV ac substation at Diele.

3.2.3 Dynamic Line Ratings and Other Considerations for Maximum Utilization of Transmission Capacity (PP)

As discussed in section X, the typical capacity factor of a wind plant is between 0.25 to 0.35. In addition, wind pants run for the majority of the year at megawatt levels much lower than their nameplate capacity. Finally, it is a given fact that when the wind plant is at its peak capacity wind speeds are around 20 m/s. What does all this mean from a planning point of view?

Borrowing from reference 4, let us consider the question of line ampacity. The calculation of conductor steady-state current carrying capacity is largely based on the work of House and Tuttle [5]. The current carrying capability of a bare overhead conductor is based on the maximum allowable conductor temperature and the following heat-balance equation:

qsrIqrqc 2 (1)

where

I = the continuous steady-state current rating of the conductor

qc = convective heat loss (primarily a function of ambient temperature, maximum allowable conductor temperature and wind speed)

qr = radiated heat loss

qs = heat received from solar radiation

r = conductor resistance

So consider the example presented in [4], that is, assume a 795 ASCR Drake conductor, running East to West at a latitude of 50o North, on a clear day with the sun mid sky (12:00 pm), at sea-level, and that the maximum allowable conductor temperature is 100 oC with an ambient atmospheric temperature of 35 oC; i.e. a hot summer day. Thus, the conductor current rating as a function of wind speed (assuming the wind is blowing perpendicular to the conductor) as calculated form (1) is given in Figure 1. The figure shows that as wind speed increased from what is typically used in calculating line ampacity (1 to 2 m/s) to values at which a wind plant would be at full capacity (12 m/s or higher), the line ampacity almost doubles. We must indicate some caveats to this discussion:

The current rating of riser poles or other connected equipment such as breakers, disconnect switched, bus segments, wave traps can often be more limiting than the line conductor – these depend on the equipment and not wind conditions.

114


115

The clearance restrictions on a line that limit the tolerable amount of sag have not been considered here – this may of course be translated into a lower allowed maximum conductor temperature.

Modern wind turbines have towers that exceed 50 m in height, while a typical transmission tower is between 20 to 30 m in height. Thus, there may be a noticeable difference in wind conditions at the wind turbine tower height as compared to the height of the transmission conductors. We have not factored this into the simple calculation presented here.

The calculation here is based on the assumption that the wind direction is perpendicular to the line. The convective losses due to the wind will significantly decrease as the angle between the wind direct and the direction of the line decreases.

Certain lines in the system will extend across a larger geographic region and thus are exposed to different wind regimes. Thus, it may not be appropriate to apply such considerations to these long lines.

Figure 1: Line ampacity (steady-state current rating) as a function of wind speed. It is assumed that the wind direction is perpendicular to the conductor (this yields maximum cooling effects from the wind) and that we has a 795 ACSR Drake conductor, at sea level, running East to West, at 50o North latitude, at 12:00pm and with ambient air temperature of 35 oC – the maximum allowed steady-state conductor temperature is assumed to be 100 oC.

None-the-less, from the simple example above it is clear that there is some potential for increased transfer capacity that may be yielded from increased line ampacity during high wind conditions. To be able to effectively use this potential increased capacity, dynamic line rating techniques need to be utilized. That is, by measuring in real time the actual temperature of the


conductor and providing this information to the system operator, then the operator can more accurately assess the proximity of the line to its thermal limit and thus allow greater loadability. In most present transmission planning methods the allowable line steady-state current limit is calculated similar to the example described above, including consideration for line sag etc. These methods yield a single static limit for each season based on a set of assumed weather conditions (typically average to extreme conditions) for the season. In the case of dynamic rating, the approach is more direct. This method is aimed at measuring the actual average conductor temperature in real-time and thus being able to assess the actual conductor conditions and its proximity to the conductor thermal limit. In this way, one can more accurately assess the thermal limit of the line and more fully utilize its capacity for power transfer. Methods for performing such measurement on major EHV lines due exist [6], [7] and [8]. In [7] it was shown that providing real-time transmission line ratings to the system operator is both feasible and reliable. In addition, the real-time line ratings for the transmission lines monitored in the study up to 40 to 80% more power transfer capacity than the calculated static transmission line ratings used in existing transmission planning and operations. This approach allows using existing transmission lines to their full capacity.

3.2.4 Identify Improvements to Transmission Planning Approaches (PP)

A major challenge with transmission planning, as it relates to wind power generation, is accounting for the variability of the resource. Wind plants spend only a few hours of the year at there peak name plate capacity. Furthermore, when studying a large interconnected system with numerous wind plants, though it may occur, the probability that all wind plants will be at their peak name plate capacity at any given point in time is low, particularly at system peak load. However, in the early 2000’s most system impact studies in North America were performed using a peak load power flow cases married with peak capacity on the generation under study. Though, this assumption may be quite valid for conventional plants (e.g. fossil fuel, nuclear and even hydro in some cases) it is unlikely when considering wind generation. There is, some merit to this approach since it helps to identify fatal-flaws in the transmission system such as if there is adequate transmission to be able to simply inject the peak capacity of the plant into the grid, however, if used to plan for transmission it may result in overbuilding the transmission system.

Form a planning perspective, there are two key questions when performing power flow analysis related to wind plant interconnection:

1. Can the megawatts from the wind plant be injected into the system and reliably transmitted to the nearest major transmission substation?

2. Does the introduction of the wind plant create or further exacerbate transfer capabilities and major transmission paths on the bulk transmission grid?

To answer the first question one can perform analysis similar to what is typically done at present, namely, assume peak output from the plant and identify with N-1 contingency analysis if the peak megawatt output of the wind plant can be reliably carried to the transmission grid. In the simplest case the wind plant is connected to a major transmission substation through a radial line. In this case the radial line should have a continuous thermal rating equal to or greater than the peak capacity of the plant, and the remaining lines emanating from the major substation should ideally not be thermally overloaded under an N-1 contingency when the plant is at peak load.

116


To answer the second question, however, a fresh approach is perhaps needed for variable generation sources such as wind. One approach is as follows [4]:

1. Define the size (number and type of wind turbines) and geographical location of the wind plant(s) to be studied.

2. Using a detail meteorological model as well as wind turbine data (i.e. hub height, power curve, etc.) perform simulations to determine the megawatts generated by each wind plant for 8760 hours in the year. This can be done both for a historic year, where recorded wind data is available, and then using a meteorological model forecasted for future years. This type of work can be done by a number of wind engineering experts.

3. Using a Monte Carlo simulation tool (such as ABB’s Gridview or GE’s MAPS), one can perform a security constraint, economic dispatch for each hour of the year assuming that all available wind power is purchased and injected into the system. This analysis can be performed for both the historic year as a benchmark and for future years. As an input to the simulation tool one would need to enter the daily load curve throughout the year as well as fuel costs, outage rates etc. for all other generating units.

Based on the results of 3 above one can identify what transmission paths are limiting and for how many hours of the year. This analysis can then help to identify what transmission bottleneck may exist that require upgrading or otherwise enhancing to facilitate the full utilization of the wind resources. Further power flow and stability analysis may then be required for those specific limiting cases that are identified potential solutions.

In addition to the above detailed looks at sub-hourly effects may be needed. For example,

1. Typical and extreme ramp-rate scenarios (up and down) in wind power to assess whether the remaining generation and interconnected transfer capability of the grid is able to maintain stable frequency control.

2. Typical and extreme ramp-rate scenarios (up and down) in wind power to assess the effect of such power fluctuations (real and reactive) on voltage control and regulation in the transmission grid.

3. Typical second by second fluctuations in power to assess voltage flicker at the point of interconnection – this is typically only an issue at sub-transmission and weak (low short-circuit) nodes of the system.

Output Levels for Wind Plant for Different Scenario Analysis (AE, PP)

Generator interconnection studies are typically conducted with the wind power plant at full output. Other wind power plants in the study area may also be assumed to be at full output, or at some other output level, depending on their location with respect to the project being studied.

For regional transmission planning studies, the output power level of wind power plants in the region is also dependent on the purpose of the study. In general, the power level should be based on the average output level during the time frame of interest. Average output during a certain time frame varies depending on the location of the wind power plant. For example, the output of in-land wind power plants tends to be low (5% to 25% of nameplate capacity) during the during peak summer load hours. Average output increases during off-peak summer evening hours. Average wind power plant output is significantly higher during the spring, winter and fall. In

117


locations near the coast, wind resource may be driven by other factors such land-water temperature differential, resulting in different seasonal output patterns.

High or low wind output scenarios may need to be considered to identify constraints in certain situations.

Figure 2 illustrates an example of the total wind power distribution in Spain over the years of 2001 to 2005 [9]. This figure illustrates the typical and interesting fact that the total wind generation in a system is rarely at this peak capacity. In this particular example, the median is 26%. That is, the total wind generation is 50% of the time below 26% and 50% of the time above 26% of its capacity. In fact, the probability that a wind power plant is near maximum output or near zero output is high, considering the steepness of the wind-power curve (Figure 2). Thus, it is clear that studying wind generation scenarios just at peak output is not sufficient, and is typically the least likely scenario. Much of this tends to suggest a move towards probabilistic and risk-based planning methods.

0%

10%

20%

30%

40%

50%

60%

20% 30% 40% 50% 60% 70% 80% 90% 100%

Total Wind Generation Output

Per

cent

age

of T

ime

at a

Giv

en O

utpu

t Le

vel

Figure 2: Wind power distribution during the years of 2001 – 2005 in Spain (source: reference [9], © CIGRE 2007).

Another valuable reference related to the subject of power system planning in light of the uncertainties related to variable wind generation is reference 27.

3.2.5 The Need to Consider Interactions Issues (PP)

There are a number of potential interaction issues that may occasionally require detailed analysis [9].

Subsynchronous resonance is a phenomenon whereby series compensation of a transmission lines leads to electrical resonance frequencies in the subsynchronous frequency range that can thus lead to destabilizing modes of mechanical torsional vibration on the turbine-generator shaft that fall in the frequency range of the electrical resonance [10], [13]. Such resonance is less likely to affect wind turbines since the typical torsional mode for a wind turbine is quite low

118


(around 1 to 4 Hz). As such, it would be quite unlikely that the level of series compensation in a system would be high enough to result in an electrical resonance that would interact with such a low mechanical frequency (Note: that the electrical resonance needs to be in the range of 56 to 59 Hz on a 60 Hz system such as North America). The bigger concern is that of induction machine self-excitation [11], [12]. Thus, some detailed 3-phase analysis and discussions with the wind turbine manufacturer on a case by case basis is prudent when installing wind plants near series compensated lines.

Another potential phenomenon related to torsional mechanical modes is device dependant subsynchronous oscillations, often referred to in the literature as subsynchronous torsional interaction (SSTI). This was first observed for the Square Butte HVDC project in 1976 [13]. SSTI is a phenomenon by which controls associated with power electronic based transmission equipment, such as SVC or HVDC, may introduce negative damping torques in the frequency range associated with the torsional mechanical modes of oscillation of nearby thermal turbine-generating units. Again, due to the relatively low frequency range for torsional modes of wind turbine, this may not be a concern in most cases, however, where wind plants are closely couple to a HVDC system some analysis to ensure that controls and/or torsional interaction do not occur is prudent. Such analysis will typically require detailed three-phase models for both the wind plant and the HVDC system. Also, note that as shown in [14] and [15] SSTI is not necessarily always detrimental, in fact in some cases torsional damping can be markedly improved through the application of power electronic devices – one thermal power plant in the Western US grid used a dedicated SVC for this purpose as a means of mitigating the effects of SSR. A good recent publication on the subject of torsional issues related to wind turbine generators is [16]. A practical example of this is the Taiban Mesa wind plant located in New Mexico. This wind plant is located electrically adjacent to a back-to-back HVDC station – Blackwater. The detailed interconnection studies performed by ABB during the design of the wind plant showed that there was little risk of torsional interaction between the HVDC controls and the wind turbine generators. This analysis required detailed equipment level (3-phase) models of the wind turbines, the HVDC and transmission network. In this particular case, there was no identifiably significant interaction between the wind turbines alone and the HVDC.

Other phenomena that may expose the shaft of a wind turbine to cyclical and significant transient torque pulsations may also be a concern. For example, nearby arc furnaces, or high-speed re-closing on a transmission line emanating from the wind plant substation, or repeated commutation failures on a nearby conventional line-commutated HVDC. As a first step, some simple transient stability analysis may be performed to estimate the expected step change in the electrical torque on a wind turbine generator due to the electrical event, and the wind turbine manufacturer consulted to identify if the observed level of transient torque is a concern. Based on consultation with the wind turbine manufacturer, more detailed analysis may be required to assess if a potential problem exists and how it may be remedied.

Finally, due diligence show be done on transient stability studies to ensure that basic control loops in the wind plant (e.g. central voltage control systems often deployed in doubly-fed and full-converter based wind plants that regulate voltage at the interconnecting substation by adjusting the reactive output of all wind turbines in the wind plant) do not interact or interfere with other nearby transmission and generation controls. This is often a matter of simply proper tuning of the controls.

Sections Missing

119


Waiting for the Following write-ups from DS

o Potential reliability impacts of wind generation predominantly requesting energy resource interconnection service and non-firm or conditional firm delivery service DS

o The equivalent of "conditional firm" service for interconnection (not just transmission). That is, allowing transmission development with N-0 reliability instead of N-1 reliability, in recognition of wind being an energy resource as opposed to a capacity resource. DS

o System requirements to support ancillary service and reliability requirements DS

o Recommendations for reliability assessments. DS

3.2.6 Wind Generation Modeling for Power System Stability, Power Flow, and Short-Circuit Analyses [18] (YK, PP)

Modeling wind plants for power flow studies

The objective of a power flow calculation is to determine flows on transmission lines and transformers and voltages on power system buses. This calculation is essential in the planning and design of the interconnection of the wind plant to the system, to ensure that existing equipment is operated within its capabilities and new equipment is properly sized. Such calculations are performed under base case (normal conditions, all equipment generally in service) and contingency conditions (one or more power system elements such as lines, generating units, or transformers out of service) for different system conditions such as peak load, light load, different seasons, or different power transfer. System performance is compared to operating limits and criteria.

From the standpoint of the wind plant, these studies are primarily to determine if the generated power can be transmitted successfully to the loads or purchasing entity without loading or voltage problems. The modeling of the ability (or lack of ability) of the wind plant to control voltage through control of the reactive power output of the units is very important as well.

The model of the wind plant can be considered to have two potential levels of representation:

o A detailed model of the wind plant, representing individual units and the connections between these units and the system. A large wind plant may have over a hundred units. These units are generally spread over a large area, typically connected by a series of feeders. These feeders typically are connected at a “collector” bus which is connected to the power system. Figure 3 shows a typical wind plant topology.

120


P O I o r

c o n n e c t io n to th e g r id C o lle c to r S y s te m

S ta t io n

F e e d e r s a n d L a te r a ls (o v e r h e a d a n d /o r u n d e r g r o u n d )

In d iv id u a l W T G s

In te r c o n n e c t io n T ra n s m is s io n L in e

Figure 3. Wind plant Topology

o The detailed model would thus consist of, say, a hundred or more buses and a similar number of lines. Very detailed data on the system connecting the wind turbine/generators would need to be supplied by the developer. This data is often not available in the early stages of the planning process. These detailed models can be used to determine voltages and flows within the wind plant, as well as the injection into the utility grid. The detailed model can be used to check/design voltage control or reactive power strategies in the wind plant. One example of such a strategy is coordination between a central voltage controller and local power factor controllers.

o The wind plant can be modeled as seen from the system. Here, the concern is not on the individual wind turbines but on the aggregate effect of the entire plant on the power system. The individual generators are lumped into equivalent machines, generally represented at the collector buses. Thus the size of the system representation of the wind plant is reduced to a few buses and the data requirements are significantly reduced. This level of modeling is often used in system studies where the effects of the injection into the system on system flows and voltages are the concern, and internal wind plant conditions do not need to be determined. For the so called interconnection studies, when design of the wind plant is not yet available, usage of the single machine equivalent is a reasonable compromise. The same approach will be the most likely solution when the load flow model is used as a starting point for the stability analysis.

For the latter approach, aggregating different elements of a wind plant can be challenging. Figure 4 depicts the single machine equivalent. Not all the components shown in Figure 4 are necessarily present in all wind plants.

W

Pad-mounted Transformer Equivalent

Wind Turbine Generator Equivalent

PF Correction Shunt Capacitors

Collector System

Equivalent

Interconnection Transmission

Line

-Plant-level Reactive Compensation

POI or Connection to the Transmission

System

Station Transformer(s)

Figure 4. Single-Machine Equivalent Power Flow Representation for a wind plant

121


A wind turbine generator equivalent for the load flow depends on the type of the machine. Despite the large variety of utility-scale wind turbine generators in the market, each can be classified in one of four basic types, based on the generator topology and grid interface, as listed below:

o Type 1 – Fixed-speed induction generators

o Type 2 – Variable speed induction generators with variable rotor resistance

o Type 3 – Variable speed doubly-fed induction generators with rotor-side converter

o Type 4 – Variable speed generators with full converter interface.

For type 3 and 4 units, smoothly controlled reactive capability is achievable within the equipments limits. For the type 3 unit this can be modeled as a P/V bus with the appropriate reactive limits. For the type 4 unit, to be exact, at the limits of the converter the unit is a constant current device an so representation as a P/V bus is not necessarily fully accurate, but it is a reasonable compromise [9].

To accurately represent the type 1 machines a power flow model of the induction machine based on its equivalent circuit would be necessary – an example is discussed in [9]. At the absence of such functionality a widely used approach is to represent the unit by a constant P/Q bus; that is as a constant real and reactive power component. This is a reasonable approach, especially if for the analysis to be performed the machine’s terminal voltage does not vary significantly for base and contingency cases.

To model type 2 machine in load flow, the same approach as for type 1 can be used. However, the reactive power consumption of this machine is even more sensitive to changing system condition because of the variable rotor resistance. This necessitates the need to account for these reactive consumption variations due to voltage sensitivity. Some programs have this feature included in some wind turbine models.

Typically, type 1 and 2 wind turbines are supplied with a power factor correction system comprising several steps of switched capacitors, which are to be modeled explicitly as fixed or switched shunts at the machine bus.

Development of the collector system equivalent impedance is far from trivial task. Usually this is a multi-feeder meshed network that may include overhead and/or cable lines. The goal for achieving a good equivalent is that both the total losses and reactive power flow at the point of interconnection as functions of the wind plant total real power output be reasonably accurate. The cluster approach is, widely used: determine groups of wind turbines that can be considered as radially connected and aggregate them into equivalent units keeping grid elements between them intact.

Some manufacturers supply their wind turbines with local or central (at the point of interconnection) voltage or reactive power controllers. They could be of different implementation, like SVC or STATCOM. The respective power flow model(s) must be added to the case.

Modeling wind plants for stability studies

Wind turbine generators (which convert wind energy to electrical energy) are comprised of a mechanical turbine, an electrical generator and their associated controls. Modeling these systems

122


for power system stability simulation studies requires careful analysis of the equipment and controls to determine the characteristics that are important in the timeframe and bandwidth of such studies. Just as important, the characteristics must be reviewed to put aside factors that can be important for wind turbine/wind plant design but do not play a decisive role for the wind turbine/wind plant response from a system standpoint or whose characteristics are not relevant to the frequency range typical for power system stability performance

The programs used for stability studies usually contain two parts. The first part comprises all dynamic simulation models including models of the equipment physically connected to the grid. These models calculate and update at every integration step current injections from this equipment to respective network buses. The current injections are used by the second part of the program which is responsible for the algebraic network solution and for updating the bus voltage vector at each integration step.

There are a number of components that contribute to the dynamic behavior of a wind turbine generator (Figure 5).

Figure 5 – Generalized wind turbine model with control elements and their hierarchy

These components are [9]:

o Turbine aerodynamics

o Turbine mechanical controls (i.e. pitch control or active stall control)

o Shaft dynamics

o Generator electrical characteristics

o Electrical controls (such as converter controls, switching of shunt capacitor banks etc.)

o Protection relays

o Measuring equipment (such as filters in the transducers for measuring terminal voltage, PLL etc.)

123


The aerodynamic conversion module calculates the mechanical torque applied to the shaft. In the models, this function is typically implemented based on empirical data as a look-up table, with two-dimensional interpolation. Sometimes a problem with the manufacturer’s data is that it does not cover the whole operating range and needs to be extrapolated.

The pitch control, which may seem relatively slow at first sight, noticeably contributes to wind turbine unit response for either wind fluctuations or electrical grid disturbances, especially for wind turbines with no inherent electrical controls (type 1 above). Several versions of pitch control modules have been developed to fit different manufacturers’ implementations.

The shaft module usually simulates the long “soft” shaft as a two mass system with inertia constants, stiffness and damping factor to be determined from manufacturer’s design and field tests. The first mass with the larger inertia refers to the rotor blades and the second mass simulates the combination of the machine and a gear box (if a gear box is used). When faults occur in the grid or there is an abrupt change in the wind speed the potential energy stored in the twisted shaft is released and this mechanical system oscillates at frequencies in the range of 1 to 4 Hz. These oscillations can affect the machine terminal current (see Figure 5).

Figure 5 – Response of the rotor slip of the type 2 machine to the wind ramp with different damping factors

Since the lower range of mechanical resonance frequency is within the bandwidth of interest for stability studies, one can expect that shaft oscillations might adversely affect the system performance. In this regard, the accuracy of derivation of mechanical parameters is critical as it can be seen from Figure 5.

The machine/converter module. The bandwidth typical for stability studies determines those features of the real equipment that must be taken into account. This statement can be illustrated for the example of a doubly fed induction generator (DFIG). The 3-phase rotor terminals of the DFIG are connected to the rotor side power converter. Terminal voltage of the DFIG is determined by controls. In the available implementations, the actual macro control objectives,

124


e.g. real and reactive power, are met separately by controlling respective components of the rotor current [18, 9]. These current control loops are often controlled at frequencies in the kilohertz range that is orders of magnitude faster than power system stability phenomena. Thus simplifications have to be made when performing typical power system stability analysis. Similar simplifications can be considered with regard to the model of the machine itself. It is well known that, for the purpose of stability studies, the machine stator flux linkage dynamics can be neglected. For the conventional induction machine, without any controls, taking the rotor flux linkage dynamics into account is a must. Availability of fast acting controls changes this approach with respect to DFIG. The dynamics of the controls determining the output of the power converter, along with the dynamics of the machine, are so fast with regard to the stability analysis bandwidth that only controls dynamics may be taken into account. The only role of the machine model will be converting commands from controls into the current injected to the network bus in an algebraic way, along with simulating the mechanical rotor movement.

For the synchronous or induction generator decoupled from the grid by a full size power converter, the frequency of the line-side converter current will follow the utility voltage frequency, hence, the unit remains in synchronism with the grid. Both real and reactive power generation and their combination are subject to limits related to the power converter rating and/or limits imposed by the generator and the drive train.

For the directly connected induction generator, the machine model should be similar to standard dynamic model used for stability studies.

For wind turbine models the issue of correct initialization is especially important. The manufacturer’s provided wind-power curve should be used to get the right operating point. For variable speed machines the power - rotor speed curve provided by a manufacturer should also be utilized. A special provision to calculate the initial value of the external rotor resistance, upon given dispatch and terminal voltage for type 2 machines, should be included into the machine module.

Under/over voltage – frequency protection modules. The unit protection trips the generators or the generator along with the power factor correction system when voltage or frequency deviates outside a permissible range. Generally, voltage and frequency are monitored either at the terminal bus or the remote bus and the protection exercises several threshold levels. The voltage protection module should provide for the flexibility to emulate the wind turbine LVRT curve set by a manufacturer.

It is well known that, since the bus frequency is calculated by the stability program as a derivative of the bus voltage angle, the bus frequency may show some fictitious spikes at the instant of fault occurring and/or clearing. The frequency protection model should filter out such spikes. Moreover, it is not quite clear yet if the bus frequency is used as a criterion for tripping off the unit or it is the rotor over/under-speed measured with reference to a frequency.

Fault (Low Voltage) Ride Through Capability

Trying to meet power system needs, manufacturers keep developing sophisticated controls to expand the band of voltage and frequency where wind turbines remain in operation despite severe transients.

For fixed speed wind turbines employing conventional induction machines, a combination of turbine blade pitch control modifications, together with additional smoothly controlled reactive

125


power support (SVC or STATCOM) is needed to solve this problem [9]. For DFIGs modifications such as an active-crowbar circuit is needed to protect the converter power electronics and thus allow the unit to ride through faults [9]. For type 2 machines, again a combination of additional smoothly control reactive support and attempts to reduce further stabilize the unit by increasing the effective total rotor resistance helps achieve fault-ride through. For type 4 full size converter units, the mechanism of fault ride-through is achieve much the same way as with other voltage-source converter technologies – the generator over-speed may need to be sustained by dynamics breaker resistors. What is important from a power system stability modeling stand point is the ability to properly emulate the behavior of these fault ride-through schemes. Modeling the details of these systems may not be practical, particularly in the context of positive-sequence stability models. Reference 9 provides an in-depth discussion on the physical principles involved and the methodologies utilized for ensuring fault ride-through and the issues surrounding modeling of this phenomenon.

Dynamic model validation

Unfortunately, at present this is a very limited amount of field or factory test results for validation of wind turbine generator models. The primary approach to model validation has been to compare results of the simulations using stability models as mentioned above versus results of the simulations obtained by manufacturers using their 3-phase equipment level models used for design – such models are typically develop in software packages such as PSCAD®/EMTDC or MatLab®/Simulink.

More work is being done by some utilities and entities like the WECC MVWG to setup field measurements and collect field recording data.

Short circuit analysis

Both platforms used in the USA for system planning studies, namely Siemens PSSTME and GE PSLFTM, deal with fundamental frequency phasors of voltages and currents. It is accurate enough to say that results of the dynamic simulation represent the positive sequence components of voltages and currents. The same is true when simulating the response to a fault of the system with wind plants.

When using these platforms for a classical fault analysis the power flow data must be supplemented by sequence data for all pertinent equipment including machines.

ANSI/IEEE Standard 37.04 is generally used to examine or determine the short circuit duty of ac high voltage breakers rated on a rms symmetrical current basis. This standard makes recommendations shown in Table 1 regarding values of a reactance, resistance and X/R ratio to be used for calculating the fault current contribution from conventional induction machines split into 2 groups:

o large machines: above 1000 HP at 1800 RPM or less & above 250 HP at 3600 RPM

o medium machines: from 50 to 1000 HP at 1800 RPM or less & from 50 to 250 HP at 3600 RPM

Fault current contribution from small machines, notably below 50 HP, is generally neglected.

Table 1. Recommendations of the ANSI/IEEE Standard 37.04

Induction Machines Positive Sequence Reactance for calculating Interrupting Duty Closing and Latching

Duty Large 1.5 Xd” 1.0Xd”

Medium 3.0Xd” 1.2Xd” Approximate Resistance

126 Large 1.2 times the dc armature resistance Medium 1.2 times the dc armature resistance

X/R Ratio


For type 1 wind turbine generators the recommendations of Table 1 should be used to populate the machine sequence data.

For types 2 and 3 the conservative assumption is to consider these machines for the fault analysis as conventional induction machines. Initiated by the rotor current during the fault, the crowbar protection for type 3 machines may activate and thus may block the power converter. If it does the DFIG machine is being transformed into a conventional induction machine. If it does not, the equivalent rotor resistance may be larger or smaller than the dc rotor resistance depending on whether the machine was operating at under- or over-synchronous speed prior to the fault. However the effect on the fault current compared to the conventional machine will be negligible.

For synchronous or induction generators decoupled from the grid by a full power converter (type 4) the fault current contribution is in the order of or less than 1 p.u., due to very fast blocking capability/controllability of the grid side converter.

3.2.7 How Should Models be Utilized for Traditional System Impact Studies (AE, PP)

The purpose of system impact studies is to identify transmission system reinforcements to ensure acceptable system performance is maintained after a proposed generator is interconnected to the system. System impact studies encompass power flow, dynamic and short circuit analyses. In some cases, it is prudent to also perform interaction or harmonic analyses. Each of these analyses requires different models. The bulk of the system impact study analyses consist of power flow and dynamic simulations, and positive-sequence models are the most common simulation tool.

Model limitations are often misinterpreted. Evaluation of fault tolerance is a good example. When a single-machine representation is used, individual generator terminal voltages are not observable; therefore, it is not possible to determine whether a wind power plant will comply with fault tolerance requirements. Control actions critical to fault tolerance may not be represented in the models due to the short time frames involved. For some types of wind generators, tolerance to unbalanced faults may be more restrictive than tolerance to balanced faults. More detailed studies using 3-phase detailed models and simulation platforms may be needed to ensure compliance.

Identification of reactive reinforcements is an important aspect of system impact studies. Unfortunately, there is confusion about the base-line requirement. For synchronous generators, a power factor (pf) range of +/- 0.95 and ability to continuously control voltage are commonly required. This capability is standard offering for synchronous generators; however, it is not clear whether wind power plants are held to the same standard. For example, interconnection requirements in FERC 661-A state that reactive power range, up to a maximum of +/- 0.95 pf,

127


can be required if the need is shown in the system impact study. FERC 661-A further states that the need for steady-state voltage control also needs to be demonstrated by the system impact study. What is missed by these statements is that system impact studies focus on stability and fault ride-through issues and it is not necessarily a trivial task to illustrate the need for voltage regulation and reactive power needs in a system impact study (unless of course these needs are for maintaining system stability), since the problem with an inability to regulate voltage often leads to day-to-day operating issues – see the system operation section of this report. Most wind power plants require external reactive power support to provide a power factor range of +/- 0.95 at the point of interconnection. In the absence of a clearly applicable standard, the +/- 0.95 pf and ability to control voltage are not base-line requirements.

3.2.8 Modeling requirement for positive-sequence models (SP)

The models for wind turbine generators must represent accurately the dynamic response of the wind turbine relevant to power system stability analysis. There are a number of components that contribute to the dynamic behavior of a wind turbine, as shown in Figure 5.

The models must reflect the inherent characteristics of the machines and the actions of the wind turbine control systems. They must include all components that have a significant impact on the dynamic response. There are a number of simulation packages developed to perform system studies. The models must be compatible with the simulation package required by the transmission planner.

When a user model is delivered, complete technical documentation data and parameters are required. The technical documentation should include block diagrams of the different components, detailed descriptions of the strategies used for real power, reactive power, voltage and speed control systems under different operating conditions or any other particular information that can help to understand the behavior of the wind turbine (e.g. expected short-circuit ratio at the point of interconnection at which the wind plants can operate properly, the maximum delay that a wind plant can remain in service without tripping during a three-phase fault at the point of interconnection). For computational reasons, the models must allow all wind generators in a plant to be represented as a single generator (aggregated model). Also, the models should be robust for a simulation time step of no less than ¼ cycle (4.16 ms for 60 Hz systems) to avoid time consuming simulations.

In order to perform time-domain stability studies, the models must be designed faithfully and reasonably represent the machine dynamics for both short term (few seconds) and long term (tens of seconds to minutes) simulations. In addition, some utilities use extended term simulation software (i.e. software that allows for large simulation integration time steps by having only the longer term dynamics of devices modeled). As such, there is a need to develop models adequate for use in such tools. Furthermore, the procedure to initialize and import a wind plant model for power flow and stability studies should be straightforward, i.e. similar to that of conventional generators. The models should have the same flexibility as any models available in simulation programs and should not be dependant on the utilization of a user-written or unique and separate routine program (e.g. Python script etc.). The models must be valid for all levels of active power settings.

In order to take into account specific system behaviors (such as power swings, slow voltage recovery, overvoltages), the wind turbine manufacturers should give indications to the

128


transmission planners on the degree of flexibility of the wind turbine performance. The transmission planners need to know the possible operating margin of the wind plant (what are the active power and reactive power capabilities during and after a disturbance). This information is crucial to optimize the choice of the equipment required for the integration of the wind plant to the grid, regarding stability concerns, and thus has a direct impact on the cost of investment. To improve or maintain the system behavior, the transmission planners can then choose between the optimization of the wind plant’s behavior and the addition of new equipment on the grid. In order to optimize the wind plant’s performance, each user model should ideally come with a list of parameters (gains, time constants etc…) with a given range of values provided by the wind turbine manufacturers.

Since the dynamic response of the wind plant, as obtained from the models directly, impacts the choice of equipment and the investment required for stability constraints, the validation of the models is an important step. Compliance test results must be provided with the models demonstrating that they are representative of the behavior of the real equipment under equivalent conditions. Test results are also required by transmission planners to demonstrate that the wind plants comply with voltage and frequency requirements described in grid codes.

3.2.9 Modeling wind plants for transient studies (MF, SP, PP)

Wind plants are often located in remote areas of the network where the short circuit level is weak and where problems such as overvoltages, harmonics or voltage unbalances can be observed. Furthermore, wind projects may be located near HVDC interconnections or near series compensation and interact with these equipments (see section 3.2.5).

To study the interaction between wind plants and the network, 3-phase transient models such as developed in EMTP, Matlab® or PSCAD® are sometimes required by utilities. They are not only useful to integrate a wind plant but they are also required by some utilities to cope with the evolution of the network. These models must represent not only the wind turbines but the whole wind plant with its cables, transformers and voltage regulating equipment such as SVCs or STATCOMs. They must be well documented in order to clearly know their limitations. Transient models should be able to represent the wind plant from minimal power production to full power operation. Transient models typically reproduce behaviors of phenomena that require a simulation time step from 1 to 50 µs that last up to 5 s.

Usually, 3-phase transient models for single wind turbine generators are developed and used by wind turbine manufacturers for the design of the wind turbines and are therefore available. However, manufacturers are often reluctant to share these models with utilities for confidentiality reasons since these models are far more detailed than those used for power flow and stability studies, and thus may divulge some of the proprietary details of the manufacturers design. In addition to this, such 3-phase transient models may often be developed in a proprietary software of the manufacturer – this is thus another barrier in sharing these models as it would require translating the model to a more readily available platform used by utilities. Some transient simulation software programs are now capable of compiling code developed for other software or can allow the user to encrypted models in order to keep proprietary information confidential (black box model) and modeling efforts to a minimum.

Some of the reasons why utilities could require a 3-phase transient model are listed below:

129


o Presence of series compensation for subsynchronous resonance studies (see section 3.2.5).

o Presence of HVDC equipment (see section 3.2.5).

o Integration of a wind plant on a very low short circuit network.

o Presence of harmonics or a network sensitive to harmonics.

o Networks where special protection systems (SPS) are required in order to make sure protections perform adequately, islands can be detected, current zero crossing occur and over voltages are manageable.

o Determine the fault-ride through performance of the wind plant to network events.

o Determine if protections and controls of the wind warm will interact with network equipment, controls and protection systems.

Thus, it can be seen that such detailed model are required in special situations. The manufacturers should be willing and able to work with utility planers, and their consultants, in these situations to provide the necessary detailed models and model documentation. It is understood that such detailed equipment models may be deemed proprietary. However, there are means to address such issues for example encrypting the components of the model that contain proprietary information.

3.2.10 Recommendations for Model Development (SP)

In the last few years, significant progress has been made in the development of positive-sequence models. However, based on Hydro-Quebec TransEnergie’s [25] experience (and those of others [26]) a lot of progress still needs to be made. Since 2004, Hydro-Quebec TransEnergie (HQT) has validated 15 Siemens PTI PSSTME models coming from nine wind turbine manufacturers. HQT received up to 6 releases for one model for a total of 35 releases in total for all models. The totality of the models received showed either software problems or did not represent well the true dynamic response of the real wind turbines under specific system behaviors. The wind turbine manufacturers and model suppliers need to work more closely to ensure validated and functionally robust models. Listed below are the main aspects for which there are some gaps that the wind turbine manufacturers or model suppliers need to address. Based on HQT’s experience, the totality of the 15 models received had to be modified to correct the software problems or inappropriate dynamic response. For the first three aspects listed below, more than 20% of the models validated by HQT displayed this problem.

The positive-sequence models should:

o show no problem of any kind with initialization

o prove to be robust during disturbances (no “not converged” during and following a fault simulation)

o take into account the system frequency (under frequency deviations, the models must represent the behavior of the real wind turbines)

o provide some flexibility to the users to allow the tuning of some parameters (gains, time constants, etc.) so the dynamic response of the model can be optimized for specific system behaviors, and still reflect the true behavior of the wind turbines.

130


Finally, some utilities (e.g. HQT) use extended term simulation software (i.e. software that allows for large simulation integration time steps by having only the longer term dynamics of devices modeled), thus there is a need to develop positive-sequence models adequate for use in such tools.

Since the transmission planners select the additional equipment required for stability constraints based on the positive-sequence models, it is crucial that these models are validated. However, HQT’s experience shows that the field test reports are not easily obtained. Since they have a direct impact on the cost of investment and on the system behavior, the models should be submitted to a validation process44. Transmission planners should receive compliance tests demonstrating that the models are representative of the behavior of the real equipment under equivalent conditions as well as test results proving that the wind plants are compliant with voltage and frequency requirements described in grid codes.

Utilities who believe transient models are required for wind plant integration should request these models in their grid code specifying the reasons why they need the model in order for the manufacturer to make and give them a model that suits their needs.

3.2.11 Section Summary and Recommendations (PP)

In this section we have presented a discussion on the impacts of wind generation on the transmission planning process. From this discussion it is apparent that the following new tools, techniques and standards need to be further explored in order to facilitate the proper and reliable large scale integration of wind (and likely other variable) generation into the North American power grid.

o Standard, generic, non-confidential power flow and stability (positive-sequence) models for wind turbine generator technologies (and other variable generation) for use in the standard commercial power system simulation programs. Such models should be readily validated and publicly available to power utilities and all other stakeholders in the utility industry.

o Methods need to be developed and identified for suitably validating such models.

o It should be understood that more detailed 3-phase equipment level models for wind turbine generators are also sometimes needed for cases such as the application of wind generation near series compensated line or high-voltage dc transmission. In these cases such detailed models are needed to investigate the potential for complex control and resonant interactions. Thus, manufacturers need to have such models available for working with utilities, consultants and researchers on such potential concerns. This area requires further research.

o There is an imminent and urgent need for new tools and techniques for transmission planning in light of large scale integration of variable generation. This is because the uncertainty in the resource leads to uncertainty in the planning goals. For example, overbuilding transmission based on looking only at peak load hours combined with peak wind generation output, which is a quite low probability scenario. As such, the new tools

44 An interesting example of the impact of the models on power system study results was documented in studies performed for the Alberta Electric System Operator in 2004. (P. Pourbeik, “Sensitivity Analysis on Low-Voltage Ride-Through Requirements”, June 21, 2004, available at http://www.aeso.ca/files/MemoFromABB_SensitivityAnalysis_Final.pdf).

131

http://www.aeso.ca/files/MemoFromABB_SensitivityAnalysis_Final.pdf


and techniques will need to consider multiple scenario analysis, possibly looking at 8760 hour a year planning methods, and ultimately probabilistic methods and tools. There is a need for further research in this area.

References: [1] ERCOT System Planning, “Competitive Renewable Energy Zones (CREZ) Transmission

Optimization Study”, April 2, 2008.

[2] www.bctc.com

[3] IEEE Task Force Report, Blackout Experiences and Lessons, Best Practices for System Dynamic Performance, and the Role of New Technologies, IEEE Special Publication 07TP190, July 2007.

[4] P. Pourbeik, “Wind plant Integration in British Columbia – Stages 1 & 2: Planning and Interconnection Criteria”, ABB Report Number: 2005-10988-2.R01.3, March 28, 2005. (www.bctc.com)

[5] H. E. House and P. D. Tuttle, “Current-Carrying Capacity of ACSR”, AIEE Transactions, pp. 1169-1177, February 1959.

[6]http://library.abb.com/GLOBAL/SCOT/scot221.nsf/VerityDisplay/EACE83BD6A60D884C12570D0002F99B4/$File/1002_LTM_PSGuard_Datasheet.pdf

[7] California Energy Commission, Dynamic Circuit Thermal Line Rating, October 1999. (http://www.energy.ca.gov/reports/2002-01-10_600-00-036.PDF)

[8] EPRI TR- 113391, Development and Field Application of EPRI Dynamic Thermal Circuit Rating (DTCR), 1999.

[9] CIGRE Technical Brochure 328, Modeling and Dynamic Behavior of Wind Generation as it Relates to Power System Control and Dynamic Performance, Prepared by CIGRE WG C4.601, August 2007 (available on-line at: www.e-cigre.org)

[10] P. M. Anderson, B. L. Agrawal and J. E. Van Ness, Subsynchronous Resonance in Power Systems, IEEE Press, New York, 1990.

[11] P.M. Anderson and R. G. Farmer, Series Compensation of Power Systems, ISBN 1-888747-01-3, 1996

[12] C. F. Wagner, “Self-Excitation of Induction Motors with Series Capacitors”, AIEE Transactions, pp.1241-1247, Vol. 60, 1941.

[13] M. Bahrman, E. Larsen, R. Piwko, H. Patel, “Experience with HVDC – Turbine Generator Torsional Interaction at Square Butte”, IEEE Transactions on Power Apparatus and Systems, Vol. PAS-99, pp. 966-975, May/June 1980.

[14] D. Dickmander, P. Pourbeik, T. Tulkiewicz and Y. Jiang-Häfner, “SSTI Characteristics of HVDC Light”, White paper by ABB Inc., December, 2003.

[15] P. Pourbeik, A. Boström and B. Ray, “Modeling and Application Studies for a Modern Static VAr System Installation”, IEEE Transactions on Power Delivery, Vol. 21, No. 1, January 2006, pp. 368-377.

132

http://www.bctc.com/

http://www.bctc.com/

http://www.energy.ca.gov/reports/2002-01-10_600-00-036.PDF

http://www.e-cigre.org/


[16] R. K. Varma and S. Auddy, “Mitigation of Subsynchronous Oscillations in a Series Compensated Wind plant with Static Var Compensator”, Proceedings of the IEEE PES General Meeting 2006, Montreal, Canada, 2006.

[17] FERC Order 661-A

[18] WECC Wind Generator Power Flow Modeling Guide

[19] NERC Standards

[20] WECC Generator Fault Tolerance Standard

[21] IEEE paper – Ellis – Muljadi – Wind Power Plant Collector System Equivalencing

[22] IEEE paper – Ellis – Muljadi – Validation of WECC generic models

[23] IEEE paper – Ellis – Behnke – Reactive power requirements for wind power plants

[24] Siemens/PTI – PSS/E model user manual or wind generator application guide

[25] General Electric – PSLF model user manual or wind generator application guide

[26] Y. Kazachkov, J. Feltes, R. Zavadil, J. Santos, “Modeling Wind plants for Power System Stability Studies,” IEEE PES 2003 General Meeting, Toronto, Canada, Paper 03GM0946, 2003.

[27] Y. Kazachkov, S. Stapleton, “Does the Generic Dynamic Simulation Wind Turbine Model Exist?,” WindPower 2005, Denver, CO, May 2005.

[28] Y. Kazachkov, R. Voelzke, “Modeling Wind plants for Power System Load Flow and Stability Studies,” IEEE PowerTech 2005, St. Petersburg, Russia, June 2005.

[29] Y. Kazachkov, D. Brown, K. Patil, J. Senthil, S. Stapleton, “Work Scope and Tools for Wind plant Interconnection Studies,” Siemens Power Technology eNewsletter, Siemens PTI, Schenectady, NY, January 2006.

[30] A. Ellis on behalf of WECC WGMG, “Generic Wind Plant Models for Power System Studies,” WindPower 2006, Pittsburgh, PA, June 2006 (with WECC WGMG).

[31] Y. Kazachkov, C. Grande-Moran, Q. Liu, B. Lam, “Voltage Stability and Short Circuit Issues when Integrating a Wind plant with the Grid,” CIGRE-Canada, Montreal, Quebec, Canada, October 2006.

[32] Y. Kazachkov, B. Lam, K. Patil, “Experiences in Dynamic Stability Modeling of Wind plants for Grid Integration Studies,” CIGRE-Canada, Montreal, Quebec, Canada, October 2006.

[33] M. Behnke , A. Ellis , Y. Kazachkov , T. McCoy , E. Muljadi , W. Price , J. Sanchez-Gasca,, “Development and Validation of WECC Variable Speed Wind Turbine Dynamic Models for Grid Integration Studies,” WindPower 2007, Los Angeles, CA, June 2007.

[34] P. Pourbeik, R. J. Koessler, D. Dickmander and W. Wong, “Integration of Large Wind plants into Utility Grids (Part 2 - Performance Issues)”, Proceedings of IEEE PES General Meeting, July 2003.

[25] Hydro-Québec TransEnergie, “Technical requirements for the connection of generation facilities to the Hydro-Québec transmission system”, May 2006

133


[26] ESB National grid, “Dynamic modeling of wind generation in Ireland”, January 2008

[27] CIGRE Technical Brochure 293, Electric Power System Planning with the Uncertainty of Wind Generation, Prepared by CIGRE WG C1.3, April 2006 (available on-line at: www.e-cigre.org)

134




Long Range Transmission Planning with Wind Energy Resources (DO)

Need to determine whether material in this section is a separate section or should be integrated into resource adequacy and trans planning sections. If separate, need to coordinate w/current section 3.2.2 and differentiate LT trans planning from Intermediate?. Pounding wind energy resources into the traditional planning methodology does not work well for wind energy levels above 10% of the load energy. Planning reserve levels, operating reserve levels, ancillary service requirements and the capacity credit of wind generation are affected significantly when transmission is added for higher wind energy penetration levels. Not only are the planning and operating processes more tightly coupled with the increased power transfer levels associated with significant energy resource integration on the power system, the need for the frequency of studies increases especially during periods before 2025 when wind generation is expected to expand rapidly based on state mandates. There appears to opportunities for mode shifts in the planning process. Combinations of modes may be in process in an area or region, but there is a tendency to progress serially through the modes also.

1. The first mode occurs for wind energy penetration levels below approximately a 10% wind energy integration level.

2. The second mode occurs between 10% and 15% wind energy integration. 3. The third mode occurs above 15%wind energy integration. 4. The fourth mode is concerned with exports of wind capacity and energy.

Modes 1,2 and 3 assume that wind energy mandates are the reason that wind generation would be constructed. The customers of the state with the wind mandate pay for the revenue requirements of the wind generators. Wind energy is a price taker in the Energy Markets. Wind generation is allowed to run if it is not transmission constrained. Base generation is bumped up higher on the economic order of dispatch list and more base load generation is then on the margin. Since wind has a relatively low capacity credit (15% for Midwest ISO generation forecasts), wind mandates tend to create fairly large supplies of energy for sale. If these supplies of energy for sale are of sufficient magnitude and there is a higher price area that the energy could be sold, then transmission expansion may be supported by the economic benefits from the energy transactions. Modes 1 and 2 produce an incremental expansion that may take advantage of some of the sales of energy potential. Mode 3 is a large step in power transfer capacity that has the potential of maximizing the benefits available from energy sales. Mode1 is the mode that most areas are currently operating. The wind penetration level is less than 10% wind energy. Wind energy penetration levels are low enough that the traditional planning processes of sequential expansion and managing wind variability in a balancing control area is satisfactory. Improvements such as the Competitive Renewable Energy Zone processes have improved the sequential nature for the future site selection of wind generation and a coordinated plan of transmission expansion. The limitation of the amount of wind penetration is due to the ability to manage the wind variability and extreme events with the generation resources in the area alone.

135


Mode 2 occurs for wind energy penetration levels between 10% and 15%. At this point concentrated wind energy in an area cannot be managed by the resources in the area. Base load generation would have to be heavily cycled for the local generation to follow the wind variations. The additional stress on the base load units and the probability of not having the generation when it is needed for the next day peak, poses reliability concerns as well as economic consequences. Consolidation of balancing areas and limited expansion of transmission to share ancillary services over a large generation base are two regional planning concepts that are introduced in Mode 2. The Midwest ISO is presently operating for the shorter term transmission expansion in this mode. An Ancillary Services Market is anticipated to start September 9, 2008 with the Midwest ISO performing as a single control area. Limited transmission expansion is planned to link the wind rich areas to a large generation base that can supply the Ancillary Services required to manage wind energy variations. While economic planning methods can effectively be used in the planning process, there is a tendency to try to use traditional planning tools for Mode 2 planning. The Midwest ISO uses both economic and traditional study methods to plan the transmission for the mid range period that Mode 2 represents for the Midwest ISO with a planning horizon for the year 2018. Recent studies are showing that the eastern part of the Midwest ISO system may be able to be planned in Mode 2 with a sequential transmission expansion integrated into the present 765 kV system. Mode 3 occurs for wind energy penetration levels above 15%. The volume of low cost energy that is available at these higher penetration levels the presence of high energy price differential to the east of the Midwest ISO present an economic opportunity for a large transmission expansion. The transmission expansion in Mode 3 requires that a high power transfer multiple line transmission overlay be constructed complete with the capability of being able to back up its own contingencies. The distance from the western Midwest ISO state to the east coast where the highest prices exist is greater than 600 miles. 600 miles is about the economic break over point between AC and HVDC systems. HVDC also has the capability of being able to collect energy and deliver energy to specific buses without imposing a loop flow across transmission systems that may not wish to participate in the project. Free rider opportunities would also be minimized since HVDC is scheduled and not subject to inadvertent or loop flows. The minimum power transfer level for an 800 kV (6400 MW rated) system of three HVDC lines is about 15,000 MW. Recent Midwest ISO Transmission Expansion Studies (MTEP 08) indicates that it may be economically feasible for a Mode 3 type transmission expansion with the revenue from energy transactions over the line paying for the revenue requirements of the line and have some margin remaining. The Midwest ISO includes the Mode 2 study results in the Mode 3 processes. The Midwest ISO, PJM, SPP, TVA, the NYISO, the New England ISO, Entergy, MAPP and other interest parties are in the process of investigating the impacts of wind energy for potential transmission expansion for the U.S. Eastern Interconnection through an open process voluntary study organization named the Joint Coordinated System Plan(JCSP) The DOE National Renewable Energy Laboratory ( NREL) through the Office of Energy Efficiency and Renewable Energy is coordination with the JCSP to produce mesoscale hourly and ten wind generation output models for the JCSP footprint to a two kilometer square resolution for the years 2004,2005, and 2006. The first year of model data is expected in late June, 2008. Talks with NREL and WEICAN of Canada have been initiated to produce wind models which would include Canada. Canada has some observers to the present JCSP process, but they are not part of

136


the planning process. The hourly wind data will be used in the transmission forecast development. There is some reluctance by the politically correct to call a conceptual plan a “plan”. Thus the use of the word forecast. Plans are said by some to be complete and ready to submit to the approval and construction processes. JCSP will produce transmission forecasts for a 5% wind energy penetration for the U.S. Eastern Interconnection based on present wind mandates. A 20% U.S. Eastern Interconnection wind energy scenario will also be studied. The study is expected to be completed near the end of 2008. NREL is also commissioning the East Wind Integration Study that uses the transmission forecasts developed by the JCSP for the years 2004, 2005 and 2006 to determine the short term (1 second to a few hours) interaction between load, wind energy and the forecasted generation mix. Reserve levels, ramp rate reservations, transmission reserves for ancillary services, and the evaluation of ACE and frequency control are some of the outputs expected from the East Wind Integration Study. Mode 3 studies were initialized with fixed planning reserves based on the present practices. The generation forecast was based on the same assumptions. The capacity credit for wind uses the ELCC method that is dependent of wind energy diversity and the LOLE calculations. The LOLE calculations are based on the available power transfer level between generation groups. The East Wind Integration Study will probably increase the operating reserve models and possibly change the unit commitment model that is used in the PROMOD economic simulation phase of Mode 3. The Mode 3 study process probably has to be repeated in 2009 to reflect these major changes in the power system to converge on the most economic evaluation. Figure 1 is a process flow diagram for Mode 3 forecasting. The inputs of wind data is shown. Mode 3 transmission may provide a 15,000 MW interconnection capability. Coordination with Modes 1 and 2 would be necessary to produce a least cost solution.

137


138

Determine FuturesRenewable Future20% Wind Energy

+Env ironmental

$25/Ton CarbonTax+..

Generation Forecastand

Generation Location

Transmission Dev elopment

Ev aluation of other Futureswith this Futures

Transmission

Selection of A RobustTransmission Conceptwith Future Specif ic

Transmission Expansions

Reliability Study

Wind Integration StudySimulations to DetermineAdequacy of GenerationControls to Fit the Short

Term

Real Time Simulation

Wind Hourly Data

LOLE

Wind Data f romSeconds to Hours

Power System Conceptual Design

Steps 1, 2 of JCSP

Step 3 of JCSP

Step 5 of JCSP

Step 4 of JCSP

Step 6 of JCSP

DOE NREL Wind Integration Study

No JCSP Plan to Perform MISO MTEP 09 Process


139

Steps 1,2 of the JCSP Process which is planning Mode 3 produces a unique generation forecast ( Figure 2) and sites the generation( Figure 3) in the forecast according to stakeholder inputs and rules such as using brown field sites, location of water, location of transportation, etc. The generation forecasts for the Midwest ISO Transmission Expansion Plan 08 four futures are shown in Figure 2. The four futures are the Reference with existing wind mandates, the Environmental with a $25/ton carbon tax, a Fuel constrained future with limit natural gas for generation, and a Renewable future with 20% wind energy. The large generation capacity surplus for the Renewable future is available mainly off peak and appears on the Energy Market mainly as displaced low cost coal energy.

Figure 2

Future Capacity Requirements 2008-2027

6,326 6,326 6,326 6,326

12,000

22,800

9,60027,600

18,000

4,8003,600

6,000

4,800

3,2001,280

4,160

2,880

13,800

13,200

40,50012,600

72,506

56,686

46,006

50,926

-

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

Ref Env Fuel Ren

Future

Ad

de

d C

ap

ac

ity

(M

W)

Queue Nulcear Coal CC CT Wind

Figure 3 is the location of generation for the 20% Wind Energy future for the JCSP. The generation forecast is linked to buses at the forecast locations in the PROMOD economic simulation program that models both the transmission and the generation of the Eastern Interconnection.


140

19 Figure 3

Step 3 of the JCSP( Mode 3) process is to determine the transmission that economically deliver energy within the study footprint. Processes determine the economic sources and sinks, the interface flows across critical geographical boundaries, formulates a least cost transmission conceptual plan or forecast that has the capability of delivering the transmission. The plan is then evaluated for its economic performance as to the benefit to cost ratio of the conceptual transmission plan. Figure 4 is a conceptual plan for the MTEP 08 Renewable Future with 40,000 MW of wind generation. The benefit to cost ratio was 1.1:1. The JCSP method provides a means of measuring the economic performance of conceptual transmission systems and provides the areas of existing constraints that could further be relieved to improve the performance. Hourly flows across interfaces can be used to determine the critical loading periods to test for the peak stress of the transmission system. Figure 5 shows the Ohio-Pennsylvania and South interface flows. The wind pattern contribution can be seen from the lower power transfer during peak periods and the high off peak transmission loadings. The variability of the wind can be seen from the flow patterns on the lines comprising the interface. The peak transmission loading periods are in January, which is not the present focus for reliability studies. Several thousand MW of unscheduled power transfer capacity is available on peak. The loading periods and on peak available power transfer capacity may change the types of studies that NERC is performing to assure reliability. The evaluation of the availability of resources may be more important for reliability assessments than the performance of the transmission system.


Transmission Expansion for Renewable Future year 2021

Black dotted - 800 kV HVDCBlue dotted - 765 kV AC Yellow dotted - 345 kV AC

Figure 4

West to East Interface Flows OH-PA

0

5000

10000

15000

20000

25000

0 720 1440 2160 2880 3600 4320 5040 5760 6480 7200 7920 8640

Hour of the Year

MW

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 5

141


Steps 4 and 5 of the JCSP ( Mode 3) planning process is a probable risk analysis assessment that is used to determine one recommended core transmission plan that could be adapted to serve any of the futures with some future specific transmission expansions. Step 6 is the traditional Reliability studies to assure that the transmission system would meet NERC standards. More off peak power transfer conditions are addressed than presently. On peak cases with a low wind probable level is also studied. The Wind Integration and the Real Time Simulation are processes that address the short term performance of the generation fleet to supply the necessary ancillary services and reserve levels associated with the level of wind integration. Extreme conditions are simulated for dynamic performance assessments. Operating reserves would be expected to increase and the specification of generation with ramping capability may change the types of generation in the generation forecast.

The LOLE calculations become an annual analysis as more wind data becomes available and the transmission power transfer capability is fed back into the subsequent JCSP( Mode 3) planning processes. Planning reserve margins would be expected to change in the early years of planning for Mode 3 transmission expansion associated with higher levels of wind energy integration to lower levels. Wind capacity credits may increase with wind generation geographic diversity. Both the lower planning reserve margins and the capacity credit increase would have positive financial affects on the value of the wind generation. Reliability issues that would need to be addressed probably would be focused on minimum load and minimum generation periods with high wind generation. If there is a successor to the JCSP, then the LOLE study should be run to provide inputs for the generation forecast and for the wind capacity credit determinations.

The ability to move ramping capability over the transmission system may be as important as the ability to meet peak load. Keeping units on line on minimum load that will be needed for the next day’s peak would be an evaluation that probably is critical to reliable operations. Mode 4 Mode 4 occurs when the export of wind energy from a high capacity factor area is justified based on economics. Merchant development is often thought to be the reason that exports would be justified. If there is wind generation that can access a high priced market and obtain sufficient revenue for the wind generation and the transmission associated with the delivery of wind energy, then the export of wind capacity and energy to the high priced region is justified economically. While the focus is on wind energy, merchant capabilities may be possible for other types of generation. As long as Mode 3 had mandated wind energy displacing low cost base load generation that would be placed into the Energy Market without having to recover the capital cost of the generation, merchant opportunities would be limited. The price of natural gas above about $9/MBtu could also trigger a merchant response from wind generation as it may be revenue sufficient displacing gas generation at the $9/MBtu level.

142


A second scenario is that if the transmission were available and low enough in cost, it may be more economical to obtain wind generation capacity and energy from a high capacity factor area. Studies by the Midwest ISO indicate that a 10-15% capacity factor difference would be sufficient to pay for the transmission associated with Mode 3. Once the large incremental step to establish a base for Mode 3 transmission (15,000 MW), the incremental cost for the next step of 5,000 MW, costs less than the initial transmission expansion. The open access transmission rules may provide a means of delivering wind from the Midwest to the east coast to satisfy wind mandates. Since the decision is not based on the relative economics of Mode 3 operation, but is based on the need to obtain renewable energy, Mode 3 transmission may enable wind generation capacity and energy exports. Planning transmission systems under Modes 2, 3 and 4 may be coincident. A combination of Mode 3 transmission based on economic expansion and Mode 4 based on meeting wind mandates more economically would forecast the opportunity remaining for merchant generation development. The amount of generation that could be used by the U. S. Eastern Interconnection in total then would have a forecast. Building transmission beyond the forecast would have to be supported by the generation developer a there would be no identified customers remaining. NERC Standards Applicability All Modes would be designed to meet NERC reliability standards. The NERC standards may not apply to the future operation with variable energy resources as well as is has with mainly capacity resources and require some modifications. Generation adequacy would no longer be a function of resource mix alone. The transmission system may change the capacity credit on wind generation. The power transfer capacity of transmission associated with the Energy Markets integrated with wind generation may change the planning reserve levels at peak conditions. The iterative nature of the study process and the uncertainty of being able to determine the next state of the operation of the power system may require probabilistic method rather than the deterministic methods used to test reliability presently. Peak transmission loading will probably occur off peak and minimum load levels may be limiting condition to power system operation, not peak load with respect to higher voltage transmission. Larger balancing areas probably seldom operate under an N-1 state. An N-4 to 6 may be considered to be normal with coordinated outage scheduling outages of critical lines assuring next day reliability. Planning criteria may have to reflect expected operating practices. Large scale wind penetration affects most of the Eastern Interconnection. Present NERC RGO’s may be too small to effectively study the impacts of large scale wind penetration without coordinated studies similar to the JCSP. Wind integration studies on an RGO level may not be applicable to an accurate assessment. The DOE NREL East Wind Integration Study, the DOE NREL West Renewable Integration Study , ERCOT Wind Integration Studies and the incorporation of Canada into the East and West Integration Studies with interface boundaries defined for Hydro Quebec may be a more reasonable geographic definition for wind integration studies with larger penetrations of wind. The Midwest ISO alone has SERC, MRO and RFC regions in its foot print and neighbors NPCC, WECC and SPP.

143


Reserves may require that a ramping capability be maintained to handle schedule variations up to multi-hour windows for periods when wind generation is forecast to decrease rapidly. The data to evaluate the geographic diversity impacts of large wind penetrations will not be able until late June 2008 for the Eastern Interconnection. The DOE NREL Eastern Wind Integration study will not be complete until mid 2009. A better answer should be available if the JCSP process and the DOE NREL Eastern Wind Integration study processes are repeated in late 2009 and early 2010 including all of the Eastern Interconnection. WECC is performing similar studies, but probably has a different schedule and process. ERCOT has its own processes. Hydro Quebec has its own processes. Both reliability and economics are considered in the planning forecasting or planning processes.

144


Appendix: Current Methods Utilized in U.S. Jurisdictions to Calculate Capacity Credit of Wind Plants Pennsylvania-New Jersey-Maryland Regional Transmission Organization PJM is an RTO that encompasses all or parts of Delaware, Illinois, Indiana, Kentucky, Maryland, Michigan, New Jersey, North Carolina, Ohio, Pennsylvania, Tennessee, Virginia, West Virginia and the District of Columbia. PJM includes over 56,000 miles of transmission lines and more than 1,200 generating units. PJM has almost 165,000 MW of capacity, and it serves about 145,000 MW of peak demand (PJM 2007). In general terms and on an annual basis, PJM requires LSEs to have a reserve margin of capacity above what is required to serve load. To meet that requirement, LSEs can self-supply capacity, enter into bilateral arrangements with generators for capacity or purchase capacity through a PJM-administered capacity market. In both PJM East and PJM West, the capacity market consists of daily, monthly, interval, and multi-month markets. PJM’s current required reserve margin is 15%. In 2007, PJM put in place a forward-capacity market called the Reliability Pricing Model. This market includes an annual Base Residual Auction allowing LSEs to acquire power three years in advance of the delivery year, as well as three optional incremental auctions prior to the delivery year. In these auctions LSEs meet their load obligations which are based on historical summer peaks and an additional 15% reserve margin.45 RPM is meant to send price signals that will encourage the development of new capacity resources.46 Generators are not required to submit bids into the RPM auction to operate in PJM, but they will not receive capacity revenues. Those that do bid into the RPM auction but produce less than bid will either have to make it up through the bilateral market, or pay the higher of PJM’s estimate of the cost of new entry or two times the RPM market clearing price.47 The capacity credit for wind in PJM is based on the wind generator’s capacity factor during the hours from 3 p.m. to 7 p.m., from June 1 through August 31. The capacity credit is a rolling 3-year average, with the most recent year’s data replacing the oldest year’s data. Because of insufficient wind generation data, PJM has applied a capacity credit “class average” of 20% for new wind projects, to be replaced by the wind generator’s capacity credit as noted earlier once the wind project is in operation for at least a year. As an example, a new wind generator will receive a capacity credit of 20% the first year; the average of 20% and the wind generator’s capacity factor during the hours from 3 p.m. to 7 p.m. from June 1 through August 31 in the second year; and the average of 20% and the wind generator’s capacity factor during the hours from 3 p.m. to 7 p.m. for June 1 through August 31 for years two and three, and so on. In addition, wind generators are also required to bid into PJM’s day-ahead energy market, along with other generators receiving capacity credit in PJM.

45 LSEs can opt to meet their required capacity obligation requirements through self-supply, subject to PJM approval and for a minimum period of five years. 46 PJM. "Reliability Pricing Model." www.pjm.com. (accessed 29 Jan 2008), http://www.pjm.com/services/courses/c-reliability-pricing-model.html 47 For example, the cost of new entry in New Jersey is $72,207/MW/year; $74,117 MW/year in Maryland and $73,866/MW/year in Illinois.

145

http://www.pjm.com/


In May 2008, PJM will replace the 20% capacity credit class average with 13%, based on the average capacity factor during the 3-7 p.m. hours from June through August for all wind generators that have been in operation for three years or more in PJM. The revised capacity credit will take effect for the 2011/12 period; the 20% class average will remain in effect until then. A higher project-specific capacity credit may be obtainable if the wind developer provide evidence that the wind turbine design and wind patterns justify the use of a higher capacity credit than the PJM class average for wind.48 PJM also set a minimum and maximum amount that wind generators can bid into PJM’s RPM auction, setting as a minimum of 85% of the capacity value of a wind project, and the maximum as the capacity value of either the individual wind generator (if more than three years of operational experience is available) or the capacity credit class average for wind at the time of the auction. Considering a 100 MW wind project, for example, then the maximum it can bid into the RPM auction is 13 MW (0.13*100), and the minimum is 11.05 MW (0.85*13). The 15% approximately represents the standard deviation from the mean of the annual capacity value of wind generators now operating in PJM. The minimum and maximum bid amounts for wind were implemented in order for wind generators to minimize the potential for being penalized for under-delivering, such as lower-than-expected wind resource patterns.49 New York ISO The New York ISO (NYISO) consists of the transmission assets of eight transmission owners located in New York and a small part of New Jersey. The NYISO includes about 43,771 MW of available capacity (including in-state and out-of-state capacity and demand response resources) and had a peak load of 32,169 MW in 2007.50 The NYISO also requires LSEs to have capacity reserves over their load requirements. The NYISO has a capacity market and obtains capacity through three auctions: a 6-month strip auction held twice a year, prior to the summer and winter capability periods; a series of monthly auctions; and a monthly spot auction for LSEs that have not met their reserve obligations. The summer capacity credit for existing wind projects is determined by a wind project’s capacity factor between 2 p.m. and 6 p.m. during June, July and August from the year before. The NYISO also offers a winter capacity credit and for wind, this is determined by the capacity factor of wind during the hours of 4 p.m. and 8 p.m. during December, January and February from the year before. New wind projects are assigned a summer capacity credit of 10% and a winter capacity credit of 30%. In addition, variable energy generators such as wind are exempt from having to bid into the day-ahead energy market in the NYISO, a requirement for other non-variable energy generators. ISO New England

48 PJM. PJM Manual 21: Rules and Procedures for Determination of Generating Capability. April 1, 2008. http://www.pjm.com/contributions/pjm-manuals/pdf/m21.pdf. (Accessed April 11, 2008). 49 PJM RPM Working Group. Offering Wind Resources in RPM Auctions. January 22, 2008. http://www.pjm.com/committees/working-groups/rpmwg/downloads/20080122-item-03-offering-wind-resources.pdf. (Accessed April 10, 2008). 50 New York ISO. Summer 2007 Electricity Review. October 2007. http://www.nyiso.com/public/webdocs/newsroom/press_releases/2007/summer_2007_electricity_review_102507.pdf. (Accessed April 10, 2008).

146

http://www.pjm.com/contributions/pjm-manuals/pdf/m21.pdf

http://www.pjm.com/committees/working-groups/rpmwg/downloads/20080122-item-03-offering-wind-resources.pdf

http://www.nyiso.com/public/webdocs/newsroom/press_releases/2007/summer_2007_electricity_review_102507.pdf


ISO New England operates in six states and includes more than 34,000 MW of capacity and serves about 27,000 MW of load.51 ISO New England is in the midst of changing from an installed capacity market to a forward capacity market auction, as described more below. Wind generators under 5 MW in capacity participate in the ISO New England energy market as “settlement only” resources, a category for generating resources under 5 MW. Settlement-only resources sell electricity into the grid at real time and receive the real time market clearing price. These resources are not assessed any operating charges for schedule deviations or imbalances and through May 2010, will receive a capacity credit equal to the unit’s capacity, multiplied by 1 minus its forced outage rate.

Wind generators over 5 MW would be classified as variable power resources and can schedule into the ISO New England’s day-ahead market. If variable power resources do not submit bids into the day-ahead market, then before the next operating day, these resources must self-scheduled the capacity amount for each hour. If in real time the capacity amount is different than the self-schedule amount, the variable power resource must contact the ISO and re-declare its schedule. As with settlement-only resources, variable power resources are not assessed operating charges for scheduled deviation or imbalances until May 2010, will receive a capacity credit equal to the unit’s capacity, multiplied by 1 minus its forced outage rate. As with PJM, ISO New England administers a forward capacity market, with an annual auction set three years before actual delivery is due. All demand and supply resources can participate in a descending clock auction to meet the ISO New England’s installed capacity requirement. The first auction was held in early 2008. Over 39,000 MW of new and existing generating resources competed for the projected 32,305 MW of load in the 2010 to 2011 timeframe.52 At least eight wind projects (five existing and three new) participated in the auction, bidding a collective range of 10 MW to 18 MW. Four of the wind projects bid zero for capacity.53 A second auction is scheduled for December 2008. New variable energy projects, including wind, that wish to participate in the forward capacity market auction can claim a summer and winter capacity credit but must provide supporting summer and winter wind speed data for wind; water flow data for run-of-the-river hydro; and irradiance data for solar facilities. Typically, this will range from 15% to 30%. Actual data will be used to adjust the capacity credit once the variable energy project is in operation. For determining the summer capacity credit for existing variable energy projects, ISO New England, for the first forward capacity market auction that occurred in 2007, took the average of the median net output of the variable renewable energy from 2:00 through 6:00 from June through September for the previous four years. The same terms apply for subsequent forward

51 ISO New England. 2007-2016 Forecast Report of Capacity, Energy, Loads, and Transmission. April 2007. Available at: http://www.iso-ne.com/trans/celt/report/2007/2007-celt_report.pdf (accessed January 29, 2008). 52 ISO New England. “Wholesale Marketplace Helping to Achieve Long-term Power System Reliability Goals.” February 13, 2008. http://www.iso-ne.com/nwsiss/pr/2008/press_release_fcm_auction_results_02_13_08.pdf. (Accessed April 10, 2008). 53 See http://www.iso-ne.com/markets/othrmkts_data/fcm/doc/fca_monthly_obligati on_including_zeros_final.xls.

147

http://www.iso-ne.com/trans/celt/report/2007/2007-celt_report.pdf

http://www.iso-ne.com/nwsiss/pr/2008/press_release_fcm_auction_results_02_13_08.pdf


capacity market auctions, except that ISO New England will use five years instead of four. Also, after June 2010, variable energy resources will also receive capacity credit during summer peak hours when ISO New England has declared a system-wide shortage. Furthermore, if the variable energy resource is in an import-constrained capacity zone, then capacity credit will be granted for all shortage events in that zone. ISO New England also provides a winter capacity credit for existing variable energy generators for the median output between 6:00 and 7:00 between October and May for the past four years for the 2007 forward capacity market auction, five years thereafter. As with the provisions for summer capacity credit, additional capacity credit may be available after June 2010 if ISO New England has declared a system-wide shortage event, and partial years will be used to determine a variable generator’s capacity credit.54 Southwest Power Pool The Southwest Power Pool (SPP) adopted a method to calculate wind capacity contribution (SPP GWG 2004). The SPP uses a monthly method that results in 12 capacity measures for the wind plant. The process first examines the highest 10% of load hours in the month. Wind generation from those hours is then ranked from high to low. The wind capacity value is selected from this ranking, and it is the value that is exceeded 85% of the time (the 85th percentile). Up to 10 years of data are used if available. For the wind plants studied in the SPP region, the capacity values are typically about 10% rated capacity.55 According to SPP’s Generation Working Group (SPP GWG) presentation, this method is used for long-term planning. Although it appears counter-intuitive to us, the SPP GWG believes that ELCC/LOLP methods are better used to determine the level of desired spinning or operating reserves and not to determine the reliability impacts of wind. Minnesota Department of Commerce/Xcel The Minnesota Department of Commerce (MN/DOC) study examined the impact of 1,500 MW of wind capacity distributed at various locations in southwest Minnesota. This represents approximately 15% wind penetration, based on the ratio of rated wind capacity to peak load. One of the tasks of this study was to calculate the capacity contribution of wind. The study used a Sequential Monte Carlo method, which performed repeated sampling of an annual state transition matrix that was calculated based on the wind data used in the study. The intent of this approach is to capture some of the impact of the inter-annual variation of wind so that estimates of ELCC may be more robust. The SMC cases found a 26.7% capacity contribution for the prospective wind plants. For comparison, the study also used a simple “load-modifier” method that calculates reliability based on a simple netting of the wind generation against hourly load. When this approach was used, the prospective wind capacity value was 32.9% of rated capacity. Another study, conducted by the EnerNex Corporation in November 2006 on behalf of the Minnesota Public Utilities Commission, analyzed the effect of 15%, 20% and 25% wind penetration levels on the Minnesota Electricity grid using wind data from 2003, 2004 and 2005.

54 ISO New England. Market Rule 1. April 3, 2008. http://www.iso-ne.com/regulatory/tariff/sect_3/index.html. (Accessed April 10, 2008). 55 See, for example, the discussion in Westar’s Energy Plan. Westar Energy. Meeting Our Customers’ Energy Needs: A Strategic Plan for Uncertain Times. Undated. http://www.westarenergy.com/corp_com/contentmgt.nsf/resources/CEP/$File/CEP.pdf?openelement. (Accessed April 10, 2008).

148

http://www.iso-ne.com/regulatory/tariff/sect_3/index.html

http://www.westarenergy.com/corp_com/contentmgt.nsf/resources/CEP/$File/CEP.pdf?openelement


Using 2004 as a base year, the study found that a 25% wind penetration level would only require an increase in reserve requirements from 5% to 7.05%. The study also estimated capacity values of wind generation using an ELCC/LOLP method, finding it to be between 5 and 20% of nameplate capacity depending on penetration level and wind year.56 PacifiCorp PacifiCorp recently completed a new Integrated Resource Plan.57 In the plan PacifiCorp used ELCC as the standard calculation of capacity contribution from wind generation for planning purposes. Wind generation was modeled using the same Sequential Monte Carlo approach used by Enernex in the MN DOC study. For the several prospective wind locations analyzed by PacifiCorp, the capacity contribution of wind averaged approximately 20% of rated capacity. The capacity value from the IRP is used as part of an evaluation to determine how much additional capacity is needed to meet future load forecasts. Electric Reliability Council of Texas (ERCOT) ERCOT evaluated the operating wind plants to determine the capacity contribution of wind. The analysis was based on wind generation on weekdays from 4:00 p.m. to 6:00 p.m. during July and August, the peak period for ERCOT. Using historical data and an adjustment for Equivalent Forced Outage Rate for a combustion turbine and a confidence factor, ERCOT calculated a factor of 2.9% of nameplate capacity. Beginning in 2005 this factor is used to include wind power in reserve margin calculations. The method of evaluation of this confidence factor is unclear from the document. ERCOT’s Generation Adequacy Task Force recently made a recommendation to alter the capacity contribution factor suggesting a number of new methodologies which would raise the capacity contribution factor from 5.3% to 16.3%.58 The document concludes that “the ELCC methodology should be used until better (i.e., more) actual performance data becomes available to make an accurate determination of the true capacity value of wind in ERCOT.” (page 5). Unfortunately, the ERCOT approach utilized data from a long-term NWP model that did not retain the synchronization between the load and wind. As described by the ERCOT documents, the wind data was not coincident with load data. In fact, the wind data was chosen based on a random selection of a 24-hour period. Although this represents an improvement over past methods, contemporary work on wind integration and capacity value has stressed the use of time-coincident wind and load data so that underlying weather patterns are picked up by the data set. Mid-Continent Area Power Pool (MAPP) The Mid-Continent Area Power Pool (MAPP) approach is a monthly method that calculates wind capacity value based on the timing of its delivery relative to peak. Up to 10 years of data

56 EnerNex Corp. 2006 Minnesota Wind Integration Study. November 30, 2006. http://www.enernex.com/staff/docs/windrpt_vol%201.pdf and http://www.enernex.com/staff/docs/windrpt_vol%202.pdf (accessed January 30, 2008). 57 PacifiCorp. 2007 Integrated Resource Plan.” May 2007. http://www.utahpower.net/File/File74765.pdf. (Accessed January 30, 2008). 58 Electric Reliability Council of Texas’ Generation Adequacy Task Force. Recommended Changes to the ERCOT Reserve Margin Calculation Methodology. March 7, 2007. http://www.ercot.com/meetings/tac/keydocs/2007/0330/11._Draft_GATF_Report_to_TAC_-_Revision_2.doc (accessed January 31, 2008).

149

http://www.enernex.com/staff/docs/windrpt_vol%201.pdf

http://www.enernex.com/staff/docs/windrpt_vol%202.pdf

http://www.utahpower.net/File/File74765.pdf

http://www.ercot.com/meetings/tac/keydocs/2007/0330/11._Draft_GATF_Report_to_TAC_-_Revision_2.doc


(wind and load) can be used if available. For each month, a 4-hour time window surrounding the monthly peak is selected. Any contiguous 4-hour period can be selected, as long as the peak hour falls within the window. The wind generation from that 4-hour period in all days of the month are then sorted, and the median value is calculated. The median value is wind’s capacity value for the month. If multiple years of data are available, the process is carried out on the multi-year data set. The results of these calculations are used in operational planning in the power pool. Portland General Electric (PGE) Portland General Electric (PGE) assumes a 35% capacity factor in their 2007 Integrated Resource Plan for the planned 325 MW Bigelow Canyon wind farm, and a 32% capacity factor for all other wind projects. PGE estimates the capacity contribution of wind generation to be between 5 and 15%. The IRP predicts almost 1,000 MW of wind power by 2015.59. Nebraska Public Power District The Nebraska Public Power District (NPPD) serves nearly a million Nebraska customers. In 2006, it completed construction on its 36 turbine, 60 MW Ainsworth wind farm. NPPD assigns a capacity credit of 17% of nameplate capacity to wind energy. Little information was provided on how this capacity credit was determined. NPPD’s draft 2008 IRP calls for completing negotiations of a power purchase agreement for up to 150 MW of wind by 2008 or 2009; build or contract for another 100 to 150 MW of wind power for 2014 to 2016; conduct a wind integration study; and perform a study on how much transmission is necessary to facilitate major development of new wind generation in Nebraska.60 Idaho Power According to its 2006 IRP, Idaho Power gives wind a 5% capacity credit, based on a 100-MW wind plant’s projected output that would occur 70% or more of the time between 4:00 p.m. and 8:00 p.m. during July, Idaho Power’s peak month.61. Therefore, Idaho Power’s method is similar to SPP’s by multiplying a subjective statistical number by actual capacity factor values. Puget Sound Energy (PSE) Puget Sound Energy just released its 2007 Integrated Resource Plan that includes a wind integration study as an appendix.62 Although not specified in the plan, a personal communication with a PSE representative determined that PSE’s determination of a capacity credit for wind is the lesser of 20% of nameplate capacity or 2/3 of the capacity factor of a wind project in January, which is PSE’s peak month. Pacific Northwest

59 Portland General Electric. 2007 Integrated Resource Plan. June 29, 2007. http://www.portlandgeneral.com/about_pge/current_issues/energy_strategy/pge_irp_2007.pdf. (Accessed January 30, 2008). 60 Nebraska Public Power District. 2008 Integrated Resource Plan (Draft). http://www.nppd.com/irp/additional_files/irp_draft.pdf. (Accessed April 10, 2008). 61 Idaho Power. 2006 Integrated Resource Plan. October 12, 2006. http://www.idahopower.com/pdfs/energycenter/irp/2006/2006_IRP.pdf. (Accessed January 30, 2008). 62 Puget Sound Energy. 2007 Integrated Resource Plan. May 2007. http://www.pse.com/energyEnvironment/energysupply/Pages/pse2007irpView.aspx. (Accessed January 30, 2008).

150

http://www.portlandgeneral.com/about_pge/current_issues/energy_strategy/pge_irp_2007.pdf

http://www.nppd.com/irp/additional_files/irp_draft.pdf

http://www.idahopower.com/pdfs/energycenter/irp/2006/2006_IRP.pdf

http://www.pse.com/energyEnvironment/energysupply/Pages/pse2007irpView.aspx


The Northwest Resource Adequacy Forum, an initiative of Bonneville Power Administration and the Northwest Power and Conservation Council, are working to create a consensus-based resource adequacy framework for the Pacific Northwest. The Forum assigned a proxy capacity credit of 15% for wind. However, a recommendation of the Northwest Wind Integration Action Plan is to reassess that proxy value, and the Forum is in the process of doing so. California The California Public Utilities Commission has a local resource adequacy requirement that requires load-serving entities under the CPUC’s jurisdiction to provide evidence that at least 90% of the capacity needed to meet demand is available, plus a planning reserve margin of 15% to 17%, on a year-ahead basis for the following May through September. The CPUC determines these capacity obligations annually and has just started a new proceeding for determining resource adequacy requirements for 2009.63 The monthly net qualifying capacity credit of wind is determined by the three-year average of monthly hourly production between noon and 6:00 p.m. on weekdays. Therefore, for June 2007, the monthly capacity credit would be determined by the average of monthly hourly wind generation for June 2004, June 2005 and June 2006.64 For wind projects with less than three years of operation, a “class average” of all wind generation within a transmission zone will be used, supplemented with project-specific data when available.65 A CPUC staff paper determined that the monthly net qualifying capacity value of wind in summer 2007 ranged from 20 to 60% of nameplate capacity in June to between 15% and 30% in July and August. There also was considerable variation between the different wind development areas in California, with Tehachapi in southern California generally having the highest ratio of net qualifying capacity to nameplate and Solano in northern California having the lowest, but with Solano producing the highest fraction of its net qualifying capacity during summer peak hours.66 PNM PNM provides power for 1.3 million customers, from its eight power plants, including the New Mexico Wind Energy Center, and from the wholesale market. Public Service of New Mexico examined wind’s contribution during its peak time of between 4:00 p.m. and 5:00 p.m. during July for the 200 MW New Mexico Wind Energy Center project. PNM determined that the wind plant was at 1-5 MW 16% of the time, followed by 0 MW, 6-10 MW, 11-15 MW and 16-20 MW, each close to 8% of the time. PNM determined that the most capacity wind can provide is

63 California Public Utilities Commission. Order Instituting Rulemaking to Consider Annual Revisions to Local Procurement Obligations and Refinements to the Resource Adequacy Program. Rulemaking 08-01-025, February 22, 2008. 64 California Public Utilities Commission. Order Instituting Rulemaking to Promote Policy and Program Coordination and Integration in Electric Utility Resource Planning. D.05-10-042, October 27, 2005. http://docs.cpuc.ca.gov/word_pdf/FINAL_DECISION/50731.pdf. (Accessed April 9, 2008). 65 California Public Utilities Commission. Order Instituting Rulemaking to Consider Refinements to and Further Development of the Commission’s Resource Adequacy Requirements Program. D.07-06-029, June 21, 2007. http://docs.cpuc.ca.gov/WORD_PDF/FINAL_DECISION/69513.PDF. (Accessed April 9, 2008). 66 California Public Utilities Commission Staff Report. 2007 Resource Adequacy Report. March 20, 2008.

151

http://docs.cpuc.ca.gov/word_pdf/FINAL_DECISION/50731.pdf

http://docs.cpuc.ca.gov/WORD_PDF/FINAL_DECISION/69513.PDF


152

less than 1 MW (16%*5 MW). Therefore, PNM concluded wind cannot be assigned a capacity credit at this time.67 Tri-State Generation and Transmission The Energy Policy Act of 1992 requires customers of the Western Area Power Administration such as Tri-State Generation and Transmission to file an integrated resource plan every five years. For its 2007 IRP, Tri-State compared monthly on-peak, off-peak and average capacity factors for wind projects in southeastern Colorado. Tri-State noted that the capacity factors for wind from May through August are lower than in other months, and determined that the capacity value of wind, as measured by wind’s contributed to Tri-State’s monthly coincidental peak, ranges from 2 to 12%.68 Colorado PUC/Xcel Energy Xcel Energy issued an ELCC study in 2007, as required via a settlement Xcel entered into with several parties. The company used hourly wind energy production profiles for 1996 through 2005 for several locations in eastern Colorado; historical loads from 1996 to 2005; forecasted load from 2008 through 2012; planned maintenance schedules and plant outage rates. Xcel modeled three scenarios of 280, 755 and 1,035 MW of wind, respectively. Unfortunately, the modeling software adjusted the 1996-2005 load data to meet projected monthly peak demand for 2008 through 2012. That, in turn, disconnected the load profiles from the wind profiles, affecting the final results and causing the ELCC values for wind to vary dramatically from scenario to scenario, and from year to year. Ultimately, Xcel Energy recommended adopting a capacity credit of 12.2% for wind.69 In its 2007 IRP plan, Xcel Energy used 12.5% capacity credit for wind.70 Rocky Mountain Area Transmission Study The Rocky Mountain Area Transmission Study (RMATS) is a multi-stakeholder, regional transmission study in the west. RMATs encompassed Colorado, Idaho, Montana, Utah, and Wyoming and was established by the governors of Wyoming and Utah to assess the feasibility of investing in new transmission to either access remote coal and wind resources or to export generation to other areas in the West (RMATS 2004). RMATS used 20% of rated capacity for all wind plants in the region. Although this is clearly a simplification and does not take account the significant differences between wind delivery profiles and the match to load, the wind capacity contribution is an important factor in determining the required capacity of other generation resources to meet loads during the study period. Because the RMATS modeling was based on local/regional load modeling and respected transmission constraints, it is likely that the wind capacity contribution across the RMATS region would vary, perhaps considerably.

67 Public Service Company of New Mexico. 2007 Electric Resource Plan. February 8, 2007. http://www.pnm.com/regulatory/pdf_electricity/irp_electric_resource_plan.pdf (accessed February 1, 2008). 68 Tri-State Generation and Transmission Association, Inc. Integrated Resource Plan. February 15, 2007. http://www.tristategt.org/NewsCenter/NewsItems/Tri-State%20IRP%2002-15-2007.pdf. (Accessed April 9, 2008). 69 Xcel Energy. An Effective Load Carrying Capability Study for Estimating the Capacity Value of Wind Generation Resources. March 1, 2007. http://www.xcelenergy.com/docs/PSCoELCCFinalReport030107.pdf. (Accessed April 11, 2008). 70 Xcel Energy. 2007 Colorado Resource Plan. November 15, 2007. http://www.xcelenergy.com/XLWEB/CDA/0,3080,1-1-1_41994_48216_48221-42116-0_0_0-0,00.html. (Accessed April 11, 2008).

http://www.pnm.com/regulatory/pdf_electricity/irp_electric_resource_plan.pdf

http://www.tristategt.org/NewsCenter/NewsItems/Tri-State%20IRP%2002-15-2007.pdf

http://www.xcelenergy.com/docs/PSCoELCCFinalReport030107.pdf

http://www.xcelenergy.com/XLWEB/CDA/0,3080,1-1-1_41994_48216_48221-42116-0_0_0-0,00.html

http://www.xcelenergy.com/XLWEB/CDA/0,3080,1-1-1_41994_48216_48221-42116-0_0_0-0,00.html

Operational Planning & System Operations

OOppeerraattiioonnaall PPllaannnniinngg && SSyysstteemm OOppeerraattiioonnss

Outline:

The operational issues created by variable generation result from the uncertainty created by the variable output and the characteristics of the generators themselves, such as the inertial response and dynamic response during fault conditions. The impacts are also affected by factors specific to the particular variable generation site, its interconnection to the power system, the characteristics of the conventional generators within the system being operated, and the rules and tools used by the particular system operator. The operational issues created by variable generation can be considered in terms of various time frames: seconds to minutes, minutes to hours, hours to day, day to week, and week to year and beyond. The operational experience of most utilities in the United States and Canada with variable generation has been limited to a small portion of the total generation within a power system or balancing area. Historically, most generation on the system could be scheduled and the output controlled by the system operator.71 Historically, for most utilities the most uncertain element in the power balancing was the load forecast; and the larger imbalances were created by contingency events. However, this situation is changing rapidly and in the next few years, variable generation and wind plants in particular are or will be significant producers of the energy portfolio in many North American regions. The integration of large amounts of variable generation in balancing areas and power systems poses new challenges to the power system operation. The operational practices to address them are at this time a work in progress, being developed and adjusted as more experience is gained.

71 Nuclear plants provide a significant exception though their output is not variable. Run-of-river hydro and increasingly fish-constrained hydro are also exceptions.

153


154

1. Operational Issues with Variable Generation


155

Figures from John Adams from their GE Study Report


1.1 Seconds to Minute Timeframe

Automatic Generation Control (AGC), Area Control Error (ACE), and Frequency Regulation: The changing output from variable generation creates an imbalance between production and demand. This imbalance will create area control error, which represents the error in the amount of actual interchange versus scheduled interchange (for an interconnected area) and error in the amount of frequency error, in terms of MW. The system governors and any frequency dependency of loads will be the immediate response to the imbalance (in between AGC cycles). The secondary response is by the control actions initiated by AGC to correct ACE. The AGC response occurs on a processing cycle of a few to several seconds. If variable generation is a small portion of the total amount of generation in the balancing area, the impact is not significant and can be managed by the same means in place for uncertainties in load forecast. When the amount of variable generation is significant, the variability can exceed the uncertainty and variability in the load and the variable generation can become the dominant factor in the area control error. The amount of variable generation required to impact frequency control and area control performance will be dependent upon several factors. For many regions and balancing areas, studies have been conducted to predict the impact significant amounts of variability would have for this time frame, but at present levels in most cases no change has yet been necessary to operational practices. It is anticipated that as the amount of variable generation increases, there will be a larger average ACE and a greater number of excursions into emergency control regions. It will become more challenging for the system operators to meet NERC’s Control Performance Standards. Although at present levels of wind and solar production most systems in the North America have been able to meet the Control Performance Standards, it has been demonstrated for autonomous grid systems with relatively large wind plants that frequency error increases with the addition of wind generation (ERCOT, EIRgrid, Hawaii Electric Light Co). It has also been found that a high percentage of variable generation results in a larger number of control actions and greater range of control for regulating generation under AGC control, which may eventually lead to increased operational costs. The degree of impact in this time frame is dependent upon several factors:

Degree to which the variable resources are correlated within the balancing area: Diversity will lessen the impact of the power fluctuations in the balancing area. Diversity does help smooth the variability in the seconds-to-minute time frame.

Flexibility of the regulating generation for the balancing area in responding to frequency error through droop and/or AGC control, and amount of available reserves: The impact will be reduced if the regulating/responsive generators have fast control capability with sufficient reserve (up and down) to manage the imbalances.

156


Frequency bias of the balancing area: A system with a smaller frequency bias will experience a greater amount of area control error (frequency error and/or interchange error) for a given power imbalance created by variable generation.



The mechanisms available to the operator to procure and manage regulating reserves in the short term: The term used for this function or category of reserve will vary for the specific market or system, however the variable generation will require greater use of the reserve generation which is responsive in the short term and available for frequency control. A flexible mechanism which allows the system operator to add or reduce the regulating reserves can adjust the amount of reserve available depending upon the actual conditions in real-time.

It is necessary to optimize the response of regulating units to AGC when there is a large amount of variable generation. Regulating generators also have to respond to a wider range of operation, and it may be necessary to modify the area control parameters to address the latency inherent to the AGC control. AGC operates on a cycle of a few to several seconds, and thus does not reflect changes in the variable generation (and ACE) that occurred during the AGC processing and control response. NERC Control Performance Standards measure imbalances of one minute and longer. Faster control is required on the interconnected system only to the extent that it helps better meet the one minute balancing requirement. A review of present balancing area) practices finds that most system operators in the North American regions have not established a firm policy of increasing or modifying the regulating reserves by a set amount to accommodate the present levels of variable generation, although studies have reviewed the need for reserves to increase with added wind and there is work in progress for several systems to address this in the future. Most studies and system operators agree that some additional reserves for frequency control will be necessary as the amounts of wind on the system are increased. There is less information available regarding the impact of other types of variable generation on frequency control. The mechanisms to procure additional reserves for frequency control are specific to the particular system or market and may affect ancillary service markets, regulating reserve requirements, and so on. At this time several systems permit the operator discretion to increase the reserve levels in order to meet the control performance criteria (or manage frequency for autonomous grids) as necessary when variable energy is on the system. Analysis of frequency data from systems with significant wind energy on autonomous systems have verified that the increased frequency error is linked to the magnitude of the

157


energy changes on a second-to-second time frame, rather than the total amount of variable energy being produced. This variability is more pronounced when wind plants operate in the mid-range of the turbine power curve. As a means of addressing the problem, some interconnection requirements or grid codes require the wind plants to mitigate sudden changes in output resulting from plant startup, shutdown, or wind changes. There is often an “out” clause for mitigation during wind reduction. System operators on autonomous grids have employed curtailment of the wind plant as a means of reducing the variability under cases of extreme volatility, in response to the difficulties it created in managing frequency and the stresses imposed on those units responding to correct frequency. There have been a few subsidized demonstration projects illustrating that storage devices can be used to mitigate the second to second power fluctuations, acting to smooth the power exported to the grid. It is best to use storage as a system balancing resource and to first aggregate wind variability with load variability. This reduces the total amount of storage required to reduce system variability. However at this time storage solutions are not in wide-spread use or consideration, due to cost. If the amount of wind production at night, coupled with the minimum generation limits of conventional generators, exceeds the ability of the system to absorb the energy, an excess energy condition exists. This is discussed further in a later section titled “Excess Energy”. Under such conditions, the regulating generators often operate closer to their technical minimums than historically the case. If the regulating generators become constrained towards minimums, it can be difficult for the system to respond to over-frequency and the balancing area performance drifts up. It may be necessary for the system operator to re-examine the criteria for maintaining sufficient regulating reserve down to manage credible contingency events involving loss of load, to ensure that regulating units will not be driven below their technical minimums.

Voltage/Reactive Power Control As wind plants become a larger amount of the total generation on a system, and conventional generation is displaced, it is now widely recognized that wind plants need to provide the voltage regulation and reactive power control capabilities presently met by the conventional generation. FERC and AWEA recognized this need and incorporated voltage-ride-through and reactive power support requirements for wind plants in the Large Generator Interconnection Agreement as part of FERC Order 661-A. Wind plants must provide dynamic and static reactive power support and voltage control if system interconnection studies show that it is required for power system reliability. In some cases, voltage regulation is required even if the plant is not producing real power; this is in order to avoid having to operate conventional generation solely for voltage support. The requirement for active voltage control is viewed by many systems as necessary to maximize the percentage of wind production on a system. Distribution-connected wind plants typically do not provide voltage regulation. The reactive capabilities of wind turbines vary by manufacturer and model, and in some cases a supplemental reactive device is required to meet the voltage regulation needs.

158


Some systems have found voltage stability constraints. ERCOT performs voltage stability studies in the day-ahead time frame and issues operating limits for a subset of the wind farms based on those studies. Contingency Reserves Contingency reserves provide very short-term reserves to manage unplanned outages and equipment failures. One consideration in evaluating contingency reserves is whether the contribution of variable generation is included in the available contingency reserves. There is no consistent policy. Variable generation resources are unique in that a sudden loss of generation can occur in the absence of equipment malfunction, due to loss of the original energy source. If the potential loss of variable generation is comparable to the existing defining contingency, there is no change required for contingency reserve to manage loss of the variable generation resource. If there are events that could result in a coincident near-instantaneous reduction from multiple variable generation sources, the aggregate loss could become the new defining contingency. Solar has the potential to ramp very quickly and could theoretically have a similar impact on the system as a generator trip. For wind plants, usually a sustained loss or increase in power production results in a ramp event which occurs over a longer time period. However, high wind-speed cut-out can result in a near instantaneous loss of power production from a wind plant which could require contingency reserve to cover. Geographic aggregation mitigates the speed of the wind generation reduction for plants with tens or hundreds of MW of output. Large amounts of wind generation are unable to simultaneously experience a high speed cutout event and instead experience large hour or longer ramp events. The ability to fore-warn the system operator of impending high wind speed cut-out is under development but does not seem to be available to the operators in the control centers today. At this time system operators in the North American regions have not had to acquire additional contingency reserves to manage the variable generation. The inability of wind plants to remain connected following a fault on the system could result in loss of multiple wind plants for a single incident, and thus be the cause of correlated outages. Rather than being addressed by contingency reserves, this is being addressed through low voltage ride through requirements, discussed in the System Security section.

System Stability and Dynamic Response Characteristics Wind and solar have different inertial and dynamic response characteristics than conventional generators. The impact depends on the operating policy to some extent. The impact increases when variable generation such as wind plants displace conventional generation completely, such that a certain amount of conventional generation that would

159


have been connected is now left online. The system inertia is reduced, a change which results in a higher rate of change of frequency and smaller frequency bias. Short circuit levels can also be affected. These issues are studied in the planning process, and may require modification of protection schemes, load shed schemes, and operator action levels. Some European and island grid codes are requiring an inertial response and frequency (droop) response from wind plants in order to permit the largest possible aggregate amount of wind, measures which can mitigate the effect of loss of inertial response. The impact of these potential system changes on system operations may include change in procedures and modification of frequency control parameters (through AGC) to reflect the change in system bias. In autonomous systems studies have defined the minimum amount of conventional plant that must be kept online to provide system stability. Depending upon the particular characteristics of the wind or solar facility, reactive power control issues can also arise. As mentioned above, some wind plants are designed to provide reactive control. However there have been some operational constraints identified at ERCOT which require operating limits for a subset of wind farms based on voltage stability studies. System Restoration During System Operations procedures, the system is in an abnormal operating condition. If the system has been islanded, frequency regulation is difficult and at times is managed through a different mechanism than under normal operating conditions. Under these emergency conditions, the additional uncertainty and imbalance created by variable generation would add to the challenges. At this time, wind plants and other variable generation do not participate in system restoration. Some systems are studying how wind plants could potentially contribute to system restoration in the future. Planners will have to examine the potential changes to existing restoration plans to consider the impact of large amounts of wind plants especially if a region of concentrated wind energy has the potential to be islanded.

1.2 Minute to Hour Timeframe

For this time frame, the system operator is responsible for monitoring the system for system problems such as thermal overloads, and taking necessary control actions; and ensuring the balance of power production with load demand to maintain interchange and system frequency. The system operator also maintains generation reserves in various time frames of response, to manage uncertainties and contingencies. Historically, with limited amounts of variable generation, the major uncertainty for system operators came from the load forecast. Over time tools have been developed for load forecasting and most system operators have a fairly accurate load forecast. Perhaps more importantly, system operators have a very good understanding of the uncertainty in this load forecast which is included in the planning for reserves. Operator corrective

160


actions were typically required for contingency events, such as loss of transmission or generation. Tools are in place to manage such events effectively and generally result in short-term excursions, with the system rapidly being returned to normal parameters. The output of variable generation introduces a new uncertainty factor for the system operator. The variable generation output affects the generation/load balance and can also create transmission system issues such as voltage problems or thermal overloads. The key challenge at present is the uncertainty in wind plant forecasts. When variable generation comprises a significant amount of the system’s energy production, the impact on the system operator from the uncertainty and forecast errors in the power production from those sources exceeds the uncertainty of load forecast. Perhaps the greatest challenge for the system operator in this time frame comes from sustained ramping up or down from wind plants, which can occur without advance notice to the system operator. The ramps can occur over a period of several minutes to hours. Another issue facing system operators on several systems is managing transmission constraints or transfer limits. In several systems the amount of possible wind production can create overloads of tie corridors or transmission systems. These issues could be mitigated by the system operator having accurate forecasts in the near term and hour ahead for wind, and incorporating these forecasts into dispatch decisions. Very accurate forecasts in the short term are not presently available, and there is some question as to whether intra-hour events can be captured. Accurate forecasting of wind energy for hour ahead and day ahead time frames is an area of active research and development. There is not a consistent handling of wind forecasts that are available in the either the planning or operations. System operators are developing the tools and working to better integrate the available information into dispatch decisions, reserve planning, identifying constraints, and operations plans.

Near-Term Reserves The term used for the reserves supplying near-term load balancing and frequency regulation is dependent upon the particular operational and market of the system operator. In addition, the time period of response covered by near-term reserves is specific to the system, and must consider the time period to obtain supplemental reserves. For example, at CAISO, the “Regulating Reserve” used for frequency regulation and balancing area performance, must be adequate to meet deviations within a five-minute interval. ERCOT utilizes “Responsive Reserve” to restore frequency within the first few minutes of a deviation from the standard frequency. AESO manages near-term imbalance with the Ancillary Services market. Historically, this type of reserve responsive within a minute primarily managed error in load forecast and variation in load demand, or provided contingency reserve. Variable generation creates additional uncertainty, which results in a greater use of reserves. Near-term reserves must manage the short-term variability as well as the

161


maximum potential ramping events than occur within the near-term time frame. The factors influencing the impact on short-term reserves are the same as those described in the section titled Automatic Generation Control (AGC), Area Control Error (ACE), and Frequency Regulation. For most systems, the amount of near-term reserves can be adjusted by the system operator in response to observed system conditions, including the impact from wind plants or other variable resources. There is not a general guideline in place to establish a set reserve increase in this time frame based on wind or other variable generation amounts. Load Balancing / Energy Balancing, and Sustained Ramp Events One of the most significant impacts of increasing amounts of wind plant has been the result of wind ramps. Wind ramps can occur in any hour of any day, and in any month. Wind ramps can occur due to loss or increase of wind, or as a result of the rapid shutdown of wind plants in response to very high speed wind conditions. The impact of sustained ramps on a system depends on several factors, many of which are also factors for sub-minute impacts:

Degree to which the variable resources are correlated within the balancing area. It has been shown that ramp events can affect wind production over a relatively large geographic area as a result of weather fronts.

Availability of wind production forecasting to the operator. If a system operator can obtain advance awareness that a wind ramp event may occur, additional reserves can be procured ahead of time (up or down).

Flexibility of the regulating generation for the balancing area in responding to AGC control, and amount of available reserves. The impact will be reduced if the regulating/responsive generators have fast ramping capability with sufficient reserve (up and down) to manage the imbalances.

Frequency bias of the balancing area. A system with a smaller frequency bias will experience a greater amount of area control error (frequency error and/or interchange error) for a given power imbalance created by variable generation.



The mechanisms available to the operator to procure additional generation reserves or commit additional generation in real-time. A flexible mechanism which allows the system operator to add generation/energy resources to the system gives the operator more options to adjust the balance of supply and demand to compensate the for an unforeseen and large change in wind production.

162


The availability of generation in the sub-hourly energy market: Systems with five, ten, or fifteen minute energy markets are better equipped to accommodate variable renewable generation with energy market response. Policies which increase the depth of the available generation stack in the sub-hourly market further reduce the cost of integrating variable renewable generation.

Availability of dispatchable or curtailable loads can increase the options available to the system operator to manage a ramp event.

Systems with existing wind installations have incurred an increase in ramp rate requirements. For systems with moderate levels of wind energy the existing capabilities have been sufficient to manage the increase. However, as wind energy increases, the ramping capabilities have at times exceeded the ability of the system to respond. This issue has become apparent first on autonomous systems with high wind penetration, such as ERCOT, Maui Electric Co., and Hawaii Electric Light Co. It is widely recognized that wind energy can become a larger percentage of total energy on a system with flexible plants. Good characteristics for conventional plants from the perspective of helping mitigate wind energy impacts include a wide operating range, good ramping ability, cycling capability, fast start-up capability, and short minimum down time. Generators that have these capabilities, such as simple cycle gas turbines and reciprocating engine plants, are not necessarily those that have historically had the lowest cost in comparison to less flexible plants (such as coal plants or combined cycle turbines). There is question as to how to encourage the construction of flexible plants in the future for market systems. The following mechanisms have been used by system operators today, in response to wind ramp events, to balance frequency: 1) Increase available regulating and near-term reserves to manage wind energy 2) Operate sub-hourly energy markets 3) Reallocate reserve to maximize available ramping capability 4) Provide visual indication, through a trending display, to the system operator in the

control room of the aggregate wind power output 5) Provide ramp-rate alarm indication to the system operator via the EMS system 6) Call upon offline (fast starting) generating reserves when ACE approaches action

levels 7) Impose ramp rate limitations upon wind plants as an interconnection requirement (For

Maui Electric Co and Hawaii Electric Light Co, an out-clause for downward ramping has resulted in upsets for down-ramp events but helped mitigate problems for up-ramp events)

8) Arrange for transfer from neighboring utility to provide balancing (where available) 9) Remotely curtail wind production (for up-ramp events) 10) Utilize curtailable loads to reduce demand during large ramp-down events 11) Under-frequency load-shed Accurate forecasting of wind production is a key to avoiding large imbalances caused by unforeseen large-scale changes in wind. At present, the forecasting tools available to the system operator are not adequate to avoid significant system upsets as a result of sustained ramps for large amounts of wind energy. For system operators with significant

163


existing wind, most control rooms provide an indication to the system operator of the total real-time amount of wind production, at least at the transmission level. However, accurate wind production forecasts are not yet readily available to the system operator. There is not a consistent mechanism to obtain forecasting, nor is there consensus across all areas as to who should provide it. At present, some system operators have no wind energy forecasting is available. Other systems require the wind plants to supply forecasts either directly or through the market schedule. As the result of issues encountered to date, and because it is easier to forecast aggregate wind performance than it is to forecast individual wind plant performance, many system operators have procured or are developing an independent forecast without relying upon the wind plants themselves. There are some system operators using forecasts from third-party forecasters and others developing the forecast in-house. Practical issues have come up that present a challenge to developing accurate forecasts. These include difficulty in obtaining wind turbine availability, difficulty obtaining accurate wind data, concerns over data confidentiality, and difficulty in developing an actual forecast strategy. It is becoming apparent that the analytical approach used with success in one region may not apply well to another. Another problem identified is failure to capture distributed wind plant output, as a separate item from load. In some systems wind production for plants connected on the distribution appeared to the operator as a load reduction, which further complicated forecasting efforts. Even for system which has wind energy forecasting available, the process are not yet in place to incorporate the dispatch into control room decisions. For example, at this time ERCOT does not include wind forecast into the reserves planning, but presently incorporates the market participant schedules. AESO includes wind in the short-term adequacy, up to 80 MW. For both of these systems, work is in progress to incorporate wind forecasts in the dispatch decisions and reserve calculations, as well as provide forecasting information to the system operators in the control room.

Transmission Constraints and Transfer Limits Often, very good resources for wind plants are located far from load centers. In several systems the installed capacity of wind plants exceeds the transmission capacity available to transfer the power. When an overload occurs in real-time, it is standard practice for the system operators to reduce the output of the variable generation through dispatch controls (also called curtailment) to eliminate the overload, which requires a commensurate increase in conventional generation to compensate. Operators have had varying experiences with dispatch control of wind plants. The ERCOT system operator has found that the response from wind plants is often a step function rather than a ramp down to the dispatch level. Hawaii Electric Light Co has had issues with failure of the remotely controlled curtailment system at one wind pant, which has required the system operator to call the wind plant operator to effect the curtailment. It appears technically feasible to implement a smooth ramp to the dispatch level, based on experiences in some

164


locations. This is apparently easier to implement for some turbines than for others and is dependent on the degree of power limiting that can be implemented before the particular turbine must be taken offline. If turbines can reduce power to only 50% before being taken offline, for example, a smooth curtailment is more difficult than if the turbines can be curtailed to very low percentages. For operational planning, wind power forecasts are being used to varying degree in the calculation of transmission constraints. In general, the wind forecast is not used for day-ahead planning but actual wind production is considered in the real-time contingency analysis. Excess Energy (Over-Generation Condition) In some systems, the output of variable generation resources such as wind is not well correlated to system demand. At times there is high production of wind energy during low-load conditions when fewer generators are online. For systems with a large amount of wind energy relative to the system demand, the system may not be able to absorb the amount of wind energy being produced during such conditions. This excess energy condition exists on island systems for up to eight hours of a day, depending on the production from variable resources (run of river hydro and wind). With the increase in wind plants expected for many other systems, it is expected it will be necessary to curtail production during lower load conditions for larger systems. An important factor relating to this issue is the flexibility of the existing conventional plant. In order to accommodate wind and avoid excess energy, it is desirable for plants to have low technical minimums and plants which can be taken offline. Some systems have found it desirable to ban the addition of new base load conventional power plants to avoid their aggravating the minimum load situation. However in some systems; existing plants must operate continuously and have relatively high technical minimums. It is also necessary for the system operator to keep a minimum number of conventional plants online to provide system stability, adequacy of supply, regulating reserves, and address system constraints. Autonomous systems with significant wind have found that at times, additional reserves must be maintained to manage wind ramping. In order to maintain this level of reserve, some plants must be kept online that might otherwise be taken offline. This leads to an increased level of minimum load. From the perspective of the wind energy producer, this is a drawback to using increased reserves as a strategy to manage wind energy changes. For systems experiencing excess wind production, the standard practice is for the system operator to dispatch the wind suppliers to curtail the output to the level the system supports. There is not a standardize rule used for the allocating or prioritizing the curtailment. It can be based on contractual agreements (as in the island systems of Maui Electric Co and Hawaii Electric Light Co) or by pricing in Market systems. Clear rules need to be established and it is important that the administration of the rules not become overly complex. It is important for the system operator to have guidelines for regulating

165


and near-term reserves down as the system will be operating constrained towards minimums during excess energy periods. Practical issues encountered for dispatch (curtailment) signals include: 1) Dispatch being implemented as a sudden change (step function) as opposed to

ramping to the desired level 2) Changes in wind production occurring after the curtailment occurs 3) Administrative burden to system operator to manage energy curtailments 4) Remote curtailment interface becoming inoperable so that the wind plant does not

respond to the dispatch signal 5) System operator uncertain when to invoke the curtailment. Delay in curtailment

resulting in over-frequency or large deviations in interconnection schedules). 6) System operator uncertain when to remove the curtailment, uncertain as to the level

wind output will be at once curtailment is lifted.

As the timeframe increases, the operational concerns associated with the integration of variable generation resources into the power system change. While the previous section focused mostly on frequency and voltage control, issues such as large ramp rates from wind, the management of energy being delivered from variable generation resources and the ability of available resources to perform system balancing become prominent in the time range of minutes to an hour. In response, System Operators and other Industry Members have begun to development and implement tools and strategies into their control rooms.

Issues:

A large power ramp from wind resources is often the first difficulty that comes to mind when considering how to integrate variable generation resources. Studies have shown that in many regions periods of high wind output are anti-correlated with periods of high demand. This in turn implies that wind ramps and load ramps occur at approximately the same time, but is opposing direction. This has the effect of increasing the strain on traditional generation to follow ramps in net load (system load – generation from wind units). Also, as the amount of energy being provided by variable generation resources replaces that of traditional units, the flexibility of available resources may not be sufficient to meet the system ramping needs. This is of course unless tools are implemented to protect from these situations occurring.

Large wind ramps can also have the consequence of shifting the timing of daily peaks in net load. For example, if a system’s evening ramp down occurs at 7:00 PM but there happens to be a large ramp up from wind resources at 5:00 PM, the demand that must be supported by traditional generation is reduced and may reach its lower peak at an earlier time than expected. As a result, generators may have to start-up or shutdown at times other than those at which they have historically done so.

Along with increased variable generation also comes increased uncertainty, or variability, in net load. Recent integration studies which have been performed for both Canada and the U.S. have dedicated a significant amount of time in investigating the degree to which net load variability is altered by increasing wind generation in the system. While the numbers do vary by region, it is

166


clear that significant increases in variability do occur and that these increases show up in the minute timeframe. This again means added burden on dispatchable generation to follow load and, assuming no changes are made, will result in decreased balancing area performance (CPS1 and CPS2).

Outside of the concerns of what will be required of traditional units, there is also the question of how to manage the energy being delivered by variable generation resources. Some assumption has to be made about where the output of variable generation resources will be until the system again dispatches in order to balance load. At low levels of variable generation penetration, it may be reasonable to estimate that over a certain time period the output level will remain relatively constant, but this again goes back to what was just said about variability.

The second concern with the management of MW coming from variable generation resources is the speed at which they react when asked to change their level of output, such as when being curtailed in order to relieve congestion. Variable generation resources tend to have the capability to ramp very quickly and may be able to ramp faster than is beneficial to the system. This problem may be relatively simple to resolve but still requires consideration when developing system protocols.

As discussed earlier, it is also very typical for periods of high wind output to occur during periods of low system demand. This only exacerbates the issues that have already been listed. Now the capabilities of traditional units still available for dispatch are even more in question both in their ability to follow load and to provide ancillary services.

Minimum generation in excess of minimum load conditions at night are also exacerbated by inflexible conventional generation technologies. New Zealand, for example, has placed a ten year moratorium on the addition of new base load generation to prevent this problem from getting worse.

Additionally, these issues can also be made better or worse depending on the strength of interconnection that a system has. Some areas may be able risk more uncertainty in their mix of generation if they are able to increase their dependency on their ability to import/export energy to/from connected systems.

Options:

In order to manage power systems as the amount of energy being delivered by variable generation resources increases, many control rooms have created operator tools and strategies or are in the midst of developing them. As a starting point, many groups are currently trending real-time output and other data for wind resources. It is being assumed that, at least in the time frame of 5 to 30 minutes, wind will continue to move in approximately the same direction at a similar rate. Traditional units can then be dispatched to follow load under this assumption. Others are also attempting to extract more information from these real-time plots, such as when the wind levels are high (or low) and likely to be steady, when the wind is ramping at a rate greater than some threshold, or when the wind speeds are a level close to the over-speed shutdown levels.

167


Along with this, many groups are also in the middle of implementing wind power forecasts into their operations. Current forecast designs vary in latency from 15 minutes to 1 hour, average outputs from wind resources and range between a few hours and multiple days as to the amount of time in which they look forward. Forecasts are also being created to predict output levels at a particular level of confidence. For example, for a 70% confidence level forecast, the output from a particular wind unit should be equal to or greater than the forecasted value 70% of the time. These forecasts are particularly useful when the economic or reliability consequences of over or under forecasting are not symmetric. Which forecast is more conservative depends on the current state of the power system. Over forecasting wind can result in the need to use expensive quick-start units when the wind fall short of expectations if other units have not been committed while under forecasting may have a relatively minor impact on the economic dispatch under some conditions. At other times over forecasting is a relatively inexpensive error if there are ample other generators already online while under forecasting results in high costs as conventional generators hit minimum load conditions. There has also been interest in developing event forecasting for wind farms or the entire wind fleet that predicts with some probability that a severe wind ramp (up or down) will occur in the upcoming hours.

Another topic of interest is the expanded use of demand-side resources to provide ancillary services. While markets already allow at least some amount of their ancillaries to be provided in this manner, it may become a more essential option as variable generation resources begin to displace traditional resources. This could also prove particularly useful during times of low system demand when the capabilities of resources available for dispatch may be limited, as previously mentioned.

1.3 Hour to Day Timeframe

As the time progress to the next hour and on to a day out, variable generation, specifically wind generation, adds many operational decisions to a control center operator. This section will look at the impacts of ramp events, forecasting and its uncertainty, system operator tools and strategies, and how these vary in systems of different strengths.

In this timeframe, a system operator will be concerned with unit commitment decisions, how flexible the on-line generation is, and what the wind forecast looks like. Tools that integrate variable generation into their outputs will be necessary to give the operator a view of possible scenarios and also allow input into operating decision points during the day.

One of the main issues to deal with during this time is the potential for large wind ramps and whether the generation on-line is capable of meeting the changes in variable generation. If the wind forecast predicts a upward ramp event, the operator will consider the time of day it is forecast for, what the load and interchange are forecast to do, and look at the downward ramp rates of the generation scheduled to be on-line during the event and decide if the upward ramp event can be handled. If the load is increasing during the forecasted upward ramp event, then additional flexible generation is not needed. If the upward ramp event is expected to occur during a time of decreasing load, enough flexible generation with its

168


increased ramp rates will need to be left on-line to accommodate the ramp event. If Operating Guides or Special Protection Systems (SPS) are in place to curtail variable generation during times of excess generation, whether it is low load period or during transmission congestion, then operating decisions can be made with these in mind. An upward ramp event can quickly change to a downward ramp event if high wind cutout speeds are reached. Procedures and tools need to be in place to quickly change between the two and the reliability ramifications of the change. Fortunately conditions that are likely to result in these extreme events are reasonably easy to predict.

Downward ramp events can be either due to a quick drop in wind or due to high wind cutout events. Geographical diversity will play a part in each event and how quick the ramp will be. For example, once the ERCOT wind fleet reaches 11,000 MW extreme ramps of 2800 MW in 30 minutes are expected about once per year. If wind generation is concentrated in one geographical area, the likelihood of a ramp event is increased. During the timeframe of one hour to a day downward ramp events can impact reliability and economics of the system. While the system operator is looking at the unit commitment for this time, a wind forecast that includes a downward ramp event or a potential for a high wind cutout event will need special attention. Other on-line generation will be needed with enough ramp capability to offset the variable generation decrease. Like an upward ramp event, if the downward event correlates with a load change, a unit commitment can be built that more easily accommodates this correlation. If the event is in the opposite direction of the load change, additional generation will need to be committed to handle both. The reliability aspects of a cutout event show that procedures and tools need to be in place for the system operator to use to recover from an event and to use prior to a predicted event. Economics surrounding a downward ramp event include forecasting when the ramp will occur and the uncertainty around it. Again, the need for the operator to integrate the events into their unit commit, congestion management, and ancillary services tools are necessary to maintain the reliability of the system without unduly increasing the economics.

Variable generation forecasting needs to be able to show the forecast output, the probability around it and the likelihood of a ramp event. For the hour to day timeframe, this forecast can be used during the intra-day generation commitment decisions and initially in the next day commitment. An operator with confidence in the wind forecast will make decisions to maximize the efficient reliability of the system they are operating. If confidence is minimal or the wind forecast is still new to the operator, their decisions will err on the side of reliability. Tools and procedures that show the forecast, its uncertainty, impacts on congestion, and possible curtailments needed in various regions will allow the operator to gain their confidence of the wind forecast and its impacts to the system.

In all of the above situations, the capability of the non variable generation on line is important for this timeframe. A continuous and ongoing look at the economic dispatch ranges, the ancillary services availability, and any emergency ranges available for the non variable on-line generation will allow the operator to build a reliable, yet efficient portfolio. The necessary flexibility of the non variable generation increases as the uncertainty of the wind forecast increases. A forecast with a high level of uncertainty during a time of increasing load and known congestion issues requires the most flexible generation available.

169


Mention has been made of operator tools and strategies needed during the hour to day timeframe. First and foremost is a wind forecast that looks into the next hours and the next day. Initially an operator needs access to this forecast to observe what the output of the variable generation is expected to be. This tool needs to show the forecast, its uncertainty, and actual real time variable generation output. Known generation output impacts like generation maintenance, forced outages of wind turbines or interconnections to the transmission or expected curtailments due to Operating Guides or Special Protection Systems need to be shown and the impacts of each to the forecast. The operator needs to be able to integrate the wind forecast into the unit commitment tool and choose between different scenarios. Scenarios built around uncertainty, ramp events (including cutout events), capabilities of the non variable generation all impact the final commitment decision. Procedures and Operating Guides fill in the gap when variable generation forecast and unit commitment decisions don’t match. Special Protection Systems are used when variable generation output, system configuration, and system capability meet conditions set up in the SPS.

Other tools needed for a system operator include congestion management tools for variable generation. Based on the wind forecast, the impacts of the variable generation on constraints can be displayed. If ramp events are predicted, the tool can show varying degrees of ramps and the effect of geographical diversity on the congestion. The ramp could increase or decrease a congested area and dispatching around the constraint based on the forecast can be displayed.

Operator strategies around variable generation factor in all of the above scenarios. Wind forecast, wind forecast uncertainty, potential for ramp events (including high wind speed cutout events), load trend, flexibility of non variable generation, scheduled interchange trend, congested areas, efficiency of starting additional generation, potential impact of transmission and generation contingencies, Operating Guides and Special Protection Systems, curtailing variable generation, strength of interconnected system wind is attached to, and/or how large of a Balancing Area wind is integrated into all are part of the strategy used in the operator’s decision making process.

1.4 Day Ahead and Longer

Most system operators do not incorporate wind and other variable generation forecasting in the longer-term planning activities such as Outage Scheduling, Security Analysis, and Available Transfer Capabilities. Though annual wind energy is more consistent and predictable than annual hydro energy the variability in when the wind will be available throughout the year is problematic. Several systems are studying what amount of wind capacity might be reasonable to include in longer time plans. Data is being collected to evaluate monthly trends which could be considered in the timing of maintenance outages for conventional generation resources. Most systems expecting a significant amount of wind energy within the next year to several years are looking to incorporate wind forecasts in near-term analysis for security constraints, transfer capabilities, and voltage constraints. System operators are in the processes of evaluating how

170


large amounts of wind plants affect existing emergency procedures and restoration plans, and how they may participate in the future, but at this time systems with significant wind do not expect participation of wind in restoration plans. For island systems, the uncertainties in wind have made fuel delivery schedules more difficult to plan accurately.

2. Differences between solar and wind variability

Both wind and solar power are variable resources and in that sense they are similar. Both have very low variable operating costs so failing to use the currently available energy stream wastes energy that can never be recovered. Neither can be dispatched above the currently available energy stream.

The seasonal and diurnal cycles of the solar resource are somewhat more predictable than wind. Solar power is never available at night and the daily pattern of maximum possible solar energy follows a predictable annual cycle. The output of the sun itself is quite constant, varying only 0.1% over the eleven year solar sunspot cycle. The rotation of the earth and the tilt of the earth’s axis, however, yield perfectly predictable daily and annual cycles of maximum possible solar energy. The exact amount of energy available depends upon the solar generation technology being employed with tracking solar systems providing flatter output than fixed arrays.

Actual solar generation is reduced below the maximum possible, and variability is increased beyond the daily cycle, by clouds and cloud movements.

Co-firing and Thermal Storage

Concentrating solar power systems provide two opportunities to reduce variability and to extend the hours of operation. Concentrating solar generators (sterling engine driven generators, solar troughs, and solar towers) use mirrors to focus sunlight, capture heat, and run a generator. These plants can be designed with thermal storage to continue producing electricity during cloudy periods or at night to produce electricity outside of the traditional solar daytime window. They can also be combined with natural gas to supplement the solar energy and eliminate or reduce the variability.

Aggregation

Both solar and wind power benefit from aggregation though wind probably benefits more. All solar plants experience night at nearly the same time because it is difficult to gain more than an hour of east-west diversity within the best solar resource areas. Both benefit from aggregation that reduces shorter term variability. Further, aggregating solar and wind provides diversity among technologies and reduces net variability.

Solar and Wind Variability

171


172

Wind variability tends to be a local phenomenon because wind loses coherence over distance. Figure 6 shows that the correlation between wind plants reduces dramatically with both distance and event speed. Wind plants that are only a few kilometers apart are completely uncorrelated at the five minute regulation level. Consequently relative short-term wind variability tends to reduce as more wind plants are added. Multi-hour wind ramps require greater geographic distance to achieve the same aggregation benefits.

0

0.2

0.4

0.6

0.8

1

0 100 200 300 400 500 600

Distance [km]

Co

rrel

ati

on

Co

effi

cie

nt

4h Average

12-hour

1h Average30min Average

5min Average

2h Average

Figure 6 Wind correlation reduces with distance and frequency.

Ronald E. Davis and Billy Quach. 2007. Intermittency Analysis Project: Appendix A: Intermittency Impacts of Wind and Solar Resources on Transmission Reliability. California Energy Commission, PIER Renewable Energy Technologies Program.

CEC‐500‐2007‐081‐APA.

With 17,000 MW of wind generation capacity currently operating in the US there is significant amount of historic data available. Wind variability has been characterized over various time


frames from seconds to years. Solar generation is not as commercially advanced and not as much operational data is available. Instead, data from individual installations must be used for analysis.

A study of Arizona solar data by Carnegie Mellon found a 80% correlation between Yuma, Prescott and Phoenix (300 KM). They concluded that while geographic diversity reduces the number of variations, it does not reduce the maximum ramp rates that are seen. (get citation)

Solar PV output can be subject to increased variability on a shorter time scale compared to wind. The mechanical inertia of spinning wind turbines tends to dampen out fast fluctuations in the wind. The thermal inertia of thermal solar plants also tends to dampen out fast fluctuations in solar resource. However, solar PV has no inertia and is very responsive. During high wind, partly cloudy climate conditions, solar PV output can be highly variable. The following 4 graphs show the output and the delta change from Tucson Electric Power’s Springerville Generating Station on an hourly, 15 minute, 1 minute and 10 second interval. (“Utility Solar Generation Valuation Methods”, Tucson Electric Power (TEP), 8, DOE Solar America Initiative project)

173


174


175

Load Correlation

The solar resource tends to be somewhat correlated with load on a seasonal and diurnal basis, with the exception of the evening load rise. Peak load days tend to occur on those days with high solar insolation and significant rainfall that reduces solar insolation to zero also tends to reduce load as well.

The correlation of wind generation with load is a local issue depending on the specific wind patterns. There is no overriding function that drives correlation of peak load with high generation as there is with solar. Diurnal wind patterns can favor wind generation at any time depending on local geography but frequently result in higher generation at night.

3. NERC Standards and Variable Generation Resources 3.1 Balancing Authority Performance Standards A principal requirement for Balancing Authorities is to support interconnection frequency and maintain Area Control Error (ACE) within acceptable limits. As we have explored in this section, these two objectives are accomplished over various time frames by controlling


generation automatically (via governors and AGC) or manually. To accomplish the control objectives, generation capacity (and sometimes load) must be reserved. NERC standards in the BAL series specify what performance is expected performance. Balancing Authority performance is measured in three ways: 1. Ability to support frequency in the interconnection (BAL-001, CPS-1) 2. Ability to maintain ACE within acceptable limits (BAL-001, CPS-2) 3. Ability to deploy reserves to respond to loss of generation within the balancing area (BAL-002, DCS) Variable generation also has limited ability to support frequency, and limited ability to control output on demand. At high penetration levels of variable generation, increased operating reserves are needed to comply with existing BAL standards. This is particularly true in small balancing authorities. New approaches to frequency control and ACE management can significantly impact costs associated with integration of wind generation. NERC has been working on a new set of BAL standards [Reference], which would replace CPS1, CPS2 and DCS with a new Control Performance Metric (CPM). The proposed Balancing Authority ACE Limit (BAAL) would allow for wider limits for ACE when interconnection frequency is near normal, and would tighten ACE limits during frequency excursions (see figure below). This concept should address both frequency support as well as ACE limits. Intuitively, the proposed CPM would reduce the need for deploying reserves to balance variable resources, considering that frequency stays within a narrow band around nominal most of the time.

3.1 Generator performance Standards In a high penetration scenario, there is a potential for diminished overall frequency support. Variable resources are technically capable of providing a limited amount of primary frequency support. Modern wind farms can reduce output when frequency is high (night-time, light-load,

176


minimum conventional generation situation). It is also technically possible for a wind farm to increase output when frequency is low; however, this would require the wind farm to “reserve” some of its capacity for this purpose. For example, the frequency response requirement depicted in the figure below is in effect in Ireland. It is debatable whether a similar standard is justifiable in North America at this time, but this concept should be one of the considerations when concerns about frequency support are brought up.

Large changes in wind farm output are also a reliability concern in some areas. In the case of wind generation, little can be done to control down ramps. However, it is technically possible to reduce up ramps (see Figure below). Currently, there are no standards in this area in the US.

Poutput

Pavailable

3.3 Resource Planning

There is little industry consensus with respect to treatment of variable generation resources in the context of resource planning and reserve sharing allocations. Different methods are used in North America to calculate capacity value and assign a capacity credit to wind generation. The results vary widely as well. The contribution of wind energy to system reliability can be adequately computed in terms of effective load carrying capability (ELCC). Generation capacity value is an important input to generation resource planning process; however, there is no consensus on how wind capacity value should be handled.

177


In general, the adequacy of existing and proposed standards with respect to high variable generation resource penetration should be studied. Guidance should come from the operating experience that the industry is accumulating.

4. Use of Contingency Reserves for Wind Events

When a power system experiences loss of supply, system frequency will begin to decline. If the decline in frequency drops below the dead-band in the governor controls, these governor controls will open valves or gates and increase the mechanical power to the turbine on those generators that have unloaded prime mover capability. Increasing the mechanical power to the power system will begin to arrest the decline in frequency. This behavior is a frequency responsive component often associated with the spinning reserve component of contingency reserves.

The role of the governor in a loss of supply event is to arrest the decline in frequency, but will not restore frequency. Restoration of frequency will be the act by system operators or generation under AGC control to further increase supply or decrease load and this is often the secondary role of contingency reserves.

Thus contingency reserves have 2 fundamental purposes for reliability.

The first is the frequency responsive component to arrest the drop in frequency as a result of a supply contingency.

The second is fast ramping component that will quickly restore frequency back to nominal.

To ensure that the power system is prepared for the “next” supply contingency, contingency reserves should be restored within 30 minutes and must be fully restored within 105 minutes of the loss of supply.

NREC and Regional Reliability Council rules require the BA or reserve sharing pool to carry sufficient contingency reserves to cover for their most severe single contingency, which in most cases often is a trip of their largest generator. The second consideration is the ability to restore contingency reserves within 105 minutes if deployed for a wind event.

Contingency Reserves and Wind Events

The NERC standard BAL-002 describes the purpose of the Disturbance Control Standard (DCS) is to ensure the Balancing Authority is able to utilize its Contingency Reserve to balance resources and demand and return Interconnection frequency within defined limits following a Reportable Disturbance. Because generator failures are far more common than significant losses of load and because Contingency Reserve activation does not typically apply to the loss of load, the application of DCS is limited to the loss of supply and does not apply to the loss of load.

178


A wind event that is in the same probability of a generator failure that can cause a balancing area’s “Area Control Error” to exceed the magnitude of a Reportable Disturbance should also have the same consideration to the use of Contingency Reserves to return interconnection frequency and preserve the reliability of the interconnected system.

5. Recommended measures to overcome challenges with variable generation.

1. Implement state-of-the-art wind forecasting services for all wind generators within a Balancing Authority to provide Day Ahead, Hour Ahead, and Real Time forecast. These forecasts will be crucial for unit commitment, scheduling, and dispatch processes in the Day Ahead, Hour Ahead and Real Time timeframes.

2. The Day and Hour Ahead wind generation forecasts (block energy schedules) should be

incorporated into the Balancing Authority scheduling processes. The Day and Hour Ahead schedules must be based on the forecasted wind generation values.

3. The Real Time wind generation forecast (average wind generation for five-minute

dispatch intervals) should be integrated with the Real Time unit commitment and Real Time dispatching applications.

4. All intermittent resources within a Balancing Authority should be requested to comply

with the interconnection rules established for renewable resources. These rules include installing meteorological towers and telemetry systems to communicate six-second or less meteorological and production data from wind parks to the Balancing Authority. This data needs to be integrated into the Balancing Authority forecasting software.

5. Develop ramp forecasting tools to help system operators anticipate large energy ramps,

both up and down, on the system. The longer the lead time for forecasting a large ramp, the more options the operators have to mitigate the impact of the ramp.

6. Balancing Authorities must work with the wind generator operators to ensure procedures,

protocols and communication facilities are in place so dispatch commands can be communicated to the wind plant operators.

7. During over generation periods, when dispatchable generation plants are already

operating at their minimum levels, the Balancing Authority needs to have an ability to curtail wind generation on an as needed basis to maintain reliability

8. Balancing Authorities need to consider developing a feed forward AGC system

incorporating Real Time wind energy and load forecasts.

9. Balancing Authorities need to evaluate the benefits of participating in a wider-area arrangement like ACE sharing or Wide Area Energy Management system.

179


180

10. Study the impact that additional cycling (additional start ups) and associated wear-and-tear issues; dispatches below the maximum unit capacity; and associated additional cost and environmental impacts will have on conventional generation due to the integration of large amounts of intermittent resources. Improvements may have to be made to Real-Time Dispatch and Regulation systems that will minimize the impacts on conventional units.

11. Encourage the development of new energy storage technology that facilitates the storage

of off peak wind generation energy for delivery during on peak periods.

12. Units with faster and more durable ramping capabilities may be required to meet additional ramp requirements.

13. Adequate short start units must be available to accommodate Hour Ahead forecasting

errors and intra-hour wind variations.

14. During off-peak hours when wind generation is at its highest, a Balancing Authority may have to commit non-intermittent resources to meet its Frequency Responsive Reserve (FRR) obligation. (The WECC is currently evaluating the FRR obligation for each Balancing Authority within the Western Interconnection).

15. All new wind resources must meet FERC LVRT requirements.

16. All new wind plants should be Type 3 or Type 4 generators that are capable of providing

dynamic reactive support to help the transmission grid meet applicable Interconnection transient stability performance standards and to prevent potential tripping due to low voltages. In the event that some of the new wind plants are of Type 1 or Type 2 with no dynamic reactive capability, the wind generator owner must provide sufficient reactive resources to meet the LVRT standards and voltage control standards.

Glossary

55.. OOtthheerr aanndd FFuuttuurree CCoonnssiiddeerraattiioonnss

Introduction: Chapter 5 will cover a number of topics which are important to understanding the planning and operation of wind power plants in the context of the operation of the larger interconnected power system with a high penetration of wind energy. As discussed in earlier chapters, the key differences between wind power plants and conventional power plants are that they exhibit greater variability and uncertainty in their output on all time scales. The power system currently exhibits variability and uncertainty with regards to the load in particular, and to a lesser extent with the generation. There are many techniques which have been developed over time to deal with these characteristics, which can also be extended to wind power plants. This chapter will begin by exploring a number of the current and future possibilities for managing the additional variability and uncertainty introduced by wind power plants, including the rapid developments in the field of wind plant output forecasting and its applications to managing the additional uncertainty introduced into the hour ahead and day ahead planning processes. The topic of wind plant design, and the rapid development in the field of application of power electronic controls to variable speed machines will be discussed. Such applications are leading to a new generation of machines with utility interface characteristics which are much more compatible with the needs for improved operation and system reliability than earlier generations of technology. This will be followed by a discussion of the critical need for generic wind turbine and wind plant dynamic models which can be used in the conventional transmission planning software for performing load flow, short circuit, and stability studies. At the present time, proprietary models for individual machines are limiting the ability of system planners to share wind plant models across multiple jurisdictions in order to carry our coordinated regional planning across broad geographical regions as required under FERC Order 890. A significant industry effort is already underway in this regard, and it needs to be strengthened. The final section of the chapter will deal with some of the considerations regarding the application of distributed generation on distribution systems which impact the reliability of the bulk power system. The primary consideration here is the current requirement contained in IEEE 1547 that generators disconnect from the system when the voltage and frequency fall outside of narrow bands intended to prevent islanding on the distribution system. These requirements are contrary to the behavior required to maintain bulk system reliability when there is a high share of distributed generation on the system. Information will also be presented on a system architecture being examined by Energinet.dk, the transmission system operator in Denmark, which already provides 20% of the country’s annual electricity supply from wind (25% in Western Denmark), and is studying how to accommodate 50% by 2020. This information will be presented as an illustration of the radical departure from conventional thinking that is occurring in an examination of how the power system of the future may need to evolve as more attention is focused on the planning and operation of the bulk power system in a carbon-constrained world.

181

Glossary

The chapter concludes with a summary of the conclusions and recommendations drawn from the issues explored in the chapter. The most significant item requiring attention in the near term is the lack of availability of suitable public domain wind plant dynamic models to be used in planning and studying the performance of the bulk power system.

Geographical diversity and aggregation of wind plants: As described in earlier chapters of this report, there are profound implications of the impact of geographic aggregation for wind, load, and wind and load together. These impacts are significant, and are an indication that barriers that inhibit the aggregation of load and wind will have potentially significant impacts on the ability of the Balancing Area operator to secure the needed ramping capability that is required to maintain CPS performance. Much of the material in this section is based on work by Kirby & Milligan.

Wind Variability and Forecasting

Economically and reliably dealing with wind’s variability and predictability requires a large, flexible power system. Physical size is important because the correlation between the power production from multiple wind plants diminishes as those plants are geographically farther apart (Error! Reference source not found.). In the graph below, correlation between wind plants is lowest (approaches 0) when distances between wind plants are large.

Larger geographic and electrical size also makes forecasting easier. Error! Reference source not found. shows that the wind forecasting error is reduced significantly when wind output from all four regions of Germany is compared with wind output from a single region. (Rohrig, 2005) According to Finley et. al, there is often a 30%-50% reduction in forecasting error that results

Figure 7. Wind generator variability loses corelation as the distance between machines increases and as the time frame of interest decreases (Ernst, 1999).

0

0.2

0 100 200 300 400 500 600

Distance [km]

1h Average30min Average

5min Average

0.4

0.6

0.8

1

Co

rrel

atio

n C

oef

fici

ent

4h Average

12-hour

2h Average

182

Glossary

from geographic dispersion of wind power, compared with the wind power being geographically concentrated. Thus, power system operators can more accurately predict and plan for changes in wind output when their systems are larger. Not surprisingly, forecasting accuracy also improves closer to real time. It is easier to forecast for short periods ahead, compared to longer periods in the future. Markets that operate closer to real-time take advantage of the improved forecasting accuracy by allowing more frequent generator schedule changes. Hour-ahead markets accommodate wind better than day-ahead markets. Sub-hourly markets have the least forecast error. A coordinated series of regularly clearing markets provides the best ability for conventional generation to adjust to changing wind conditions at least cost.

Table 1. Wind forecasting accuracy improves when larger geographic areas are considered.

NRMSE Forecasting Error %

Germany (all 4 control zones) ~1000 km

1 German Control Zone ~350 km

Day ahead 5.7 6.8 4 hours ahead 3.6 4.7 2 hours ahead 2.6 3.5

Balancing Area Size and Variability

Larger Balancing Areas have advantages even without wind or other variable generation sources. But with variable generation, the benefit of large BAs is even larger. This is verified by several studies, some of which are discussed below. These analyses are based on load and wind data from New York and Minnesota, and show that net variability declines per unit relative to BA size. Utilities have taken advantage of aggregation for decades. Since each balancing area only has to compensate for the variability in its aggregate load, and since random variations in individual loads partially cancel each other out, larger balancing areas require relatively less system balancing through “regulation” service than smaller balancing areas. The same principle applies to integrating wind: larger balancing areas are better able to integrate large amounts of wind because the random variability of individual wind generators and individual loads partially cancel each other out. This is based on the principle of statistical independence. If multiple remote wind plants are grouped and operated together within a single balancing area, their overall variability falls and it costs less to integrate their production into grid operations. Having a deep pool of flexible generation that can respond to variations in wind output helps system operators and reduces the cost of system balancing. Larger balancing areas have larger generation pools. Greater flexibility is a function of the generation mix, but larger pools always provide greater flexibility than smaller pools of the same generation mix. The ability to respond to system ramping needs is important to maintain reliability. As an example of the benefit of the larger balancing areas, Milligan and Kirby (2007) analyzed the consequences of balancing area consolidation in Minnesota, both with and without wind. Neighboring balancing areas will sometimes need to redispatch their generation in different

183

Glossary

directions at the same time. This happens when the load in one balancing area is increasing during a period when the load is decreasing in another balancing area. During such times, it would be beneficial for both systems to net their load ramping requirements, which would result in less ramping of generation in both balancing areas. Using hourly data, they calculated the ramping that could be eliminated if the four balancing areas in Minnesota were to combine. The graph for one full year of hourly load data is shown in Error! Reference source not found.. Opposite ramping does not occur in all hours, but it is apparent from the graph that 50 MW/hr or more can be reduced during much of the year, resulting in approximately a 14% reduction in ramping requirements (both up and down) annually if operations are combined. This reduction in load ramping requirements translates into lower cost of serving loads in all affected balancing areas. In Error! Reference source not found. below, the top graph panel shows the load ramping movements that would cancel out and need not be performed if the four Minnesota balancing areas were combined. Cancellation happens whenever one balancing area is ramping up while another is ramping down. Benefits are spread throughout the year, but can be seen to vary from hour to hour. The lower portion of Error! Reference source not found. reorganizes the same ramping information into a ramp-duration curve, which shows that, absent balancing area consolidation, there is as much as 75 MW of costly, unnecessary load-following generation in Minnesota attempting to compensate for the net variability of loads. This graph is based on loads only; there is no wind in the system portrayed by this graph.

-150

-100

-50

0

50

100

150

Exc

ess

Ra

mp

(M

W/h

r):

Ch

ron

olo

gic

al a

nd

Lo

ad

Du

ratio

n C

urv

e

80006000400020000

-150

-100

-50

0

50

100

150 Load Excess Ramp

Figure 8. Physical ramping requirements can be reduced by consolidating balancing areas (hourly load data).

184

Glossary

Combining balancing areas provides multiple benefits for loads, as seen in Error! Reference source not found.. Because wind is also subject to the principle of statistical independence, which says that uncorrelated or statistically independent random variables such as individual loads, individual wind turbine output, or multiple wind plant outputs do not add linearly, wind variability declines on a per unit basis when more wind is added to the system. An example of this benefit for a large wind penetration is shown in Error! Reference source not found., where the benefits of consolidated operations is more significant than portrayed for load alone in Error! Reference source not found.. What this figure shows is that excess ramping, which is unneeded and costly, is significant when balancing areas operate independently. Some balancing areas must ramp generation up at the same time that other balancing areas are ramping down. If operations could be coordinated, much of this ramping, and the associated costs, could be eliminated. The figure shows that the maximum unnecessary ramp is approximately 400 MW, and is matched by a -400 MW ramp. This bi-directional ramp requirement could be eliminated if the balancing areas would combine operations. Milligan & Kirby did a similar analysis using the 5-minute load and wind data from the Minnesota 20% Wind Integration study. Error! Reference source not found. shows a duration curve of the ramping that can be eliminated on a 5-minute basis if the four BAs were to combine, either physically or virtually. The inset shows a blow-up of the left side of the lower curve. Combining operations saves a modest amount (approximately 400 MW) of ramping over the 9-month period covered by this data set. But it is also clear that large infrequent ramp events are significantly damped by aggregation. These are the ramp events that are most likely to compromise reliability.

185

Glossary

186

Combined operations can offer a significant improvement on tail events: large but infrequent ramps

-600

-400

-200

0

200

400

600

Exc

ess

Ram

p C

ompa

riso

n: S

epar

ate

vs.

Com

bine

d O

pera

tion

s

10864205-Minute Periods

Load and Wind Load Only

This graph shows the difference in excess rampingrequirements between wind and no-wind cases.

-600

-400

-200

0

200

400

600

Ex

ce

ss

Ra

mp

Co

mp

ari

so

n:

Se

pa

rate

vs

. C

om

bin

ed

Op

era

tio

ns

100080060040020005-M inute Periods

Load and W ind Load O nly

Figure 9. Duration curve of ramp reduction (both up and down) on 5-minute wind and load data

Most of the discussion above focuses on the physical improvement on ramping needs and the cost impact of coordinated operations. There is an important link from this discussion to reliability. Because wind variability will increase the need for ramping within a balancing area, at high wind penetrations some balancing areas might find it challenging to meet the higher ramp requirements. This already appears to be an issue for Northwestern Energy in Montana, which is a small BA with a significant wind penetration. There has been an increase in CPS violations because of the confluence of several factors: small BA, significant wind penetration, and lack of flexible resources. Because adjacent BAs can pool their ramping requirements by cooperative operational arrangements, a reduction in CPS violations can be attained, resulting in greater reliability of the system. Both the reliability and the economic benefits of this BA consolidation can be realized through the operation of an ancillary services market across a broad geographical region. Balancing areas can be consolidated either physically or virtually. Physically combining balancing areas is straightforward, but may not always be desirable. Two or more balancing areas can retain their autonomy and still capture much of the aggregation benefit by electronically combining their Area Control Errors (ACE). Each can control to an allocated portion of the combined ACE, assuring that reliability is met at lower cost.

Glossary

187

-400

-200

0

200

400E

xces

s R

amp

(MW

/hr)

: C

hron

olog

ical

and

Loa

d D

urat

ion

Cur

ve

80006000400020000

-400

-200

0

200

400

Load and Wind Excess Ramp

Figure 10. Combining balancing areas can reduce ramping requirements for systems that have significant wind and load.

These results are corroborated by the New York State wind integration study and follow-on work by Miller and Jordon which found that combined operation of the eleven zones in the New York State power system reduces hourly load variability by 5% and five-minute load variability by 55%. In the NY study, hourly load variability shows the smallest reduction in variability (5%) when state-wide operations are compared with zonal operations, because loads are highly correlated on an hourly basis. Most loads increase in the morning and decrease in the evening. State-wide wind does not show that same similar pattern. Hourly wind variability is reduced by 33% and five-minute wind variability is reduced by 53% with state-wide operations. Hourly system variability is further reduced by 10% and five-minute system variability is reduced by 15% when wind and load are considered together. Note that while operating large balancing areas helps reduce the cost of wind integration, it also helps reduce the cost of serving load with or without wind (as pointed out in Error! Reference source not found. for the no-wind case). The benefits of large electricity balancing areas apply to systems around the world. In a recent report for the International Energy Agency (Holttinen et al, 2007), the authors conclude “Larger balancing area size and wind aggregation: both load and generation benefit from the statistics of large numbers as they are aggregated over larger geographical areas. Larger balancing areas make wind plant aggregation possible. The forecasting accuracy improves as the geographic scope of the forecast increases; due to the decrease in correlation of wind plant output with distance, the variability of the output decreases as more plants are aggregated. On a shorter time scale, this translates into a reduction in reserve requirements; on a longer time scale, it

Glossary

produces some smoothing effect on the capacity value. Larger balancing areas also give access to more balancing units.” (page 107).

Market Size and Flexibility

The flexibility and size of the market can have a significant impact on the ability to access faster ramping capability. Market mechanisms can increase the ability of the system operator to access the ramping or other maneuverability that may be physically present. Obtaining access to critical resources will help to maintain or improve system reliability. This is shown by Milligan & Kirby. Larger markets provide a deeper, more accessible stack of ramping capability that is critical to maintaining system reliability. In addition to market size, the flexibility of the market can also help maintain reliability. Markets help economically and reliably integrate wind both in how they treat wind generators and in how they treat conventional generators. Markets that allow variable resources to sell excess energy or purchase shortages at transparent and fair prices accommodate the natural characteristics of wind while reflecting the true real-time cost of maintaining reliability. More generally, generation scheduling rules and the energy market structure itself are the most important factors in tapping the physical flexibility of the conventional generation fleet. Sub-hourly energy markets provide economic signals that make it profitable for conventional generators to respond to fluctuations in load and wind. Scheduling rules that restrict generators to hourly movements artificially hobble the conventional generation fleet, resulting in lost opportunities for those generators and increased costs for all. Markets that encourage conventional generation movement when it helps increase reliability, and do not restrict generators to only changing output at the top of each hour, reduce costs. Markets can provide direct economic incentives for generation flexibility if the current fleet does not have enough. Some regions in the U.S. have fast energy markets, which result in a new economic dispatch every 5 to 15 minutes, depending on the market. The fast energy markets make it possible to hold the regulating units closer to their preferred operating point because they can be brought back to the mid-point of their operating range much faster than if the redispatch did not occur for an hour. Therefore, there is less need for regulation in faster energy markets. This results in a significant reduction in costs because regulation is typically the most expensive ancillary service. Thus, when calculating wind integration costs, such features that reduce balancing costs generally will lead to lower wind integration costs. Enhancing the flexibility of the conventional generation fleet helps to accommodate wind. This can involve valuing physical flexibility as other generators are built: fast start generators, lower minimum load capability, and high ramp rates are all valuable. Markets can provide signals about what characteristics are valued both to existing generators and to entities that are planning new generation. Participation in fast energy markets also encourages generators to move to the operating level that is consistent with their energy bids. This implies that load following services, which have no

188

Glossary

energy component, can often be extracted at no cost from the energy market. This can have important consequences for wind integration, and is discussed further in that context below. It is possible for a power system to have insufficient ramping capability with inappropriate consequences for energy prices if maneuverability is not directly valued. Error! Reference source not found. presents a simplified example where a fast energy market, which normally provides load following as a byproduct, may have difficulty providing ramp capacity under some conditions. Prior to 8:00 a.m., the example system is serving a 2,550 MW load with over 3,000 MW of baseload generation, and therefore clearing all energy at $10/MWh. At 8:00 a.m., a 300 MW ramp starts which the baseload generation can not follow. There is ample baseload capacity; it simply cannot ramp fast enough. Peaking generation (the only other generation in this example system) is started to meet the ramp needs. The peaking generator stays on until baseload generation can ramp up. With no explicit ramping service, the price rises for the entire energy market (all 2,850 MW) from $10/MWh to $90/MWh for 5 hours, just to follow a 30-minute 300-MW ramp. In this case, it might be better to create a separate ramping or load following service and pay the peaking generator for its response, rather than distorting the price of the entire energy market. It is very important to determine if ramping requirements can be served at a low cost as a byproduct of the sub-hourly energy market (the typical condition), or if ramping requirements impose a high cost because dedicated resources must be used. Fortunately market monitors can detect this condition and recommend the establishment of a ramping service if the condition is persistent. (Milligan & Kirby, 2007)

3200

3000

2800

2600

2400

4:00 AM 6:00 AM 8:00 AM 10:00 AM 12:00 PM 2:00 PM 4:00 PM 6:00 PM

dat

Base Load - $10/MWh

Peaking -$90/MWh

Energy Price $10/MWhEnergy Price $10/MWhEnergy Price Increases

to $90/MWh becausebase unit can't rampfast enough

Figure 11. In this simple example, load following is required from an expensive peaking generator, but energy is only an incidental product.

189

Glossary

In large parts of the western part of the U.S., there are numerous small BAs, and energy markets are not robust. However, there is significant interest in developing cooperative agreements among balancing areas that would provide some of the benefits of consolidation. The Northern Tier Transmission Group (NTTG) (www.nttg.biz) developed an ACE Diversity Interchange (ADI) pilot program that allows for the sharing of regulation across regions. There are several ways to allocate the combined ACE among the participating balancing areas, but in all cases the required control actions are smaller than if the ACE signals were not combined. There is significant interest in this project, and WestConnect (www.westconnect.com) has joined the NTTG ADI project, and continues to investigate wholesale market enhancements and seams issues in the footprint. The National Renewable Energy Laboratory (NREL) has begun a large Western Wind and Solar Integration Study. The focus of the study is the WestConnect footprint, but the entire U.S. portion of the Western Electricity Coordinating Council (WECC) will be modeled, and high wind and solar penetrations will be analyzed. Error! Reference source not found. shows the study footprint. One of the scenarios will consider the benefit of consolidated balancing area operations and examine the potential benefit for integrating a high penetration of renewable energy sources. In the Pacific Northwest, the Bonneville Power Administration convened stakeholders and utilities to examine how the region could best position itself to integrate up to 6,000 MW of wind that may be developed in the next several years. The result of this effort is the Northwest Wind Integration Action Plan (http://www.nwcouncil.org/energy/Wind/library/2007-1.pdf). Among the items on the agenda for the Northwest Wind Integration Action Plan are “(1) developing more cooperation between regional utilities to spread the variability of wind more broadly; (2) developing markets that will reward entities who choose to market their surplus flexibility.” Other parts of the report indicate a need for “developing more robust markets for control area services that will provide needed electric services for smaller control areas with substantial wind resources” (page 13).

190

http://www.nttg.biz/

http://www.westconnect.com)/

http://www.nwcouncil.org/energy/Wind/library/2007-1.pdf

Glossary

191

Figure 12. NREL's Western Wind and Solar Integration Study focuses on the WestConnect footprint and models the U.S. portion of the WECC footprint.

Although the outcomes of these various initiatives cannot be precisely predicted, they are further indication that when analysts consider how to integrate wind, market structure and design changes can offer significant benefits. The combination of regions in the Northwest and in WestConnect covers nearly the entire West that is not currently part of the California ISO or the Southwest Power Pool (parts of eastern New Mexico).

Transmission expansion- a tool for aggregation and variability reduction

An article by Christina Archer and Mark Jacobson of Stanford University often quoted in the popular press, entitled “Supplying Baseload Power and Reducing Transmission Requirements by Interconnecting Wind Farms” published in the Journal of Applied Meteorology and Climatology, Nov 2007 concludes that with proper blending of geographically diverse wind plants, the average capacity factor from the wind generation could be 33%. What is often overlooked is that in order to achieve this vision in actual operation, it is necessary to have sufficient transmission to aggregate the total wind plant output plus load, and provide the ancillary services necessary to balance the net load and system generation. In a pioneering application of this technique to a broad geographical region, the DOE-National Renewable Energy Laboratory wind plant output data base for the Joint Coordinated System Plan (JCSP) footprint that includes the New England ISO, the NYISO, PJM., MISO, SPP, Entergy, MAPP and TVA will provide hourly wind plant output time synchronized data for the years 2004, 2005, and 2006 by the end of 2008. The DOE-NREL East Wind Integration and Transmission Study (EWITS) will use this data to define the ancillary services and reserves

Glossary

necessary to serve the JCSP area under a 20% wind penetration scenario. The JCSP will define transmission that is needed to supply the JCSP ancillary service requirements, as well as the transmission to obtain benefits from the Energy Markets. Sufficient transmission capacity has to be available to blend the output variations of individual wind plants into the summation of the output of a very large aggregated wind power plant across a broad geographical region to enable smooth energy delivery to the power system. Wind integration studies such as the New York, Minnesota, California and Texas studies show that there are significant economic benefits from being able to obtain services from a large market with Ancillary Service offerings. Transmission is the critical enabling technology necessary to deliver ancillary services for regulation and load following from a large base of generation to a high wind energy producing area across a broad geographical region.

Dealing with variability through demand response: Demand response is an underutilized reliability asset which can operate in every time frame of interest to wind integration: from seasons to seconds. Different loads have different response capabilities, and different costs to respond. Demand response programs can be designed to utilize various loads to achieve various response objectives. Demand response programs exist in many regions; these can be expanded to assist wind integration. It may also be attractive (to the responsive loads, the power system, and wind generators) to establish demand response programs in regions where they do not exist or where different response characteristics are needed than current programs offer. Broadly speaking, there are five basic types of demand response as shown in Error! Reference source not found.. (ORNL/TM-2006/565: Demand Response For Power System Reliability: FAQ) Energy efficiency programs do not provide controlled response so are of limited use in accommodating variable wind. Peak shaving and price response programs typically obtain multi-hour response either through power system operator command (peak shaving) or through customer response to changing power prices (price response). Customers receive a few hours to a day’s advance notice. The response duration and/or the number of times response can be called upon are often limited. Conceptually these types of programs could be modified to respond to variable wind. The multi-hour response duration corresponds with the multi-hour nature of large wind events. The types of loads which respond might change since response would not be focused exclusively on times of system peak load. Advanced notification times might change as well to better accommodate wind variability. Reliability response and regulation are relatively new types of demand response. Both of these provide fast response with little notification though the response duration is typically relatively short. Loads stand ready to provide spinning or non-spinning reserve for reliability response. When a generator or transmission line trips and the system operator calls for response, the loads immediately reduce consumption, fully responding typically well within ten minutes. Regulation from loads is newer still. Here loads adjust consumption minute-to-minute in response to system operator automatic-generation-control signals (AGC).

192

Glossary • Energy Efficiency programs reduce electricity consumption and usually reduce peak demand

• Price Response programs move consumption from day to night (real time pricing or time of use)

• Peak Shaving programs require more response during peak hours and focus on reducing peaks every high-load day

• Reliability Response (contingency response) requires the fastest, shortest duration response. Response is only required during power system “events” – this is new and slowly developing

• Regulation Response continuously follows the power system’s minute-to-minute commands to balance the aggregate system – this is very new and may have the potential to dramatically change production costs, especially for aluminum and chlor-alkali

193

Figure 13 Load can provide five basic types of response that may be useful for wind integration.

Price responsive load markets and demand response

A robust Economic Load Response program enables Demand Resources to voluntarily respond to prices, by reducing consumption and receiving a payment for the reduction. This can be initiated by the customer (by monitoring real time pricing signals) or by the Control Area (requesting load relief). Some areas of the country offer this opportunity in a day-ahead market. The Day-Ahead alternative provides a mechanism by which any qualified market participant may offer Demand Resources the opportunity to reduce the load they draw from the system in advance of real-time operations and receive payments for the reductions based on day-ahead prices. Typically, these programs are for long periods of time (hours) and would be primarily of use in responding to forecast changes in output from variable resources (see 5.3.3 Responsive Load Characteristics below).

rket. The Day-Ahead alternative provides a mechanism by which any qualified market participant may offer Demand Resources the opportunity to reduce the load they draw from the system in advance of real-time operations and receive payments for the reductions based on day-ahead prices. Typically, these programs are for long periods of time (hours) and would be primarily of use in responding to forecast changes in output from variable resources (see 5.3.3 Responsive Load Characteristics below).

Demand response as a wind-specific ancillary service Demand response as a wind-specific ancillary service

In order to be of greatest use in dealing with variable generation, demand response must be able to be utilized rapidly and repeatedly, responding to generation variations with little warning. Variability in the minute-to-minute time frame requires regulation response. Variability in the hourly and longer time frame requires energy market response. Variability in the sub-hourly time frame may be addressed with sub-hourly energy markets in regions where they exist or may require a distinct ramping response if sub-hourly energy markets do not exist. A distinct ramping service may also be required if the ramping capability of the energy supply stack is exhausted.

In order to be of greatest use in dealing with variable generation, demand response must be able to be utilized rapidly and repeatedly, responding to generation variations with little warning. Variability in the minute-to-minute time frame requires regulation response. Variability in the hourly and longer time frame requires energy market response. Variability in the sub-hourly time frame may be addressed with sub-hourly energy markets in regions where they exist or may require a distinct ramping response if sub-hourly energy markets do not exist. A distinct ramping service may also be required if the ramping capability of the energy supply stack is exhausted. Variable generation and variable load are typically not perfectly correlated. That is, individual variable generators may be moving up or down when other individual loads and generators are separately moving up or down. Individual random generator and load variations partly cancel out and the total system variability is typically significantly less than the sum of the individual loads and generators’ variability. Consequently it is almost always significantly more efficient to compensate for total system variability rather than compensating for the variability of individual

Variable generation and variable load are typically not perfectly correlated. That is, individual variable generators may be moving up or down when other individual loads and generators are separately moving up or down. Individual random generator and load variations partly cancel out and the total system variability is typically significantly less than the sum of the individual loads and generators’ variability. Consequently it is almost always significantly more efficient to compensate for total system variability rather than compensating for the variability of individual

Glossary

generators or loads. This is true on every time scale but the effect tends to be greater on the shorter time scales (minute-to-minute and sub-hourly) because individual load and generator variability is least correlated at short time intervals. Responsive loads can effectively provide response for total system variability in all of these time frames.

Responsive Load characteristics

Characterizing both the need for load response and the capabilities of various loads to respond is necessary if load response programs are to be designed to facilitate wind integration. Specific residential, commercial, and industrial loads will likely all be attractive if they are understood fully. A discussion of the full range of available loads is beyond the scope of this paper. We can only mention the characteristics of concern and provide a few examples. A common characteristic of all responsive loads is that the load is discretionary in some sense. Some loads are discretionary on price. The same necessary function can be performed with another fuel source or another facility if the price of electricity changes. Other loads are discretionary in time. Electricity consumption can be delayed (or advanced) a few minutes or even a few hours with little or no customer impact.

How much load is available to curtail? The availability of some responsive loads varies with time. Air conditioning, for example, can provide excellent load response but it is only available when temperatures are high. This is the same time when overall power system load is high and other generation resources are needed to meet load and are not available to respond to changes in wind. So the changing availability itself may not be a problem but there is a need to be able to accurately predict how much response is available both seasonally and for each hour day-ahead. How much notification does the load require? Some loads require significant notification, such as industrial loads which adjust shift labor. Other loads (air conditioning, for example) require no notification once they agree to participate. How fast can the load curtail? Many loads can provide full response immediately. The aggregate load response can be ramped by curtailing individual loads in sequence. A few industrial loads have ramp rates comparable with thermal generation. How long can curtailment be maintained? This is an area where load response is frequently different than generation response. Most conventional generators incur initial response costs but can maintain their response indefinitely. Responsive loads often incur little cost for the initial response (once the capital cost is paid for) but incur large costs if response is held for too long. Aluminum smelters, for example, can provide up to about an hour’s response without damaging the process but incur very high costs if a pot line freezes. Steel arc furnaces can often provide minutes of response but not hours. Air conditioning exhibits similar duration limitations.

194

Glossary

How often can the load be curtailed? Demand response programs often limit the number of responses the utility can call for in a season. Actual limitations are often harder to quantify than duration limitations but they are quite real. Customers tend to withdraw from demand response programs if they are called on too often. This means that capabilities and needs must be carefully matched. What does curtailment cost? There are both initial costs for setting up the demand response capability and ongoing response costs. Residential air conditioning programs have significant initial communications and controls costs but little costs for actual response. Industrial response is often lower cost for communications and controls but there can be significant opportunity costs when response is needed.

There is great potential for demand response to aid wind integration. Responsive load can bridge between wind variations and the energy market cycle. Existing demand response programs provide a basis to build from. Work is required to identify load response opportunities and to develop commercial arrangements to obtain significant aggregate response.

Future outlook for dealing with uncertainty through forecasting: The field of wind plant output forecasting has made significant strides in the past 10 years. The progress has been greatest in Europe, which has seen a much more rapid development of wind power than the US due to a more favorable and stable policy environment. However, the situation is now changing in the US. Some jurisdictions have already implemented forecasting systems, others are in the implementation process, and many more are in the information gathering and fact-finding stage.

Value of forecasts

The value of wind plant output forecasting has been explored and quantified in a number of wind integration studies. Forecasts are valuable not only to assist the balancing area operator in performing his duties, but also in increasing the economic value of the wind energy to the system and minimizing any adverse economic consequences on the operation of the remainder of the generation in the system. Results from two of the recent studies are described below. The first study mentioned is the NYSERDA-sponsored study conducted by GE for the NYISO in 2005. In that study, a future NY system with 10% wind by capacity (wind generating capacity as a percent of system peak load) was examined for a scenario with 3,300 MW of wind on a 33,000 MW peak load system. In the base scenario, a unit commitment was carried out ignoring wind energy and dealing with it as it showed up in real time. A second case was carried out modeling a state-of-the-art (SOA) wind plant output forecast in the unit commitment, which led to a variable cost reduction compared to the base case, of $95 million, or $10.70/MWh of wind energy generated. A third scenario was carried out using a perfect next day wind plant output forecast in the unit commitment. This scenario led to an additional savings of $25 million, or an additional $2.80/MWh of wind energy generated. Lack of a wind forecast in the unit commitment leads to an over-commitment of fossil generation and inefficient use of that

195

Glossary

capacity when the wind shows up. As can be seen, most of the benefit of a wind forecast can be realized with a SOA forecasting system. Similar findings were obtained in another study concluded in 2007 by GE for the California Energy Commission. For a future high penetration scenario for California, in this case a simulation of 30% energy from renewables (large fraction from wind) in 2010, the value of a SOA forecast was found to be $4.37/MWh of wind energy, with an additional $.95 added from a perfect forecast. The simple message is that significant operating cost savings are available with a good wind forecast, and most of the savings is available with a SOA forecast.

Different forecasts for different time periods

The previous discussion was related to hourly wind plant output forecasts for the day ahead. Actual experience from utilities using forecasting techniques indicates that there are generally four different forecasting products which are useful for improved power system operations and reliability. The first of these is a severe weather event alert to improve situational awareness in the control room. This is a web-based real-time system which will enable operators to visualize and react to high wind events. An example is the high wind warning system based on a geographic information system platform being developed for Xcel Energy. It includes U.S. Storm Prediction Center watches, warnings, and convective outlooks in both graphical and text format. It also provides high wind forecasts for winds exceeding 20 m/s and real-time color-coded high wind observations. Most importantly, operators can quickly identify the amount of their wind generation that could be impacted by an extreme wind event. The second product is an hour-ahead forecast which provides finer time resolution for the next few hours. It is used by operators for next-hour planning, and as input for defensive operating strategies during large ramps in wind power production. This forecast is updated hourly using a rapid update cycle technique, and provides 10-minute power values for the first 3-4 hours, and hourly power values thereafter out to 6-9 hours. The value of this forecast, and the measure of its accuracy, is its ability to identify the magnitude and phase of significant wind events in time for the operators to prepare for them and do something about them. Such actions might include curtailing the output under some scenarios, limit the up-ramp in other scenarios, or procure additional reserves. The third product is the day ahead forecast, as described above in the discussion of the value of forecasts. The day-ahead forecast provides hourly power values typically for an 84-96 hour time horizon and is updated every 6-12 hours. This forecast is used in the unit commitment process and can be used for scheduling fuel purchases and deliveries for systems with significant natural gas generation. The uncertainty associated with the wind plant output forecast in this time frame is important to know, and is an area in which significant developments are occurring with ensemble forecasting techniques. The fourth forecasting product useful in system operations is a nodal injection forecast for use in the transmission congestion planning process. Separately optimized forecasts are generated for each delivery node in the transmission system. By using the real-time nodal power data along with weather model forecast data, an artificial intelligence computational learning system can be

196

Glossary

trained to make more accurate forecasts. This technique can also be used to improve the hour ahead and the day ahead forecasts.

Forecast error and uncertainty

Forecast error for wind plant output predictions can be measured in many different ways. It is fair to say that there is not yet common agreement in the industry on the best error measurement metric for the different forecast products. One common measurement is the Mean Absolute Error (MAE), which is simply the absolute value of the error divided by the predicted or reference value. It is common practice to measure the MAE of the energy with reference to the predicted value, and the MAE of the power with reference to the rated value. Typical hour ahead values of forecast error for a single wind power plant are 10-15% for energy, and 4-6% for capacity. Typical day ahead values are 25-30% MAE for energy and 10-12% for capacity. For a large aggregation of wind plants spread across a broad geographical region (400-500 mile major dimension), typical hour ahead values of forecast error are 6-11% MAE for energy and 3-6% MAE for capacity. Typical day ahead values are 15-18% MAE for energy and 6-8% MAE for capacity. These types of numbers can be achieved for a broad range of sites with SOA forecasts today, but there are additional sites which can be found on both ends of the range.

Forecasting data requirements

Real-time wind turbine power output, availability, and curtailment information is critical to the accuracy of the wind plant output forecast. The reason for this is that once wind speed and direction values have been obtained from the forecast, it is necessary to translate the model output value with what is actually happening at the wind plant. This is typically done with the assistance of a physical model, a statistical analysis process, an artificial intelligence based learning system, or some combination of these techniques. All of these techniques rely upon historical meteorological and wind plant output time series data from the site to perform the analysis, correlations, and training of the AI learning system in order to be able to produce an accurate forecast. In order to improve the accuracy of the forecast, it is critical to know if any of the turbines is experiencing an outage, or if the output of the wind plant was curtained for any period of time due to a limitation on the transmission system. Otherwise, the forecasting system will have bad data being used to train the learning system or make the correlations between forecast wind speed and plant output. In summary, to do an accurate forecast, it is critical that the forecaster have access to:

the real-time meteorological information, power output, wind turbine outage information, and plant curtailment information.

197

Glossary

This is critical for the accuracy of the forecast and the reliable operation of the system. In addition, it is critical that the BA operator have real-time knowledge of the state of the wind plant and the ability to control its output under emergency situations. The need for this information was clearly illustrated during the restoration of the UCTE system following the disturbance of Nov. 9, 2006, during which the system split into three islands and shed load. During the restoration process, as the operators were attempting to stabilize system frequency in the three islands, wind turbines which had tripped off during the event were automatically reconnecting without operator knowledge or ability to intervene. In the zone with the largest installed wind capacity, the north of Germany and the Jutland region of Denmark, the system frequency continually drifted upward during the restoration process as the wind turbines reconnected, complicating and extending the restoration process unnecessarily. The communication and control requirements for wind plants of the future need to take this situation into account; this issue should be considered in the review of the reliability standards.

Central vs plant forecasts

As discussed above regarding forecast error and uncertainty, forecasts are significantly more accurate when they cover large geographical areas than only single plants. Because of this, and for other reasons having to do with improved efficiency of both system operation and market operation, the trend around the world today is toward central forecasts done for large BA or market operators. This can be seen in the high penetration wind areas in Europe (Germany, Spain, Portugal, Denmark, Ireland), as well as in the US and Canada (CAISO, NYISO, ERCOT, PJM, AESO, HQ). The accuracy improvement available from a forecast over a large geographical region compared to a single plant can be seen in Fig. 8. For example, over a distance of approximately 1000 km, typical for a German forecast, the error is reduced to 42% of that for a single point forecast.

198

Glossary

Figure 8. Reduction in forecast error with region size (source: Focken, 2008)

Moving forecasts into the operating environment.

While a significant amount of effort has gone into developing good wind plant output forecasts for real-time, hour ahead, and day ahead planning purposes, a significant effort remains to integrate the forecasting products into the operations planning software used in the short-term planning and operation of electricity markets and systems. The major software vendors are just beginning to pay attention to this emerging need. Based on modeling and simulation work done to date, there are significant savings to be realized through the incorporation of a state-of-the-art wind plant output forecast into the day-ahead market planning and unit commitment procedures. Similarly, there are significant operations concerns with reliability implications which can be addressed through the incorporation of rapid-update hour ahead forecasts into the system operations procedures, as illustrated through the recent ERCOT event of Feb 26.

Energy Storage considerations for wind energy:

Sources of system flexibility

Time variant generation sources provide an additional source of system variability. This variability must be met with an equal level of flexibility in order for the power system operator to maintain system frequency. Sources of flexibility can be described as any source of electric energy that is available to the operator to balance system variability. This includes traditional dispatchable generators, demand response, and energy storage.

199

Glossary

Traditional dispachable generators that have ramp rates suitable to variable generation include simple cycle gas turbines, hydro turbines, and reciprocating engines. With the exception of hydro, these units typically also have the highest marginal cost of the generation fleet and therefore are utilized the least. The low utilization (capacity) factor means there is ample ability to balance significantly more variation than it does now.

Demand Response (DR) is a subset of demand side management (DSM) which includes any load reduction activities or programs undertaken by a LSE or its customers. Of particular value to system operators with variable generators is Dispatachable Demand Response (D2R), which is directly controllable. This largely untapped source of flexibility can be utilized to reduce peak demand and provide ancillary services. In addition, utilizing DR as contingency reserves has show to be an inexpensive means to deal with low probability system events such as generator outages or other unexpected large generation changes. Creating operating reserves from D2R not as widely utilized and requires a real-time telemetry back to system control. In the future, price responsive D2R could be a natural source of large amounts of flexibility.

Energy storage brings together in one location a dispatchable generator and load. Its ability to transform energy into capacity has many advantages depending on the storage medium. However, the cost of storage devices compared to other methods of flexibility currently limits their applicability to specific and limited situations.

Dedicated vs. system storage

Dedicated storage is one in which the storage device is directly tied to a particular variable generation source with the purpose of creating deterministic energy deliveries from the combined facility. The storage device may be co-located or remote to the generator, but there must be a dedicated electrical connection and continuously acting control system linking generator and storage.

System storage is linked only to power system network controls and is responsive to system operator and market signals to provide ancillary services such as regulation, load following, capacity, etc. Since it is under network control, it is available to balance variability of any source. System storage provides greater value to power system operation than dedicated storage, which represents a sub-optimum solution to the problem of system balancing.

Findings regarding storage in existing wind integration studies

A number of recent studies have been performed to quantify the incremental increase in system variability caused from high penetrations of variable generation sources. The impact is often quantified by its effect on operating costs. These include regulation, load following, unit commitment, and fuel supply costs. Studies performed by major utilities and states across the country have concluded that these costs are both marginal and well below what would be required to justify additional infrastructure, including storage. The following table outlines results from a number of these studies.

200

Glossary

201

0-0.69***nana***trace0-0.6920GE/Pier/CAIAPFeb ‘07

4.971.453.32na0.2015Xcel-PSCoApril ‘06

Dec ‘06

April ‘06

2005

June ‘03

June ‘03

June ‘06

Sep ‘04

May ‘03

Date

4.41**31**MN 20%

3.721.262.26na0.2010Xcel-PSCo

4.60na3.01.6020PacifiCorp

2.92na1.750.151.0229We Energies

1.90na0.690.091.124We Energies

0.45nanatrace0.45*4CA RPS Multi-year

4.60na4.37na0.2315Xcel-MNDOC

1.85na1.440.4103.5Xcel-UWIG

Total Operating Cost Impact($/MWh)

GasSupplyCost($/MWh)

Unit Commit-ment Cost ($/MWh)

Load Following Cost ($/MWh)

Regula-tion Cost ($/MWh)

Wind Capacity Penetra-tion (%)

Study

0-0.69***nana***trace0-0.6920GE/Pier/CAIAPFeb ‘07

4.971.453.32na0.2015Xcel-PSCoApril ‘06

Dec ‘06

April ‘06

2005

June ‘03

June ‘03

June ‘06

Sep ‘04

May ‘03

Date

4.41**31**MN 20%

3.721.262.26na0.2010Xcel-PSCo

4.60na3.01.6020PacifiCorp

2.92na1.750.151.0229We Energies

1.90na0.690.091.124We Energies

0.45nanatrace0.45*4CA RPS Multi-year

4.60na4.37na0.2315Xcel-MNDOC

1.85na1.440.4103.5Xcel-UWIG

Total Operating Cost Impact($/MWh)

GasSupplyCost($/MWh)

Unit Commit-ment Cost ($/MWh)

Load Following Cost ($/MWh)

Regula-tion Cost ($/MWh)

Wind Capacity Penetra-tion (%)

Study

* 3-year average; total is non-market cost** highest integration cost of 3 years; 30.7% capacity penetration corresponding to 25% energy penetration;

24.7% capacity penetration at 20% energy penetration*** found $4.37/MWh reduction in UC cost when wind forecasting is used in UC decision

It is important to note that these studies assume a fairly well diversified variable generation portfolio that is not constrained by transmission. There will always be situations that will create locational reliability and pricing situations that will call for varied solutions.

Modeling storage technologies

One of the challenges facing the planning community is how to model new technologies. Modeling methods for the static and dynamic characteristics as well as economic dispatch are not well known or standardized. In order to accurately account for the benefits and costs of different storage technologies, it will be necessary for planners, operators, and project developers to be certain about how they will be utilized and react to system events.

Outlook for storage technologies

There are numerous storage technologies in various stages of development and commercialization that can be used to deal with system variability. Pumped hydro and natural gas storage combined with gas turbines are well utilized, understood, and comprise the vast majority of energy storage used in our power system today.

Battery, compressed air energy, and flywheel storage are coming to market in increasing numbers and have received renewed and additional attention due to the proliferation of variable generation. Battery storage is seeing an increase in capacity and storage capability due largely to advances in cell technology. Compressed air energy storage has been around for many years and is seeing resurgence in interest, with some new variations. Appendix 1 provides additional information on conventional CAES. Flywheel storage has been used in specific applications to

Glossary

provide uninterruptible energy supply at critical loads for many years. As the units have grown is size and linked together, they are now being tested in a large scale (20 MW) regulation project in CAISO. All three of these storage technologies have varied potential value propositions due to capacity and energy capability.

Still further out in the future are technologies such as superconducting magnetic storage and super capacitors. These devices, while technically proven, are many years from being introduced to the market at a material level.

Plug-in hybrid electric vehicles

Plug-hybrid electric vehicle (PHEV) technology may prove to be a source of flexibility for the electric power system sometime in the future. The key technology limiting its market penetration is the battery. Lightweight, high power density batteries suitable to this application are not yet available at the necessary quantity and price point. As the battery technology matures and PHEVs become available, they could not only transform vehicle fuels away from oil to electricity, but also provide services that will benefit a system with increasing levels of variability.

PHEV could provide two major benefits to the electric power grid. First, assuming they don’t charge during peak, they will increase the capacity utilization of the system. Second, assuming the charging controls are integrated with the power system, they can serve as a dynamic source of energy. The combination of these two benefits could reduce diurnal demand, reserve requirements, and the average cost of delivered energy.

Future wind generator/plant behavior

Inertial response and primary frequency control

In most grid codes addressing the integration of new generation to the system, the primary frequency control is subject to specific requirements. It generally states that all conventional generators (thermal or hydro) synchronized to the transmission system must have a speed governor system to contribute to system frequency control. The frequency aspect is a major concern in islanded systems with no external AC interconnection. Changes in system frequency are caused by imbalances due to spontaneous load variations and mismatches between planned generation and the actual load level. Large interconnected systems generally have sizeable system inertia which results in small frequency deviations for such variations. In opposition, large frequency deviations on isolated systems leaning on much smaller inertia are not uncommon. The lower the system inertia, the faster the frequency will change and the larger the deviation will be if a variation in load or generation occurs. As the share of wind power in the system increases, the effective inertia of the system will decrease considering the existing technologies. While conventional synchronous generators inherently add inertia to the system, it is not necessarily the case with wind turbines generators.

202

Glossary

The fixed speed design employing a squirrel cage induction machine and the truly synchronous machines will add some inertia to the system, whereas doubly-fed induction generators (DFAG) and full converter generators do not unless specifically designed to do so. In the case of induction machines and the truly synchronous machines, there is a direct connection between the power system and the machine. When there is frequency decay on the power system, the induction machine is able to increase its output temporarily because of the slip change. The induction machines are then able to contribute to some extent to system inertia while the truly synchronous machines will inherently add some inertia to the system the same way a hydro or thermal turbine would. The basic design of converter based technology, however, does not show any inertial response. The DFAG and full converter generators employ a back-to-back converter to connect to the power system. For the DFAG design, there is a direct connection between the system and the stator while the rotor is decoupled from the system by the ac\dc\ac converter. We can take advantage of this direct coupling between the frequency of the system and the stator with appropriate control so that a frequency deviation on the power system varies the electromagnetic torque of the DFAG, resulting in a change of its rotational speed and thus free some active power (MW) acting as an inertial response. In the case of the full converter generators, they are completely decoupled from the frequency of the system. A change in the system frequency will not have any effect on the machine. Therefore, the full convertor generators will not by their design contribute to system inertia when there is a frequency deviation on the power system. But literature, although quite limited, tends to confirm that the emulation of inertial response by means of electronic controls, is also possible with this technology. In the next few years, a large amount of DFAG are planned to be integrated on power systems, thanks to their ability to maximize power extraction and their attributes regarding general system behaviors. As soon as the rate of penetration of wind turbines into power system reaches a critical point (say more than 10% of the total generation), the displacement of conventional generators by wind turbines can decrease the effective inertia of the system, resulting in larger frequency deviations to a level that becomes critical, especially in isolated systems and in periods of low load. Following a major disturbance, this could result in cascading events. Following the loss of a large in-feed, the frequency dip could deteriorate to a point where under frequency load shedding schemes would be triggered, whereas generators could be tripped due to over frequency protection following the loss of a large load. In either case, the reliability of the power systems would then be affected, which is undesirable. A consequence of the above is that additional requirements are likely to be imposed on wind plants by system operators, as is already the case for several utilities including the Nordic grid operators, ESB National Grid (Ireland) and Hydro-Québec, who have already added a frequency control requirement in their grid codes. The Nordic and ESBNG grid operators require wind plants to be able to change the active power production as a function of the network frequency. Wind plants will have to provide frequency control only when the system requires it (ex: at low load and high wind speed). Whereas the wind plants can make downward regulation of the production while at rated power following a

203

Glossary

sudden rise of the system frequency, they have to maintain a power margin (reserve margin) that may be called upon during a frequency decline. The expected response rate of each available online wind plant to frequency changes is at least 1% of the wind plant rated capacity per second, but could be more. These requirements have some impact on commercial aspects since the power margin requires some portion of the generation to be curtailed. In large interconnected systems, where the effective inertia is sizeable, it could be more economic to maintain reserves on conventional generators. While the Nordic and ESBNG requirements mainly refer to primary frequency regulation and do not require an inertial response of the machine, Hydro-Quebec’s requirement does not specifically require a reserve margin to control the frequency upward but rather refers to the use of the inertial response. Hydro-Québec requires that wind plants be able to contribute to reducing large (> 0.5 Hz), short-term (< 10 s) frequency deviations on the power system, as does the inertial response of a conventional synchronous generator whose inertia constant (H) equals 3.5 s. This target is met, for instance, when the system varies the real power dynamically and rapidly by about 5 % for 10 s when a large, short-duration frequency deviation occurs on the power system. This requires a wind plant to utilize the kinetic energy of the rotating masses. It supposes also that the frequency control is available permanently, i.e. not limited to critical moments. In 2010, Hydro-Québec will integrate the first wind plants equipped with this feature in its network. Hydro-Québec is the only transmission owner currently requiring wind plants to contribute to frequency regulation by using the inertial response. A number of papers have been written on the frequency control capability and the inertial response of wind turbines. The strategy discussed in the papers is based on the utilization of an additional derivative controller for the variable speed machines such as DFAG to control the frequency of the system. This control uses exclusively the kinetic energy trapped in the rotating masses to support the frequency changes almost instantaneously. However, in the absence of any replenishment from an external energy source, the increase of the power output can’t be sustained indefinitely. The rotor must be accelerated again after a deceleration period during the time the generator recovers its energy. This behavior, inherent to the inertial response, doesn’t worsen the frequency behavior of the system since it occurs after the frequency dip has been reached. Rather, it contributes to the system frequency behavior. The wind turbine manufacturer Repower will install a prototype of this new feature at an offshore wind plant in Germany in 2009 (Alpha Ventus research project). It will be interesting to follow the results of their research as to whether or not this strategy is viable, robust and complies with the needs of the islanded system.

No-load var production and voltage control

Currently, wind turbines are disconnected from the grid in response to extremely high or low wind speeds. They may also be disconnected from the grid in response to severe system disturbances. Under such conditions, both real power to serve load and reactive power to support system voltage are lost.

While the delivery of real power requires wind, converter-based WTGs enable delivery of reactive power without wind. Both DFAG and full converter WTGs can perform this function when appropriate controls are provided. This ability to deliver controlled reactive power will also allow voltage regulation. Such a function can not normally be provided by conventional

204

Glossary

(e.g. thermal, hydro) generation, since production of reactive power from these generators requires that the generator (and therefore the turbine) continue to spin at synchronous speed.

From a systemic perspective, the reactive power capability is similar to that provided by various dynamic reactive devices (e.g., synchronous condenser, SVC), that are used for grid reinforcement where dynamic voltage support is required.

Performance Benefits: There are two primary performance benefits associated with this type of reactive power control feature:

Continuous voltage support and regulation, regardless of wind conditions or real power generation, and

Mitigation or elimination of any adverse voltage impact due to wind turbines tripping.

Continuous voltage support and regulation improves the quality of service to local utility customers, eliminates the need for grid reinforcements specifically designed for no-wind conditions, and enhances grid security by reducing the risk of voltage collapse.

The loss of real power from a wind plant, due to either wind conditions or system disturbances, may have an adverse impact on local system voltage. The coincident loss of reactive power and voltage regulation may exacerbate the adverse impact. However, maintaining voltage regulation can reduce or even eliminate the adverse voltage impact.

The benefits of providing voltage regulation during periods of zero active power production are greatest in systems with high levels of wind penetration, or in electrically weak systems. The latter includes subsystems that are geographically remote from the main grid, wind plants at the end of long transmission lines, or wind plants connected to low voltage subsystems.

Performance illustrations: A test system, shown in Error! Reference source not found.9, is used to illustrate the improved performance achieved with no load var production and voltage regulation at a remote wind plant. The system is relatively small, serving about 700MW of load. The bulk of the load and generation are located in the central and southern regions. All generation in this part of the system is conventional thermal generation, with a mix of steam and gas turbines. A 150MW wind plant is located in the vicinity of the remote northern system, and is connected to it by a radial transmission line. The northern system serves approximately 100MW of load, but has no generation other than the wind plant. Without the wind plant, the northern system voltage is unstable at this load level, and therefore cannot operate in this topology.

205

Glossary

Wind Farm

150 MW

North System: Generation = 0 MW Load = 100 MW

50 MW

Central System: Generation = 250 MW Load = 500 MW

South System: Generation = 350 MW Load = 150 MW

200 MW

Figure 9. One-line Diagram of Test System.

Simulations were performed to illustrate voltage regulation in response to normal load variations. One scenario had the becalmed wind plant in-service with no load var production and therefore, voltage regulation. The other had the wind plant out-of-service (i.e., no voltage regulation possible). These tests were performed with a lower load level in the north system to ensure voltage stability. The load profiles were derived from measurements.

Bus voltage response to load variations, without the wind plant and therefore without any voltage regulation in the northern system, is shown in Error! Reference source not found.10. The central system voltage (light blue line) is regulated by generating units in that system. No more than a few tenths of a percent of voltage variation is observed. By contrast, the range of variation on the northern load bus (pink line) is about 4%, and would probably violate voltage flicker criteria. The reactive power output associated with the out-of-service wind plant is zero, and represented by the lavender line.

System response, with the becalmed wind plant providing reactive power and performing voltage regulation of the north system, is also shown in Error! Reference source not found.10. Variations on the north system (red line) are now comparable to those observed on the central system voltage (dark blue line). The reactive output of the wind plant, which provides the north system regulation, is represented by the purple line. In this case, the voltage is extremely well mannered. It is important to recognize that at higher power levels the system without voltage regulation is unable to serve the local load at all.

206

Glossary

207

0.960

0.970

0.980

0.990

1.000

1.010

1.020

1.030

1.040

0 100 200 300 400 500Seconds

pu

-10

-5

0

5

10

15

20

25

30

MV

Ar

North System Voltage

North System Voltage with No-Load

WTG Var

North Voltage - No Load Var Production North Voltage - Wind Plant Out

Central Voltage - No Load Var Production Central Voltage - Wind Plant Out

Wind Plant Q - No Load Var Production Wind Plant Q - Wind Plant Out

Figure 10. Response to Load Variations with and without No Load Var Production.

Without the wind plant and its voltage regulating capabilities, the northern system is unstable at 100MW load. Therefore, a voltage collapse occurs if the wind plant trips due to low wind speed. The details of this collapse are shown in Error! Reference source not found.11. The pink line represents the north system voltage, the light blue line represents the central system voltage, and the lavender line represents the wind plant reactive power output - all without voltage regulation. Without no load var production, the wind plant reactive power output falls to zero when the plant trips due to the low wind speed. Within about 5 seconds, the north system voltage collapses, precipitating a blackout of the entire system. The impact of providing voltage regulation after the turbines stop is shown in the same figure, where the red line represents north system voltage, the dark blue line represents the central system voltage, and the purple line represents the wind plant reactive power. The wind plant reactive power is maintained, regardless of wind speed and real power output, and the system voltages remain stable.

Glossary

208

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1.10

1500 1510 1520

Seconds

pu

0

2

4

6

8

10

12

14

16

18

20

MV

Ar

North Voltage - No Load Var Production

Central Voltage - No Load Var Production

North Voltage - Wind Plant Out

Central Voltage - Wind Plant Out

Wind Plant Q - No Load Var Production

Wind Plant Q - Wind Plant Out

Stable Voltage with No Load WTG Var Production

Voltage Collapse with Wind Plant Out

Figure 11. Response to Low Wind Speed Turbine Stop, with and without No Load WTG Var Production.

The examples shown here illustrate that wind turbines and wind plants equipped with no load var production and voltage regulation capability can provide improved voltage performance for weak systems. This function provides tight voltage regulation, increased system security for grid events, reduced system voltage disruptions due to wind-induced WTG tripping, and reduced need for transmission system reinforcement and/or local must-run generation.

Wind Plant Participation in AGC

The main objectives of automatic generation control (AGC) are discussed in Chapter 2. In the future, as wind plants provide an increasing amount of the energy delivered to load, it will become increasingly necessary for them to participate in a more complete range of system operation and control functions, similar to conventional plants. This will be made possible by the increasing capability of wind plant output forecasting systems, and the integration of forecasting capability with wind plant control capability in an AGC system. With a fully integrated system, the output of the wind plant can be forecast and scheduled both hour(s) ahead and day ahead, the wind plant can participate in the volt/var control system, and it may provide regulating capacity and spinning reserves if called upon to do so. It may also provide a governor response and inertial response if required.

Glossary

In addition, it will be necessary for individual wind plants to be visible to the system operator in the future. There are times when it will be necessary for wind plants to control or curtail their output in order to ensure system reliability. Such times could occur during minimum system load periods, during severe weather conditions, or during system restoration after a major outage. The system operator must know the state of the wind plant and must be able to provide it a control signal to manage its output. A recent example of this need is provided by the system restoration process after the break-up of the European UCTE system during the event of November, 2006. While the electrical islands were being stabilized and reconnected, the frequency of the islands was changing while wind generation which had tripped was being automatically resynchronized and loaded without the knowledge or control of the system operator, greatly increasing the difficulty of his task. This event dramatically illustrates the need for the wind plants to provide system status information and control capability to the balancing area operator. A renewable energy control center to perform this function has recently been added by the Spanish grid operator, Red Electrica.

SCADA Data Requirements and Communications Protocol

SCADA data requirements: The FERC Order 661-A states that “the wind plant shall provide SCADA capability to transmit data and receive instructions from the Transmission Provider to protect system reliability”. There is increasing recognition that the wind plant data is critical to performing accurate wind plant output forecasts for the balancing area operator, and for providing the status information and control capability required for reliable system operation. The data required currently varies depending on the specific utility, but usually can be classified in four categories as listed below:

Switchyard data Met mast data Wind turbine data Wind plant generation data

The switchyard data must be accessible in real-time and includes the following signals:

Alarm signals of the protection devices Alarm signals of the remote tripping systems Others alarm signals *Status signals

o Medium and high-voltage breakers o Medium and high-voltage disconnecting switches o Centralized management system of the wind plant

Electrical measurements o *Active power at the switchyard high-voltage side (MW) o *Reactive power at the switchyard high-voltage side (Mvar) o *Voltage at the switchyard high-voltage side (kV) o *Current at the switchyard high-voltage side (A)

209

Glossary

o *Active power on the medium-voltage system of the switchyard (MW) o Number of wind turbines in operation

The three following categories include statistical data. Generally, the data are compiled and transmitted every 10 minutes considering a sampling frequency of 1/5 Hz minimum. The signals found in each category are listed below. The met mast data:

*Horizontal wind speed Vertical wind speed *Wind direction *Temperature Relative humidity Barometric pressure

The wind turbine data:

Active power (kW) Nacelle direction (degree) Rotor blade position (degree) Wind speed measured at the nacelle’s anemometer Wind direction measured at the nacelle’s weathervane (degree) Status of the wind turbine

The wind plant generation data:

Active power (MW) Available active power of the wind turbines (take into account unavailability and

equipment restrictions) Available power flow capacity of the substation (take into account unavailability and

equipment restrictions) *Available active power of the wind plants (lowest value between the available active

power of the wind turbines and the available power flow capacity of the substation) *Number of wind turbines available *Number of wind turbines at stop due to low wind conditions *Number of wind turbines at stop due to high wind conditions *Number of wind turbines at stop due to low temperature conditions

* above indicates the most common signals required by system operators. The amount of data required to operate wind plants is considerable compared to the data required for conventional generators. The wind plants are generally composed of dozens of small generation units compared to a few generation units of large capacity in the case of hydro or thermal generation facilities. Signals required for conventional generators are usually limited to

210

Glossary

the signals included in the first category (switchyard data) in addition to a few data items related to the turbines themselves like the speed and the temperature, representing a few dozen signals. The three other categories are strictly related to the operation of the wind plant. One example where the amount of signals required for one wind plant is considerable is the Anse-à-Valleau wind plants (100.5 MW) integrated in the province of Quebec in Canada. Since the data listed in the third category (wind turbine data) were required for the 67 wind turbines composing this plant, there are 1537 statistical signals (categories 2-3-4) transmitted to the SCADA in addition to the 61 real-time signals required for the operation of the switchyard (category 1). The SCADA system must have the capacity to manage this important flow of data. However, the system operators can usually operate the wind plants with a limited number of real-time signals. They also need data from a few met masts to refine (auto-adjustment) the short-term generation forecast. Finally, to realize the generation forecast for different time frames, a larger amount of data are required, but can be obtained with a time delay (10 minute statistical data). Communications Protocol: The earlier discussion of wind generation forecasting for balancing areas or markets with significant wind generation identified the provision and transfer of data from operating wind plants as critical to good forecasts, but currently a significant obstacle. While addressed in FERC Order 661-A, requirements for wind plant interoperability are currently quite ambiguous. The international communication standard IEC 61400-25 (Communications for monitoring and control of wind power plants) of the IEC TC 88 is intended to provide uniform information exchange for monitoring and control of wind power plants. It will eliminate the issue of proprietary communication systems utilizing a wide variety of protocols, labels, semantics, etc., thus enabling one to exchange information with different wind power plants independently of a vendor. It enables components from different vendors to easily communicate with other components, at any location and at any time. Object-oriented data structures make the engineering and handling of huge amounts of information provided by wind power plants less time-consuming and more efficient. Scalability, connectivity, and interoperability can be maximized to reduce cost and needed manpower. The IEC 61400-25 standard is a basis for simplifying the roles that the wind turbine and SCADA systems have to play. The crucial part of the wind power plant information, information exchange methods, and communication stacks are standardized. They build a basis to which procurement specifications and contracts could easily refer. The standard builds on existing communication standards and philosophies already in use, namely the body of work encompassed in the IEC 61850 standard. It should be noted that key concepts for implementation in power system operation and control architectures, such as the ICCP (Inter-Control Center Protocol), are fundamental to the IEC 61400-25 design.

211

Glossary

Figure 12: Conceptual communication model of the IEC 61400-25 series (source: IEC 61400-25-1: Wind Turbines – Part 25-1: Communications for monitoring and control of wind power plants – Overall description of principles and models)

The communications mapping of IEC 61400-25 comprehensively defines the type of information available from each “logical” node of the information model. The richness of the information model can facilitate the implementation of all of the interoperability schemes currently under discussion, such as wide-area wind generation forecasting systems tightly integrated with control center applications. While the standard provides for communication from a “client” to ostensibly any appropriately-configured wind turbine, it is likely that applications in North America would be of a more centralized topology as shown in Error! Reference source not found.13.

Figure 13: Implementation of IEC 61400-25 communication model with a centralized topology.

212

Glossary

Many of the major vendors, as well as users and integrators, have been and continue to be involved in the development process. The list of participants includes representatives from ENERGI E2 A/S, EnerNex Corporation, GE Wind Energy, Hydro Tasmania, KC Associates, Inc., natcon7, Q-Technology, Schwarz Consulting Company, Siemens Wind Power A/S, Statkraft, Vattenfall, and Vestas Wind Systems A/S, among others.

High speed cut-out

A sudden loss of wind generation is perhaps the greatest fear of system operators. Over the past decade, there have been a few well-publicized events where significant wind generation in a balancing area was lost due to very high wind speeds across a large region, such as the ERCOT event of February 26, 2008, or the Danish event of January 8, 2005. Most commercial wind turbines utilize pitch control or other mechanisms to “spill” wind energy when wind speeds exceed the level required for nominal maximum power production. This results in a large region of rated power production over a wide range of wind speeds, which by itself is a highly desirable characteristic. However, at excessive wind speeds, usually 25 m/s or greater, mechanical loads and stresses necessitate a sudden shutdown of turbine operation, also known as high-speed cut-out. As the events referenced above, as well as some others, illustrate, the loss of large amounts of wind generation over a few hours can place significant demands on operators, or possibly even compromise system reliability. While improved wind generation forecasting for operational situational awareness is often cited as a preventative measure, there are modifications to wind turbine operation that may also contribute positively. Error! Reference source not found. 4 illustrates the modified power curve for a turbine designed to gradually reduce production in very high winds. It should be noted that such a modification is not trivial. Continuing operations in very high wind speeds has significant implications for the mechanical and structural design aspects of the turbine; for example, while the “lift” component of the aerodynamic energy capture can be well-controlled through pitching of the blades, the “thrust” component will increase with wind speed, placing higher stresses on the tower, blades, and drive train.

213

Glossary

Figure 14: Power curve for advanced wind turbine with gradual high-speed cutout.

Continued development of generic turbine models

Dealing with new wind turbine architectures

International cooperation with the generic model development effort initiated by the WECC provides strong support for continuation of this initiative as wind generation technology continues to evolve. The need for widely available and understood models appropriate for steady-state and dynamic studies of the bulk network is not unique to North American utilities. The principal attributes of these models, non-proprietary code and parameters and well-proven behavior, also appear to be a global need. Recent discussions of this topic have been focusing on the steps beyond the initial development of the generic model architectures and the distribution of code embodying these models to a much broader audience of users. There are still questions, for example, about the appropriate use of the sometimes simplified models, as well as the converse - which studies fall outside the intended application space for the models, and how should those studies be conducted. The WECC-led effort considered the four major turbine topologies in current commercial applications. Block diagrams for each were developed to encompass the range of behavior and performance across the major commercially-available turbines. However, as capabilities and features are added to the existing fleet of commercial turbines, augmentation of the structures for the generic models may be necessary. In addition, there is the possibility of new wind turbine topologies, as exemplified by the synchronous machine-based turbines now on the market. In the very near term, best representations of specific commercial turbine models with the current generic structures must be developed. This effort will require significant collaboration between the power engineering community and the wind turbine vendors, since the measurement data or detailed simulation results that provide the best opportunities for checking the behavior

214

Glossary

and adjusting the parameters of the generic models are held by the vendors and not generally available publicly. With the growing number of commercial turbines either in service or on the market, this initial validation process will amount to a very significant effort.

Long term model development, maintenance and validation

As NERC moves forward in its role as the ERO, more stringent requirements for these types of models are expected. At present, it is recognized that existing NERC standards are not being applied correctly or enforced for wind generation. Standards MOD-011 and MOD-012, for example, mandate that reliability organizations provide guidance and requirements for power flow and dynamic models. Given the lack of accepted industry standard models for wind turbines and wind plants, enforcement here has been very difficult. The current situation, with system impact studies based on one-of-a-kind, user-written, or proprietary models, is not tenable in the long term, and has actually become a significant limitation with the current installed wind generation capacity. Development of models is critical in this respect. Existing NERC modeling standards require Reliability Entities (RE) to develop comprehensive steady-state data requirements and reporting procedures needed to model and analyze the steady-state and dynamic performance of the power system (MOD-011 and MOD-013). Equipment owners are required to provide models to the RE steady state and dynamic models (MOD-012). This information is required to build a reasonable representation of the interconnection’s system for planning purposes, as stated in MOD-014 and MOD-015. In this context, proprietary or user-written models are generally unacceptable. In lieu of the accepted standard models, the common course of action for wind plant owners has been to provide no models at all, which is contrary to the requirement of these standards. Finally, there are NERC standards which deal with periodic verification of the models, such as MOD-023, which deals with verification of reactive power limits. Again, with the current process broken because of the lack of accepted models, this provision has in essence been ignored for existing wind plants. These same issues are being dealt with in other jurisdictions around the world experiencing rapid development of wind power. The process which has been adopted by National Grid in the UK in this regard is of particular interest, and can be found in a document titled “Guidance Notes for Power Park Developers: Grid Code Connection Conditions Compliance: Testing & Submission of the Compliance Report”, dealing with the full scope of grid code compliance testing and model validation. It may be found at: http://www.nationalgrid.com/NR/rdonlyres/5F1F5F26-FD98-475A-A1EA-C7584FC5C4F7/15040/GuidanceNotesPowerParksRev16.pdf Much of the current modeling activity surrounds representations of wind turbine technologies and wind plants for positive-sequence analyses, primarily power flow and dynamic simulation. As wind penetration continues to grow, there is a growing realization that other studies and evaluations are indicated in the plant design and commissioning process, for some of which the positive sequence steady-state or dynamic representations are inadequate. At present, these studies are generally conducted with a simulation platform for which a relatively detailed transient model of the wind turbine and controls already exists or can be created.

215

http://www.nationalgrid.com/NR/rdonlyres/5F1F5F26-FD98-475A-A1EA-C7584FC5C4F7/15040/GuidanceNotesPowerParksRev16.pdf

http://www.nationalgrid.com/NR/rdonlyres/5F1F5F26-FD98-475A-A1EA-C7584FC5C4F7/15040/GuidanceNotesPowerParksRev16.pdf

Glossary

Model data reporting requirements for turbine manufacturers

NERC is in the position to be able to force clarity upon most of the modeling issues that have been challenges for both transmission planners and wind plant operators. NERC can and should play a significant role in encouraging model development activities being pursued in WECC and IEEE. NERC should clearly re-state the expectation that wind generators comply with the intent of existing standards to the maximum extent possible, recognizing that there are differences that need to be addressed going forward, but setting a fixed timetable for resolution of those differences. In summary, steps that could be taken in this regard include:

1. Clarification of the expectation that wind generators must comply with standards, and a fixed timetable for compliance, with penalties for non-compliance;

2. An assessment of existing standards to determine what modifications to standards (if any) are necessary in consideration of wind generation, especially in the modeling area and including verification of models, given the somewhat unique aspects of wind generation;

3. Definition of appropriate tests for wind plants that consider the unique operational nature; verification of reactive limits for operating plants is an example, where the existing procedure may have to be modified to account for the operational characteristics.

The transition of the generic modeling activity to the IEEE Power Engineering Society Power System Dynamics Committee should provide a broader forum going forward for the needed work in this area.

Cumulative Impacts of Distributed Generation on Bulk System Behavior The amount of generation connecting to the distribution system is increasing, including sources such as wind and photovoltaics. Combined heat and power is also connecting to the distribution system in increasing amounts and while it does not have the same variable characteristics as wind, it is not very flexible, as the heat load is often the priority. These energy sources are typically referred to as distributed generation (DG). Traditionally the distribution system was a load on the bulk transmission system and represented an energy sink at all times. For very small penetrations of distributed generation, there is a small change in the characteristics as seen by the transmission system. In particular, the net energy transfer from the transmission to the distribution system (as measured typically over a year) may decline, depending on the rate of load growth and DG growth. However, as penetration levels increase, there will come a point where at certain times the distribution system is exporting to the transmission system, although the net transfer of energy is still from the transmission system to the distribution system. Further increases in DG can lead to an extreme situation where the net transfer of energy is from the distribution system to the transmission system. This situation represents a radical change in the impact of the distribution system on the transmission system. There are also very large impacts on the distribution system itself such as voltage control issues, short circuit levels and system protection, but these are not considered further here. The impact on the transmission system can be broadly categorized as load flow alterations, supply-demand balance issues, dynamics and market aspects.

216

Glossary

Load flow alterations: With the change in characteristics from being a load to a generator, the load flow patterns on the transmission system will change. These changes must be considered in the context of additional load flow changes driven by the expansion of variable generation (e.g. large scale wind) connected directly to the transmission system, and predicted load growth patterns as well. There is some thought that with the growth in DG, there may come a time when the transmission system will simply act as a backup to almost autonomous local distribution grids. The realization of many of these “smart grid” concepts are quite far in the future, but some more realistic concepts are gaining credibility and prototypes are being developed and tested, as for example in the Danish Cell Project, to be further discussed below. Supply-demand balance: With large amounts of energy being produced by non-dispatchable sources (e.g. wind) and inflexible sources (e.g. CHP), the real time balance of the system can become more challenging. As discussed earlier in this chapter, the supply-demand balance issues must be considered in the context of increasing amounts of variable generation connecting to the bulk transmission system. Dynamics: Increasing levels of DG may impact on the system dynamics. One way in which this can manifest itself is in the possible reduction in system inertia that has been previously discussed. With low inertia, the system becomes more prone to large frequency deviations and it may become more difficult to recover from large contingency events. Large amounts of DG that contribute little or no inertial response to the system can exacerbate this situation. Low voltage ride-through for wind generators has been an evolving topic over the last five years, and the matter appears to be reaching a satisfactory resolution through the development of transmission system grid codes. However, with the increase in DG and the need to establish grid codes at the distribution level, there needs to be coordination between these codes. For example, undervoltage relays on the distribution system can be used to detect an islanding situation, but the actions taken may not be appropriate for the transmission system. Market aspects: The market value of energy produced close to the load is typically different from that produced on the transmission system. Reasons for this include impact on losses and locational issues, and the need for investment in the transmission and distribution system. A simplistic view of DG is that due to the smaller amounts of energy being transported over the transmission system, energy losses will reduce. This may well be true, but needs to be studied on a case by case basis. As the amount of DG increases beyond some point, it can be argued that losses will increases as the low voltage distribution system with inherent high losses is transporting large amounts of energy up to the transmission system.

IEEE 1547 anti-islanding requirements and conflict with LVRT requirements

At the present time, the low voltage ride-thru requirements of FERC Order 661-A for transmission system connected wind plants are in conflict with the anti-islanding voltage drop-out requirements of IEEE 1547 for distribution connected wind turbines. As has been seen from incidents in Europe (Denmark, Germany, and Spain) in regions with high wind penetration, the drop-out of distribution system connected projects during system faults or disturbances can lead

217

Glossary

to system reliability issues. The conflicting requirements of these two documents need to be resolved.

Cell architecture of Energinet.dk for high wind penetration

In the present power system of western Denmark, which receives approximately 25% of its annual electrical generation from wind energy, the following security problems have been identified:

Local grids cannot maintain normal n-1 security if local generation exceeds local demand and if separation of generation from consumption is insufficient.

Security analysis has become less accurate due to missing information on local generation and unpredictable wind power.

Protection relays trip local generators after distant faults on the high-voltage transmission grid.

Traditional under-frequency load shedding schemes will disconnect both load and generation.

Restoration after faults has become more complicated and more time consuming. The conclusion reached is that to maintain efficient and safe operation of the power system with a continuous and even increasingly high share of DG, the traditional system architecture needs to be redesigned. The targets identified so far as part of the redesign process are:

Increased security of supply. Sufficient domestic resources available to maintain a balance between demand and

generation. Improved operator knowledge of actual system conditions both locally and centrally. Efficient system control particularly during emergencies. Black start capabilities using local generators. Organizing distributed generators locally into controllable Virtual Power Plants.

To fulfill the above targets, one important element of the new system architecture will be closer integration between TSO and distribution system operators (DSOs). This requires a new communication system encompassing the entire infrastructure. Local CHP plants will generate electricity only when needed, with efficient electricity markets providing proper price signals for optimization of electricity output. In a traditional power system, the large generating units are essential for system stability. When the share of these units is falling, the power system must increasingly rely on smaller units regarding fault-ride-through capability and black-start capability. Local grids and production units must be renovated in order to meet such new requirements. Numerous other measures must be adopted, and the transition into the new system architecture will be a long process. Several international research and development programs are focusing on new power system architecture to deal with this situation, and on efficient operation of DG.

218

Glossary

Conclusions and Recommendations The major findings of the chapter are abstracted and summarized in the following Conclusions and Recommendations. Additional supporting material may be found in the relevant sections throughout the chapter.

Conclusions:

1. Wind plant aggregation across broad geographical regions significantly reduces output variability and associated operating reserve requirements.

2. Wind plant output forecast accuracy increases with the size of the region forecast. 3. For balancing areas with or without wind, per unit net load variability decreases as

the BA size increases. 4. Large market size and flexibility improve the ability to deal with variability. 5. Adequate transmission is critical to realize the benefits of aggregation and large

markets, and maintain system reliability 6. Demand response can be effectively used to provide additional system flexibility. 7. Wind plant output forecasts in multiple time frames are critical for reducing and

maintaining system reliability. 8. Meteorological and electrical data from the wind plant are critical for forecast

accuracy. 9. Bulk energy storage is a valuable, but expensive and system-dependent source of

system flexibility, and is not required for individual wind plants. 10. Wind plants are becoming more capable of sophisticated control capability, which

will be increasingly valuable in providing additional capabilities to deal with reliability concerns (governor response, inertial response, volt/var control even at no-load, voltage ride thru, AGC participation).

11. International communication standard IEC 61400-25 covers wind plant/SCADA communication issues.

12. Wind plant dynamic models are a work in progress. 13. Voltage ride thru requirements and IEEE 1547 will need to be reconciled as wind

penetration increases. 14. New system architectures may be required in the future to integrate high

concentrations of distributed resources into bulk system operations. Recommendations:

1. Wind plants should provide required meteorological and electrical data through SCADA systems using standard communication protocols for use in forecasting and system operation.

2. Wind plant output forecasts should be adopted as a standard system and market operation tool for economic operation and system reliability purposes.

3. PHEV technology should be pursued as a potential source of flexibility for the power system.

4. Wind plants should be visible to and controllable by the TSO or DSO, similar to any other power plant.

219

Glossary

5. FERC Order 661-A should be converted to a NERC performance (not relay) standard and reviewed and updated periodically to reflect evolution of system requirements and wind turbine capabilities.

6. Generic public domain dynamic models and associated parameters should be provided by turbine vendors.

7. A standard requirement for model validation across all technologies should be adopted.

References [1] Kirby, B. and M. Milligan, Facilitating Wind Development: The Importance of Electric Industry Structure, Electricity Journal, April 2008. [2] Finley, C., Stinogel, J., Ahlstrom, M., Development and Implementation of an Integrated Utility-Wide Wind Forecast System, WindLogics Inc., February 22, 2008 at the UWIG Forecasting Workshop, St Paul, MN [3] Miller, N.; Jordon, G. (2006), Impact of Control Areas Size on Viability of Wind Generation: A Case Study for New York, Windpower 2006, Pittsburgh, PA, American Wind Energy Association. [4] Milligan, M.; Kirby, B. (2007). Impact of Balancing Areas Size, Obligation Sharing, and Ramping Capability on Wind Integration: Preprint. 43 pp.; NREL Report No. CP-500-41809. Presented at WindPower 2007, Los Angeles, CA. [5] M. Milligan, B. Kirby, Sub-hourly Analysis of Ramping Requirements of Wind: Impact of Balancing Area Consolidated Operations, WindPower 2008, Houston, TX, American Wind Energy Association. [6] Miller, N.; Jordon, G. (2006), Impact of Control Areas Size on Viability of Wind Generation: A Case Study for New York, Windpower 2006, Pittsburgh, PA, American Wind Energy Association. [7] GE Energy 2005, The Effects of Integrating Wind Power on Transmission System Planning, Reliability, and Operations: Report on Phase 2, Prepared for The New York State Energy Research and Development Authority. [8] Holttinen H., et al 2007, Design and Operation of Power Systems with Large Amounts of Wind Power: State of the Art Report, VTT Working Paper 82, IEA Wind. [9] Kirby, B.; Milligan, M. (2005), Method and Case Study for Estimating the Ramping Capability of a Control Area or Balancing Authority and Implications for Moderate or High Wind Penetration: Preprint. 19 pp.; NREL Report No. CP-500-38153. [10] NERC, “Data Collection for Demand-Side Management,” December 2007, ftp://ftp.nerc.com/pub/sys/all_updl/docs/pubs/NERC_DSMTF_Report_040308.pdf

220

ftp://ftp.nerc.com/pub/sys/all_updl/docs/pubs/NERC_DSMTF_Report_040308.pdf

Glossary

[11] Brochure: “International Standard IEC 61400-25: Information and Information Exchange for Wind Power Plants”, Version 4 (2660-02-20) [12] H. Muller, M. Ppoller, A. Basteck, M. Tilshcher, J. Pfister, “Grid compatibility of variable speed wind turbines with directly coupled synchronous generator and hydro-dynamically controlled gearbox”, Sixth Int’l Workshop on Large-scale integration of wind power and transmission networks for offshore wind farms, October 2006. [13] A. Mullane, M. O’Malley, “Modifying the inertial response of power-converter based wind turbine generators”, 2006 [14] Elkraft System and Eltra Regulation, “Wind turbines connected to grids with voltages above 100 kV – Technical regulation for the properties and the regulation of wind turbines”, Regulation TF 3.2.5, December 2004 [15] Nordel, “Nordic grid code 2007 (Nordic collection of rules)”, January 2007 [16] ESBNG, “Summary and discussion document for the review of frequency issues for wind turbine generators and wind farms”, Version 2.7, April 2004 [17] Hydro-Québec TransÉnergie, “Technical requirements for the connection of generation facilities to the Hydro-Québec transmission system”, May 2006 [18] F.W. Koch, I. Erlich, F. Shewarega, U. Bachmann, “Dynamic interaction of large offshore wind farms with the electric power system”, Bologna Power Tech IEEE conference, Italy, June 2003 [19] N.R. Ullah, T. Thiringer, D. Karlsson, “Temporary primary frequency control support by variable speed wind turbines – Potential and applications”, IEEE, 2007 [20] CIGRE Working Group 601 o Study Committee C4, “Modeling and dynamic behavior of wind generation as it relates to power system control and dynamic performance”, January 2007 [21] A. Mullane, M. O’Malley, “The inertial response of induction-machine-based wind turbines”, IEEE, 2005 [22] O. Anaya-Lara, F.M. Hughes, N. Jenkins, G. Strbac, “Contribution of DFAG-based wind farms to power system short-term frequency regulation”, IEE Proc.-Generation Transmission Distribution, Volume 153 No 2, March 2006 [23] J. Ekanayake, N. Jenkins, “Comparison of the response of doubly fed and fixed-speed induction generator wind turbines to changes in network frequency”, IEEE, 2004 [24] G. Lalor, A. Mullane, M. O’Malley, “Frequency control and wind turbine technologies”, IEEE, 2005

221

Glossary

[25] Hydro-Québec TransÉnergie, Projets technologiques - Exploitation, “Spécification d’exigences - Acquisition des données éoliennes,” November 2007 [26] The Danish Cell Project - Part 1: Background and General Approach; Per Lund, Energinet.dk, Denmark. IEEE PES GM, Tampa, 2007. [27] The Danish Cell Project - Part 2: Verification of Control Approach via Modeling and Laboratory Tests; Sunil Cherian, Spirae, and Valerijs Knazkins, Energynautics. IEEE PES GM, Tampa, 2007.

222

Glossary

Appendix 1 CAES as an Energy Storage Mechanism for Wind Power: Pumped-storage hydroelectric generation is a proven and viable technology for energy storage, but it is limited in its applicability to regions where there exists ample water resources and land for developing dams to store the water. Building dams can also often raise environmental concerns related to flooding of wildlife areas. Such concerns and the fact that pumped-storage is today typically not economically attractive unless it is built in sizes larger than 1000MW, which require 8 to 10 years to build, limits its future deployment. As such, there are many other energy storage options presently under research for application in the electric utility industry. Also, energy storage options are being investigated for the purpose of utilizing them as a means of controlling the variability associated with wind generation, and other variable sources of generation [1]. A storage option that is imminently emerging as economically viable, and more environmental friendly than pumped hydro, is compressed-air energy storage (CAES) [2]. CAES plants use off-peak electricity to compress air into a storage reservoir (typically an underground cavern or porous rock reservoir). Then, during peak-load hours, when the demand and price of electricity is higher, the air is released from the air store and mixed with fuel in a combustion-turbine (CT) to drive an electric generator and produce electricity. Figure 1 shows an example of a second generation CAES plant. There are other possible configurations [2]. The compressed air can be stored in several types of underground formations, such as porous rock formations, depleted natural gas/oil fields, and salt or other caverns. CAES is also economically cost-effective for short term energy storage (e.g. less than about 2 hours), where the storage medium is an above ground, man-made air-gas pipeline or vessels-tank system. Using underground geologic formations has the advantage of being able to store large quantities of air for long hours (e.g. 5 to 30 hours) of energy storage. Also, such plants are much less expensive to build than pumped-storage hydroelectric plants. Figure 2 shows a map of the US indicating regions with geological formations potentially amenable to CAES – clearly CAES can be applied in many of the wind rich regions of the US, such as the Midwest, Texas and California. PowerSouth Cooperative (which was called Alabama Electric Cooperative) built (with EPRI’s assistance) the first U.S. based CAES plant, which came on-line in June 1991. This plant uses a first generation design and has a power capacity of 110MW and its underground air store reservoir is sized to produce this power output for a maximum continuous time duration of about 26 hours.

223

Glossary

224

CT Module

ExhaustAir

Compressed Air

Compressor

Gas Turbine

Motor

Storage

Air

IntercoolersRecuperator

Fuel

Expander

Storage

Figure 1: Advanced CAES Plant: Schematic for Second Generation “Chiller Option”.

Figure 2: Geologic Formations Potentially Suitable for CAES Plants That Use Underground Storage (Source: EPRI) References: [1] EPRI-DOE Handbook Supplement of Energy Storage for Grid Connected Wind Generation Applications, EPRI, Palo Alto, CA, and the U.S. Department of Energy, Washington, DC: 2004. 1008703. [2] R. B. Schainker, Compressed Air Energy Storage (CAES) Scoping Study for California, EPRI Report, March 2007, Prepared for California Energy Commission.

Glossary

225

GGlloossssaarryy Variable resources Dispatchability etc.

reference integration of variable generation into the bulk ... of variable... · 2.7.1 facility...

Documents