HYDROVISION 2011 Evaluation of Pumped Storage Operations for Coordination with Wind Resources and for Supplying Ancillary

Services.
Authors Diana Hurdowar-Castro, Ph.D., P.Eng, Director, Power and Water Optimization, Hatch, Ontario, Canada Dieter Matzner, MScE, Principal Power Consultant, Hatch Woodmead, South Africa Francois Welt, Ph.D., P. Eng., Senior Optimization Specialist and Mechanical Engineer, Hatch, Ontario, Canada

Abstract

Pumped storage has the capability to support renewable power projects by providing the necessary generation required to firm such supplies. Pumped storage capacity can also be used to trade between off and on-peak energy and to provide ancillary services especially for systems where conventional hydro is small or nonexistent. Spinning and non-spinning reserves can be produced in both the generating and pumping modes of operation, depending on the selected design of the pumping/generating units and control system. This paper discusses pump/generation facilities and their interaction in the market by describing specific experiences gained with pumped storage plants. Of particular interest is the growing role of wind power in the overall energy mix and the increased variability that results from its large scale implementation. To demonstrate the economic and operational viability of pumped storage in re-regulating wind energy, an optimization model has been setup for a potential pumped storage plant and wind farm in Lesotho, Africa. Results were generated over a full year of operation for various wind scenarios and an associated pumped storage plant. With a growing demand for energy storage and rapid expansion of renewable energy, the analysis discussed herein is applicable to the evaluation of any future large scale renewable project.

Introduction Pumped storage stations are used worldwide as a means to increase on-peak power delivery capability by storing energy during low demand periods. This is particularly useful in areas that do not have a high percentage of conventional storage hydro power in their generation resource mix and which are heavily base loaded with nuclear or coal fired plants (e.g. parts of North America, Asia and Europe). With the increased penetration of non firm renewable energy such as wind, the issue of energy storage is becoming even more significant as wind generation is highly variable and can take place at any time of the day or night. In addition, the use of renewable energy requires a higher amount of regulation and spinning reserves. Pumped storage stations can significantly contribute to the ever increasing reserve requirements. In some particular cases, pump storage stations serve a dual purpose, i.e. not only do they firm up the on-peak capacity, but they also provide water for irrigation purposes.

Design Considerations Pumped storage stations use an upper and one lower reservoir to re-circulate water in a close loop cycle. A pumped storage station can have two artificial reservoirs; however, many facilities have the lower reservoir located on an existing waterway. There are also concepts and preliminary designs where the lower reservoir is an (abandoned) underground mine or an underground excavation made specifically for the pumped storage project. It is preferable that the two storage areas are quite close to each other to minimize losses in the tunnels and conduits, and the cost of long penstocks. A typical pump storage station configuration is shown in Figure 1. Pumped storage stations operate on a daily, weekly or even seasonal basis, the difference being the size of the upper reservoir and inflows. When the upper reservoir is part of the watershed and receives natural inflow from upstream, it is considered to be an in-stream pump storage station and the operation may be affected by the changes in precipitation. Typically, the elevation differential between the upper and lower reservoir is quite large to minimize the amount of water that is required to store. Plant heads are usually greater than 100 m, and there are many instances where plant heads are in excess of 300 m (e.g. Bleinheim Bilboa, NY, USA). Pumped storage stations with 700 or 800 m heads are also increasingly common worldwide (e.g., Kazunogawa, Japan, LaCoche, France).

For a given amount of energy storage, the volume of water in the system is inversely proportional to the head. Therefore the physical size and cost of reservoirs, pump-turbine equipment and water conduits is reduced as head increases. For this reason higher head pumped storage developments are generally more economic. Large pumped storage stations are more cost effective and plant capacities tend to be quite large. Most plants have a capacity that is at least a few hundred MW, and a significant number of pump storage stations have a capacity in excess of a 1000 MW (e.g., Ludington with 1800 MW, Michigan, USA) especially for the newer plants constructed in the 1960s and beyond. The (cycle) efficiency of a pumped storage station is the ratio of energy output to energy input. Efficiency for plants constructed from the 1960s to mid 1980s is usually 68% to 74%. More modern plants have an overall cycle efficiency in the 72% to 78% range, with some designs approaching 80%. Traditionally, very large motors and generators in the power industry (and other industry) have been single speed synchronous types. This means pump-turbines will operate at a single point of operation in the pump mode, e.g., the flow and MW range reduces to a single point at any given head. However, over the past 20 years, variable speed technology has advanced to large hydro units. This is an increasingly popular design for pumped storage station as the variable speed permits a wide range of operation in both pumping and generating modes. This allows for additional reserve capability, especially in the pumping mode of operation, as well as modest increases in efficiencies. However, the cost of such units can be as much as 30% higher than the fixed speed type. Such units have been installed in Europe (e.g., Goldisthal, Germany) and Asia (e.g., Kazunogawa, Japan). Plant Operation In most cases, units are used in pumping mode during off-peak periods, typically at night, and in generating mode during the day. For the vast majority of units, there is a minimum amount of time required to switch from the pumping to generating mode, or vice versa. For newer units the minimum time for the changeover is fairly short (e.g., of the order of 10 minutes), but it can be of the order of 1 hour and can be even two hours for certain operations. In older plants, it may also be required that no more than one pump cycle per day be used, and some restriction may also exist on the generating side. These rules of operation are imposed to prevent excessive wear and tear on the units, but also to facilitate scheduling of the operation in a smooth and predictable fashion. Newer plants have been designed for a much more flexible operation and may not exhibit such restrictions (e.g., Dinorwig, UK, with up to 35 mode changes per day). Operation in a Coordinated Hydro-Thermal Environment In this regulated environment, the use of available resources is well coordinated and planned ahead of time according to the best available load forecast. The utilization of a pump storage station is then heavily dictated by the difference between on-peak and offpeak load demand.

The cost savings comes from the ability to displace more expensive non hydro peaking units required to meet load, such as diesel or natural gas fired units, during on-peak hours. The savings in production costs has to be greater than the sum of lost revenue from the 20%-30% loss in cycle efficiency, amortized debt payment and expected equity cost for the plant operation to be profitable. As a result, pump storage stations have usually a fairly low utilization factor (0.08 to 0.18). A typical schedule of operation is shown in Figure 2, where the plant is operated in pumping mode for a few hours at night, and similarly for a few hours in generation mode during the day. Operation in an Open Energy Market Environment In this environment, the utilization of a pump storage station is heavily dictated by the difference between on-peak and off-peak market prices. Similar to the hydro-thermal environment, the price differential has to be greater than the sum of losses in pump generating efficiency (e.g., 20-30%), amortized debt payment and expected equity cost for the plant operation to be profitable. As a result, there is a strong incentive to pump at low price hours and generate during high price hours. Because of price variability, the economic operation of a pump storage station may lead to a more dynamic schedule with an increased number of starts and stops or mode changes, as compared to the operation within a regulated environment such as described previously. Operation to Provide Ancillary Services Conventional hydro power plants are very effective at providing regulation and spinning reserves, as compared to other non hydro resources. The same characteristics apply to pump storage units when operated in generating mode. In general, pump storage units can provide the following ancillary services: Operating Reserve Load Following. Regulation Reserve (up and down regulation). Black Start. Supply or Absorb Reactive Power.

Newer units have been designed to be able to start very quickly, which means that they have excellent capability to respond to reserve requests.

When operating in pumping mode, fixed speed pump storage units cannot provide ancillary services. This is where variable speed units can be very useful, as they can provide the same amount of operating, load following or regulation reserve as compared to the generating mode. Operation with Water Management A few pump storage stations are part of a more complex hydro system and as such require more elaborate water management. This is the case when there is significant upstream inflow in one of the reservoirs. Plants with significant storage can also be used for flood control. In other cases, pump cycles may be interrupted during the high flow periods, as the water must be carried downstream while generating at all hours of the day. Other plants may not always be free to pump and generate according to market incentives, e.g., when they must follow a very well defined irrigation schedule with pumping required over extended periods of time. Although not all pumped storage stations can be used to provide ancillary services, especially when considering older plants, many other plants in various parts of the world do contribute significantly to the ancillary service requirements (e.g., Ludington in the USA, Dinorwig in the UK, etc.) Operation with Wind Energy Storage of wind energy (as is for the case study described below) has been the focus of considerable attention over the last few years. A number of options have been considered such as the use of batteries, compressed air energy storage technology, fly wheel inertia, only to mention a few. Pumped storage is the most established technology, particularly when considering large amounts of energy storage. However, pumped storage may also incur the highest initial capital costs and investment requirement. The design of pump storage stations for storing wind energy requires some special consideration due to the random nature of the wind energy resource, as follows: Sufficient reservoir storage capacity to handle extended periods of high or low winds. Flexible range of operation as wind can fluctuate rapidly over time, with the use of multiple units so that high efficiencies can be maintained over a wide range by changing the unit allocation. Variable speed designs are generally considered necessary to accommodate varying pumping power that is available from wind energy generation. Good ancillary service capability to contribute to the higher reserve requirements that the use of renewable energy imposes on the system. The ability to provide ancillary services can be a key factor in the profitability of the plant operation.

Case Study Lesotho Highlands Lesotho is a country land locked by South Africa as shown in Figure 3. The landscape is characterized by hill (almost cliff-like) type terrain and characteristically high winds. Apart from the logistics associated with installing wind in this geographic area, these regional characteristics provide for an ideal opportunity for wind power production. The opportunity to export wind energy to South Africa appeared economically viable with the introduction of the South African Feed in Tariff program in 2009 which provides for attractive tariffs for renewable resources such as wind. However, exporting of large scale wind, produced within Lesotho, would require firming for delivery to South Africa. The basis of the study was to gain insight into the operations that a pumped storage station (PSS) could provide to improve the delivery reliability and firming of renewable energy to the South African electricity market. Specifically the study included a preliminary evaluation of operations of a wind-PSS hybrid system and the wind power capacity required to support an on-peak firm delivery of energy over a specified period of time each week. Various peak deliveries were evaluated along with a staged PSS development of 1,000, 2,000, 3,000 and 4,000 MW. Simulation of the wind-PSS system was carried out using the Vista DSSTM model. This model optimizes the hourly operations of all reservoirs, units, and spillways in view of external energy sources like wind and other factors like market prices, load, inflow, and operational constraints. For the study herein, the model was used to produce an operations pattern of pumping and generating on an hourly basis over a year-long period. The Model The Vista model simulates the hourly operation of the hydraulic system using detailed operational procedures analogous to those used in actual practice. The model uses detailed physical system and operational constraints to schedule plant dispatch, by unit, in a manner which optimizes revenue but which abides by defined system constraints:

Reservoir physical and operational limits Unit characteristics and operational limits Historical inflow sequences Channel lag and route characteristics River flow constraints

Market price forecasts Transaction opportunities Firm contracts Transmission constraints tie lines Maintenance

The model uses a series of arcs, which link plants, power canals, spill channels and river reaches as shown in Figure 4. Generating facilities are connected to buses and interconnected to the system via transmission lines which are tied to load centers and available markets as shown in Figure 5

PSS Setup The features of the PSS system configuration are described below and shown in Figure 6: Two storage reservoirs were defined, with the upper reservoir located 800 m above the lower reservoir to provide the desired head. Four units (250 MW each) were specified at the PSS capable of pumping and/or generating up to a maximum capacity of 1,000 MW, depending on the net head and water volume moving between the two reservoirs. A controlled spillway was placed at each reservoir to handle extreme flow conditions. The plant tailrace was defined to provide a water level that is equal to the downstream reservoir elevation. No natural inflows at the upper or lower reservoirs were considered in the analysis.

Characteristics of the PSS In order to perform the analysis several assumptions on the characteristics of the PSS were required. The pumping and generating capability of the plant was modeled as four reversible units with a rated head of 800 m. The units were assumed to be able to operate in a continuous manner in both pumping and generating modes, so that it can continue to operate even under erratic wind conditions. Typical generating and pumping characteristics were adopted. In the generation mode a peak efficiency of 89% is achieved at rated head and a flow of 32 m3/s. In the pumping mode, 92% efficiency is achieved at a flow of 30 m3/s. The variation of efficiency versus flow for both generating and pumping modes are shown in Figure 7. Pumps are typically available as single speed types, however variable speeds are available but tend to be more expensive. For the purposes of this preliminary level study, the continuous mode was employed to ensure maximum use of incoming wind power.

The system was modeled to include up to four pumped storage stations each comprising a set of two reservoirs and corresponding pumped storage facilities. Storage Characteristics The upper and lower reservoirs were assumed to have a storage capability of up to 13 Million Cubic Meters (MCM), which would permit accumulation of energy over a sustained period even under a continuous pumping mode of operation. The elevation to volume curve has been assumed to be linear, with both reservoirs sharing the same characteristics. Plant Tailwater Characteristics The plant tailwater elevation was assumed to be the same as the downstream reservoir elevation. As a result, the plant head is simply, Plant Head = upper reservoir elevation lower reservoir elevation. Wind Data Mesoscale data indicating the general spatial wind speed patterns over the terrain was available. For modeling purposes however an hourly wind time series was required to
8

analyse wind coordination with the proposed PSS. Procurement of general hourly time series data (satellite derived) for each of three wind farm locations was obtained. In order to perform the analysis the Goldwind 77 1.5MW wind turbine with an 85m hub height was assumed. The power curve from the manufacturer was adjusted to the site conditions and used to develop an hourly power delivery pattern over the designated yearlong period for each site. Transmission The wind energy was modeled as a set of external energy sources. Three buses, representing the three different wind farms were defined to incorporate the wind spatial diversity. A bus collecting generation from the various sources (wind and PSS) was linked to the South African power market. The various buses were all connected via transmission lines of unlimited capacity. The wind energy can be transmitted to the main collecting bus where it can be sold directly to the South African power market under the preferential tariffs (e.g., REFIT), or it can be used to power the pumping units. Simulation Methodology The operations simulated were formulated on the basis that the Lesotho wind and pumped storage system would be isolated with only the ability to export energy i.e., South African off-peak energy was not available for pumping. Consequently pumping for reservoir refill could occur using wind energy resources only during the off-peak hours. In the off-peak o wind energy is used to pump water to the upper reservoirs. o once the reservoirs are full, excess wind energy is sold as secondary energy into the S.A. market. In the on-peak o wind energy is sold directly to the market as part of the firm delivery. o shortfalls in meeting the firm are supplied by the generation cycle of the PSS. o if the reservoirs are empty and firm can not be met the model records the magnitude of the shortfall.

Shortfalls in meeting the on-peak firm requirement were recorded and subsequently used to determine system reliability for the level of wind and PSS installed. A number of runs were performed which simulated operations for delivering 25, 30 or 35 hours of on-peak energy to a firm value of 1000, 2000, 3000 and 4000 MW corresponding to a PSS staged development of equivalent scale. The reliability associated with meeting the firm commitments were assessed as a function of the wind capacity installed.

Simulation Results As the model simulates hourly operations over an entire year the interaction between the incoming wind, pumping, generating, reservoir usage and energy export delivery was closely monitored and preliminary conclusions formulated on the reliability of supply. Simulations provided an opportunity to specifically determine the reliability of meeting firm commitment, the average shortfall, the unit operations and usage of the reservoirs. Figure 8 shows the reliability results for a 25 hour per week on-peak delivery of 1000, 2000, 3000 and 4000 MW and the associated wind farm capacity required to achieve various reliability levels. For example, results indicated that to achieve a 95% reliability level, in terms of delivering 1000 MW for a 25-hour on-peak weekly commitment, approximately 600 MW of wind would be required. The reliabilities curve asymptotically at higher reliability levels due to the magnitude of wind required to meet firm demand at that level. This trend can be attributed to specific weeks during the year when the wind is unusually low and the reservoirs are completely depleted. These specific periods of time are seen to require significant additional wind capacity to increase firm reliability. The number of shortfalls encountered during on-peak hours over the duration of the yearlong simulation and their associated magnitude, were sorted and categorized into a plot as presented in Figure 9 showing probability of occurrence and average shortfall for the various wind farm capacities modeled. With increasing wind supply the average shortfall and likelihood of occurrence decreases.

10

For all simulations, the operation of the pump storage plant was essentially optimized to minimize the number of shortfalls, given the wind input energy and the firm delivery requirement. A typical operation, in terms of generation and pumping operations over a week long period for the 25hour 1000 MW firm requirement is shown in Figure 10. The capability of the pumped storage plant to firm wind energy is further illustrated in Figure 11, which shows the wind energy input over a week long period along with the net energy output. The on-off shaping is visibly present on the output side, while being totally absent in the raw wind input energy. Upper and Lower reservoir utilization is recorded throughout the yearlong simulation period. Figure 12 shows the upper reservoir usage for a 2-week period for the 25-hour 1000 MW case. In the cases modeled herein the storage of the reservoirs was set and the operations limited by the storage capability defined.

Conclusions Storage of wind energy has been the focus of considerable attention over the last few years. A number of options have been considered such as the use of batteries, compressed air energy storage technology, fly wheel inertia etc. Storage can be an important component of any generation system when export of the wind energy, as part of a firm commitment, is required. Pumped storage is the most established technology, particularly when considering large amounts of energy storage. However, pumped storage may also incur the highest initial capital costs and investment requirement.

11

As part of a study to assess the coordination of operations, between proposed wind farms and PSS facilities in Lesotho, an analysis was performed to determine the likely operations, reservoir usage, reliability of firm delivery and average shortfall for several proposed systems. In the analysis, a review of the coordinated operations of a large scale wind resource combined with a pump storage facility ranging from 1000 to 4000 MW was presented using an optimization model to project the level of coordination. Estimates on the size of wind farm, for specific pumped storage facility sizes, was determined for several reliability levels. This information was subsequently used to assess the economic viability of the proposed works.

Dr. Hurdowar is Director of Power and Water Optimization, Hatch, Canada, with over 20 years of engineering experience. Her most recent emphasis has been on hydro operations which includes optimization analyses for power generation, power system planning, energy deregulation, relicensing, and most notably coordination between renewable power sources and hydro-thermal coordination. Mr. Matzner is principal power consultant, Hatch, South Africa. Dieter leads the renewable power team of Hatch in Africa with its' main focus of developing wind, solar (PV, CPV and CSP) and hydro-electric projects for clients. Dr. Welt is a Mechanical Engineer specialized in the development of software solutions for the hydro industry. He has held lead technical positions in the design and implementation of water management systems for electricity producers.

12