Cost-Effective Grid Integration of Offshore Wind Plants
Offshore wind power is a relatively new market that has only really emerged on a commercial scale in the last decade. There are many technical and other challenges with the design, construction, operation, and maintenance of large offshore wind plants; this is especially true when, as is often the case today, they are built with capacities of hundreds of megawatts.
The first offshore wind plants were built in Europe. A Danish project, Vindeby, was commissioned in 1991 and consisted of 11 450-kW wind turbines. The commissioning of the Horns Rev I wind plant (11 2-MW turbines) in 2002, also in Danish waters, marked the beginning of the rapid commercialization of the business. This project was the first to utilize a high-voltage connection to the mainland power system and an offshore substation on a platform. The major design considerations of Horns Rev I still form a blueprint for current projects. Presently, the 300-MW Thanet wind plant, commissioned in the United Kingdom in 2010, tops the list of such projects ranked by installed capacity. At the same time, even larger projects are under construction, mainly in the United Kingdom and in the German North Sea regions.
Offshore wind power in the United States is still in its infancy, compared with land-based wind. Under the scenario outlined in the National Offshore Wind Strategy issued by the U.S. Department of Energy (DOE), the development of 10 GW of offshore wind capacity by 2020 and 54 GW by 2030 is predicted. On 7 February 2011, unveiling a coordinated strategic plan to accelerate the development of offshore wind energy, U.S. cabinet secretaries Ken Salazar (Department of the Interior) and Steven Chu (DOE) announced major steps forward in support of offshore wind energy in the United States, including new funding opportunities for up to US$50.5 million for projects that support offshore wind energy deployment and several high priority “wind energy areas” in the mid-Atlantic that will spur rapid, responsible development of this abundant renewable resource. In September 2011, Secretary Chu announced that the DOE is awarding US$43 million over the next five years to speed technical innovations, lower costs, and shorten the timeline for deploying offshore wind energy systems. This funding will go to 41 projects across 20 states. Some of the projects include modeling and analysis tools to assess offshore wind turbine technologies and offshore wind plant system design studies.
Grid Connection Costs
The costs of offshore wind power depend on many factors, including weather and wave conditions, water depth, and distance from the coast, as well as technology maturity and availability. The larger physical structures involved (foundations, towers, turbines, platforms, and substations), the specifics of grid connection, and the complex logistics encountered in offshore wind result in considerably higher capital costs per megawatt installed than for onshore wind (up to twice as high). For example, offshore turbines are generally 20% more expensive than those designed for onshore use, and towers and foundations cost more than 2.5 times what those used for a similar onshore project would run. Figure 1 compares the representative cost structures of onshore and offshore wind projects, based on sources that include industry papers, presentations, and actual project information.
In recent years, rapidly growing demand and supply chain constraints, coupled with currency and commodity price impacts, have raised the installed project costs of offshore wind, in spite of a decline in wind turbine prices since 2008. It is highly likely, however, that overall costs will fall in the future as experience grows, innovations occur, technology advances, and competition increases. The overall cost of offshore wind energy is expected to drop by between 15% and 33% over the next decade.
Many developers and financers are wary of the risks and up-front investment costs of building offshore wind plants costing billions. These multigigawatt projects take many years to construct and may have transmission connection agreements that are phased over several years and multiple onshore connection points, all of which affects project finances.
Numerous large offshore wind plant developers are currently in the early stages of planning and designing their projects, and the market is seeing detailed requests to manufacturers and potential subsuppliers for comprehensive information related to the potential optimal design of their particular project. Such designs must consider multiple factors and take into account available technologies, capital costs, reliability, regulatory issues, and supply chain availability. For example, a team including some of the authors of this article recently conducted a study of optioneering and power system analysis for the U.K.-based Smart Wind consortium. This study included a conceptual design of the electrical power system for a 1.0-GW U.K. offshore wind plant. Various ac and dc configurations were assessed in terms of life cycle costs, technical performance, losses, reliability, operation and maintenance, supply chain constraints and risks, and other factors. The primary objective was to identify an optimum standard configuration for the grid connection to shore, the offshore platform or platforms, and the structure of the collector system.
Offshore wind plants are complex, integrated systems. But through implementation of forward-looking design processes and state-of-the-art engineering studies, there are numerous opportunities to make designs that are reliable, efficient, and cost-effective. This article examines some of these opportunities in the design of the collector system, the wind power plant, and the wind power plant integration into the mainland transmission system.
Collector System Design
The design of the electrical collector system starts with the location of the individual wind turbines, which is largely determined by wind and seabed conditions. Also, the number of offshore substations and their locations are usually given. The electrical system designer thus has the task of finding the most reliable and economical design for the collector system that connects the turbines to the offshore substations.
Radial Network Design
In almost all offshore wind projects, a radial collector system has proven to best fulfill the requirements for a reliable and economical design. In a radial system, a number of wind turbines are connected to medium-voltage cable strings that terminate at the offshore platform. This is the most straightforward concept with regard to the amount of switchgear required and the application of protection schemes. The designer can optimize the layout in terms of string routing and cable cross-sections.
The selection of medium-voltage cable for the collector system is based on the availability of standard medium-voltage switchgear. Since a higher voltage level enables more power to be carried by a single cable feeder, designers will opt for the highest voltage possible. At present, the majority of offshore wind plants utilize a collector system voltage in the range of 30–36 kV, the highest operating voltage for which standard medium-voltage switchgear is available. The increase in the power rating of modern wind turbines leads to a decrease in the number of turbines that can be connected to a single cable string, complicating the design as more strings and thus more switching bays on the offshore substation are required. Switchgear that allows higher collector system voltages and is compact enough to fit inside a wind turbine tower will certainly be an important direction for future development.
The design of the collector system must take into account all possible current flows under different generation scenarios and operating arrangements. The selection of cable cross-sections must therefore be supported by extensive power flow studies. The number of cases to be considered can easily be in the hundreds. The number of cases considered, combined with multiple iterations of the design optimization process, means that a sophisticated simulation program with sophisticated results presentation is critical for an efficient design.
The auxiliary supply to the wind turbines when a high-voltage grid connection is unavailable also needs to be considered in this stage of the design. Typically, one or more diesel generators rated at several hundred kilowatts is installed on the offshore substation platform. Islanded operation in these emergency conditions imposes additional constraints for reactive power compensation in the medium-voltage collector system. Although the modest amount of reactive power produced by the collector system when the wind turbines are not operating can normally be absorbed by the high-voltage grid connection, it may overload the diesel generators. Therefore, reactive power compensation of the collector system by fixed or switchable shunt reactors is often considered.
Another topic receiving much attention is the inclusion of a cable section that interconnects the remote ends of two adjacent radial feeders to improve the reliability of the collector system. In case of a cable failure along one of the feeders, power supply to the wind turbines behind the fault location can be rerouted through this bypass link. There should be a clear distinction between the case in which the bypass link is only designed to supply auxiliary power to the turbines during low-wind or maintenance conditions (for example, for control, yaw, and pitch systems) and the case where the bypass link is also intended to deliver power generated by the turbines. The latter case will lead to the selection of larger cable cross-sections to allow more power to be transported in emergency conditions. This highlights the importance of including reliability aspects early in the design process, in order to achieve a technically and economically optimized design.
Reliability-Centered Design Optimization
The collector cable network can be optimized by considering the locations and types of switching elements. Circuit breakers enable the fast clearance of a faulted part of the cable feeder, whereas load disconnectors only allow changing of the topology during normal running conditions. Probabilistic reliability calculations have been proven to form a sound scientific basis on which such design decisions can be made. These studies make possible a quantified decision about to what extent increased investment costs may increase availability and reduce the energy not supplied.
Probabilistic reliability analyses provide indices, such as interruption frequency (system average interruption frequency index, or SAIFI) and probability of supply interruptions (system average interruption duration index, or SAIDI) and thus allow quantification of the reliability level of individual wind turbines as well as of the whole plant. Additional reliability indices important to wind plants include various loss figures, such as cumulative energy not supplied. Computation of the net present value of the costs associated with these energy losses over the project’s lifetime permits a quantitative comparison of different design options.
Past system behavior, as observed and recorded in suitable statistics, provides information on relevant characteristic failure models and related component reliability data. These models and data enable the classification and mathematical description of failure occurrences on network components, which are then used in the actual reliability calculations to derive various supply reliability indices. Such indices represent a quantitative measure of the expected supply reliability in the network and are the basis for reliability assessment using suitable planning criteria, as shown in Figure 2. For offshore wind, due to the rather limited experience to date, it is particularly difficult to obtain comprehensive component reliability data. Therefore, a sensitivity analysis that takes into account spare part availability and weather conditions (especially relevant for offshore wind) should be performed. In addition to the technical-economical design evaluations discussed above, a reliability assessment may also assist in the development of maintenance and spare part provision strategies.
As an example, the optimization of the switching concept with respect to availability by means of probabilistic reliability calculations is presented below. Figure 3 shows five possible switching concepts—a base design and four variants—for a wind plant collector system. In this example, the feeder string splits into two strings after the first wind turbine. In the base design, a circuit breaker is installed only at the offshore platform end of each feeder. In variant 1, an additional circuit breaker is installed after the first turbine for each feeder section. In variant 2, an additional circuit breaker is installed at the midpoint of each feeder section as well. This concept is further evolved in variant 3, where circuit breakers are installed at one end of each cable section, that is, between each turbine. In variant 4 there is only a circuit breaker at the beginning of the feeder, but load disconnectors are installed at the beginning of the feeder section and at the midpoint of each feeder section. The objective is to minimize the expected cumulative energy not supplied in MWh per annum. Results are given in Figure 4. As can be intuitively expected, variant 3 was found to lead to the lowest cumulative energy not supplied. After comparing the additional investment costs to the additional reduction in energy not supplied, however, it was found that variant 4 was the optimal design, technically and economically.
Offshore Wind Plant Design
When wind power became more noticeable in their systems, network operators started to demand technical requirements from wind plants that are similar to, or in some aspects even exceed, the requirements demanded from conventional generation. These technical requirements, usually set out in grid codes, normally relate to the connection point, which is the border of responsibility between the network operator and the wind plant owner. It is a major challenge to design the wind plant, consisting of many turbines distributed over a large geographical area, so that it behaves like a single power plant as seen by the system at the connection point. The main requirements relate to reactive power and/or voltage ranges and control, dynamic behavior, and the emission of harmonics.
Depending on how the grid connection of offshore wind plants is embedded in regulatory requirements, there are two possibilities for the connection point. In some places, most notably in Germany, the national utilities are responsible for extending their transmission network offshore to facilitate the connection of offshore wind. In such environments the connection point is offshore, generally close to the offshore substation(s) of the wind plant. All the offshore-to-onshore transmission assets are in the scope of responsibility of the network operator. Alternatively, it is sometimes the wind plant developer that is responsible for facilitating the whole grid connection up to a connection point onshore. In the United States, for example, the point of interconnection would typically be onshore, at a high-voltage substation owned by the local utility.
Reactive Power and Voltage Range and Control
Network operators demand that wind plants operate within a predefined voltage range at the connection point. Moreover, they require the wind plant to be able to control its reactive power output over a predefined range. Examples of these requirements are shown in Figure 5.
The design of the wind plant should provide continuous operation within these voltage and reactive power ranges at any production level from no wind through maximum power output. Of course, the voltage levels and power flows within the wind plant collector system and the high-voltage components must also be consistent with the design capabilities of the equipment, and transmission losses must be minimized. As modern wind turbines equipped with power electronic converters have the ability to control active and reactive power independently, the turbines are the primary means to supply and absorb reactive power. A supervisory control is normally present within a wind plant that translates the reactive power or voltage demands at the connection point to set points for the individual turbines. Such a control system may limit the control range (for instance, because identical set points are dispatched to all turbines instead of having an optimized set point for each turbine, a compromise that keeps the design of the controller simple). Transformers equipped with on-load tap changers are another system component that may affect the voltage profile and reactive power flows.
When the connection point is far from the offshore substation(s), a significant length of submarine ac cable will be part of the wind plant transmission system. This is an issue for the wind plant design, particularly when the connection point is onshore and the offshore-to-onshore transmission system is part of the wind plant. These cables produce high charging currents and thus inject a large amount of reactive power. This greatly reduces the reactive power supply and absorption margin of the offshore wind turbines under different operating conditions. Therefore, reactive power compensation elements connected at the high-voltage level can usually be found in such designs. Such elements may include fixed or switchable shunt capacitors or reactors or continuously controllable devices, most notably static var compensators (SVCs) or static compensators (STATCOMs). Harmonic filters, discussed in the next section, may also serve as a source of reactive power compensation. Reactive power compensation elements may also help compliance with certain grid codes when there is no wind, when the wind turbines may have limited or no capability for reactive power compensation.
Power flow calculations are applied to assess whether a wind plant design complies with the steady-state reactive power and voltage control requirements and to size the reactive power compensation equipment. Although the collector system design work described above may be considered a separate activity, some iteration will usually be required. Depending on the complexity of the project, hundreds to thousands of power flow calculations may be necessary. An example is shown in Figure 6, where the actual reactive capability of a wind power plant (WPP) transmission system (blue) is compared with the grid code requirements (green). It can be seen that, in the overexcited region, the plant fails to inject sufficient reactive power. In this particular example, it was found that adding a switchable shunt capacitor, as well as changing the on-load tap changer target voltage, resulted in an improved capability (red line) that meets the grid code requirement.
Whereas the reactive power capability of the wind plant is largely determined by the steady-state design of the collector and transmission systems, the exact operating point at which the plant will run is determined by its controls. In general, there are three possible reactive power control modes that wind plants can be required to operate in: fixed reactive power mode, in which a set point reactive power flow (defined by the network operator) is maintained; fixed power factor mode, in which the ratio between active and reactive power is maintained; and voltage control mode, in which the wind plant contributes reactive power to regulate the voltage magnitude at the connection point. The choice of control mode is generally determined based on local regulations or by the network operator.
Although the control mode is clearly focused on the quasi-steady-state behavior of the wind plant, network operators also impose certain dynamic performance requirements. For instance, in the United Kingdom, wind plants are required to contribute to voltage control with a predefined reactive power–voltage droop characteristic. If a sudden voltage change occurs in the network, the wind plant is required to start reacting no later than 200 ms after the change and should provide at least 90% of the required reactive power within 1 s. Two seconds after the event, the oscillations in the reactive power output may be no larger than ±5% of the final value.
Because the wind plant control is implemented at many levels (e.g., wind turbines, the plant controller, reactive power compensation elements, and on-load tap changers), each with their different characteristics, it can be very challenging for the wind plant to meet the required dynamic response under all possible operating conditions. A transient stability study representing all individual components and their controls is necessary. This type of study demands a level of precision for which generic or standard models are no longer sufficient. Vendor-specific simulation models—preferably validated with measurements—are essential. These detailed transient stability studies are crucial, especially for wind plants that employ switchable shunt capacitors or reactors or compound reactive compensation where the output range of SVC or STATCOM devices is extended by switchable shunt elements. These studies form the key input for determining the exact switching conditions and delay times. Experience has shown that it is essential to model phenomena that are not usually considered in planning-level transient stability studies, such as the details of interlocking schemes and some representation of the limitations of the mechanical drives of the circuit breakers. These additional efforts have frequently led to more cost-effective designs of wind plant transmission systems and reactive power compensation.
Other important dynamic phenomena, such as fault ride-through and participation in frequency control schemes, are determined primarily by the wind turbine designs. These controls are an important component of the stability study performed in the interconnection analysis.
The influence a wind plant may have on the harmonic voltage distortion at the connection point is a critical factor that has to be considered early in the design process. The maximum allowed emission level of specific voltage harmonics and the total harmonic distortion are defined by the utility or by applicable engineering standards. Simulations to determine the harmonic performance of the wind plant, taking into account the frequency dependencies of all components, are therefore an essential part of the design process. If these calculations show that the wind plant will exceed the harmonic emission limits, mitigation measures must be investigated. These will typically lead to the installation of harmonic filters.
In offshore wind projects, there are three main sources contributing to the level of harmonics at the point of interconnection: the background harmonics already present in the grid prior to the connection of the wind plant, the wind turbines, and—if applicable—the dynamic reactive power compensation equipment.
The background harmonics should obviously be lower than the harmonic emission limits, but they may become amplified by the wind plant’s own electrical network. Long, high-voltage ac submarine cables have frequency characteristics that could easily cause critical resonances with the power system at relatively low frequencies. These resonances could amplify the harmonic distortion present in the utility network to such an extent that the emission limits are exceeded, even without a single wind turbine online. This very different from onshore wind projects, where the interconnection distances are usually shorter or realized through overhead transmission lines.
To accurately determine the amplification of background harmonics, adequate modeling of the impedance of the utility network seen from the wind plant is of key importance in the harmonic performance study. It is not sufficient to have only a single value of the network impedance for each harmonic frequency, as this impedance is highly dependent on the switching state and loading level in the grid and, hence, may change considerably over time. The usual approach is therefore to obtain (through simulation) the network impedance for a large number of switching states and loading levels of the utility network. These can be summarized as a set of impedance loci that encompass all impedance values found. The loci are areas in the complex impedance (R–X) plane that enclose the impedance values the network may have over a certain frequency range and can be described by straight lines or circular sections. The impedance loci are usually provided by the network operator. An example is given in Figure 7.
During the harmonic performance analysis of the wind plant, the challenge is to select, from the provided set of loci, the grid impedance that leads to the highest amplification of the existing harmonics at the connection point. As this has to be done for every harmonic order that could realistically occur, the harmonic analysis is typically performed up to the 50th order.
Offshore wind turbines are almost exclusively of variable-speed design and are therefore equipped with power electronic converters. Power electronics actively inject harmonics into the network. These harmonics can be divided into characteristic harmonics, which are mainly the sidebands of the switching frequency, and noncharacteristic harmonics, which are dependent on the actual operating point and the network impedance seen from the turbine terminals. Turbine vendors can usually provide a current or voltage spectrum that has been obtained during type testing. In the harmonic performance study, the turbine can then be modeled simply as a current or voltage injection having this spectrum. A caveat of this approach is that the noncharacteristic harmonics may not be realistically represented and more elaborate modeling may be required in some cases.
The wind turbine’s contribution to the harmonic voltage distortion at the connection point has to be determined by assuming the worst-case network impedance in terms of resonances. It should be noted that the grid impedance value leading to the highest amplification of the background harmonics is, in general, different from the value resulting in the highest wind turbine contribution.
If the harmonic contributions of all wind turbines are assumed to be in phase, which is the case when the spectrum obtained during type testing is applied for modeling, the aggregated contribution at the connection point will be greatly exaggerated. Therefore, summation laws can be applied to more realistically account for angular differences, the randomness of the harmonics, and, in particular, the noncharacteristic harmonics. IEC Standard 61400‑21 provides such a summation law. Some vendors equip their turbines with specific control schemes whose objective is to displace the phase angle between turbines to minimize the impact on connection point harmonic distortion. Such schemes have a large influence on the harmonic performance study and require close cooperation with the vendor to determine the net harmonic contribution.
Reactive power compensation elements will also affect harmonic performance. Power electronics-based devices, such as SVCs and STATCOMs, inject harmonics into the grid just as the wind turbines do, and therefore similar modeling should be applied. Mechanically switched shunt capacitors or reactors are passive components that do not actively inject harmonic currents, but their impedance may have a significant impact on both the amplification of background harmonics and the contribution of the wind turbines.
The total harmonic voltage distortion at the connection point is dependent on the three sources described. These contributions can be determined separately but should not be added arithmetically because they are not in phase (as with the individual wind turbines) and because they have generally been calculated with different values for the equivalent grid impedance. Superposition can still be applied if one keeps in mind that the results will be conservative.
If the harmonic performance analysis indicates that emission limits are likely to be exceeded, one or more filters will typically be added to the wind plant design. A C-type filter tuned to a single frequency is a popular choice, since in many cases it can sufficiently suppress the dominant problematic harmonics and at the same time it supplies reactive power. Because such filters influence the reactive power capability of the wind plant, the steady-state design and the harmonic performance analysis will usually require some iterations.
Integration of Offshore Wind Turbines into the Grid
Studies are performed to investigate the impact of any new generation addition to the grid. Many of the studies performed are the same for an offshore wind plant as they would be for an onshore one or even a conventional thermal or hydroelectric plant. Studies typically performed include:
- power flow studies to see the impact of the new generation on system steady-state flows and voltages
- contingency analysis to see the impact of the new generation for events involving loss of transmission circuits, transformers, or other generators so as to ensure that postcontingency flows and voltages are within their respective limits
- power flow studies to see the impact of the generation on power system transfer capability, particularly across defined system interfaces
- short-circuit studies to ensure that the addition of the generation does not cause overduty of breakers or other equipment in nearby substations
- transient stability analysis to test the response of the new generation and nearby system to faults occurring on the power system and to ensure that generation remains in synchronism and performs in an acceptable manner.
Studies particular to wind plant interconnection analysis would include:
- reactive power analysis to ensure the wind plant meets grid codes for supply of reactive power and/or voltage control
- harmonic analysis, as described above
- flicker analysis to determine the potential impact of changes in the wind plant output on the voltages at nearby substations to ensure that excessive voltage fluctuations (flicker) do not affect local loads
- low-voltage ride-through analysis to show compliance with grid codes, demonstrating that the wind plant can remain connected to the system for faults of particular durations and severities.
Other types of studies may be necessary. For example, the studies needed will be affected by the type of interconnection—that is, whether the interconnection consists of an ac transmission line or cable or a dc inverter and cable system. AC transmission is well understood and globally proven. The equipment is relatively compact and readily available. The length of an ac cable connection to an offshore wind plant is limited by cable charging current. Offshore platform design is driven by reactive power compensation and harmonic filter requirements, as discussed above. For ac interconnections, special design studies include load rejection analysis, which evaluates the impact on the wind plant if the ac cable is opened at the system end and the large amount of cable charging capacitance is left connected to the isolated wind plant.
Use of HVdc to connect offshore wind is a less mature technology. But dc cables do not suffer from the length limitation that comes with the use of ac cables. In many cases, dc cables will have significantly lower losses than ac and can be designed to have a higher capacity. Substations associated with HVdc are generally much larger, however, and costs are significantly higher. Special design studies are often needed for HVdc interconnections. For example, studies may need to look at the interaction of the dc controls with other nearby power electronic controllers (HVdc, SVC, and so on). Subsynchronous torsional interaction with nearby units should also be considered.
The outlook for offshore wind is very favorable. Offshore wind is an important component of national plans to increase the amount of renewable generation and reduce dependence on the use of fossil fuels. Government incentives will assist in the development of offshore wind projects and foster research needed to advance the related technologies.
Such technological developments are continuously occurring. Wind turbine sizes are growing, with 3–6-MW turbines being used for offshore wind projects today and 10-MW turbines being designed. There are also advances occurring in power electronics, which are expected to reduce cost, increase reliability, and give better controllability and system-friendly characteristics.
System analysis tools, techniques, and experience are also improving. The design and analysis techniques discussed above will continue to evolve to incorporate new technologies and changing system needs.
For these reasons, we believe that offshore wind will play a significant role in the supply of energy to many of the world’s power grids in the relatively near future and will play an even larger role as experience is gained, technologies evolve, and costs are reduced.
For Further Reading
U.S. Department of Energy. (2011, Feb.). A national offshore wind strategy: creating an offshore wind energy industry in the United States. Available Online
National Grid. (2011, Sept.). 2011 offshore development information statement, offshore electricity transmission: Possible options for the future. Available Online
U.S. Department of Energy. (2008, July). 20% wind energy by 2030. Available Online
U.S. Department of Energy. (2011, June). 2010 wind technologies market report. Available Online
OffshoreGrid. (2011, Oct.). Offshore electricity infrastructure in Europe. Available Online
European Wind Energy Association. (2011, July). Pure power—Wind energy targets from 2020 and 2030. Available Online
U.S. Department of the Interior. (2011, Feb. 7). Salazar, Chu announce major offshore wind initiatives. Available Online
James Feltes is with Siemens Power Technologies International, Schenectady, New York.
Ralph Hendriks is with Siemens Power Technologies International, Erlangen, Germany.
Steven Stapleton is with Siemens Power Technologies International, Manchester, United Kingdom.
Ronald Voelzke is with Siemens Power Technologies International, Erlangen, Germany.
Baldwin Lam is with Siemens Power Technologies International, Schenectady, New York.
Nancy Pfuntner is with Siemens Power Technologies International, Schenectady, New York.