Advanced Control Strategies to Integrate German Offshore Wind Potential into Electrical Energy Supply

Transcription

1 Advanced Control Strategies to Integrate German Offshore Wind Potential into Electrical Energy Supply Author: Dr. K. Rohrig Institut für Solare Energieversorgungstechnik e. V. Königstor 59, D-3119 Kassel, Germany Co-Authors: F. Schlögl, R. Jursa, M. Wolff Institut für Solare Energieversorgungstechnik e. V. Königstor 59, D-3119 Kassel Dr. F. Fischer, S. Hartge Enercon GmbH R&D Dreekamp 5, D5 Aurich Dr. B. Valov, Dr. S. Heier Universität Kassel Wilhelmshöher Allee 71-73, D-311 Kassel Topic: IT technology for an large-scale integration of wind power ABSTRACT: The paper describes the technical status and presents new concepts, models and approaches for the technical assistance of large scale wind power integration into the electrical energy supply system in Germany. It shows the technical and economic problems with the integration of high penetration of wind power plant in the network and the total electrical system. After an analysis of the effects of increased wind power generation on system planning and grid operation, we discuss innovative operational control concepts for large (offshore) wind farms. These concepts will provide a multitude of new possibilities to enable simple, flexible and undisturbed reaction to the demands of participants (wind farm operators, grid operators). These operational control strategies are mainly based on ISETs successful employed wind power prediction tools, operated by all German TSO s, and will enable energy and power control, as well as the provision of reactive power, in order to achieve comparable properties to conventional power plant types. 1 INTRODUCTION At the end of, more than 1.3 Wind Turbines (WTs) with an installed capacity of 1.5 MW generated approx. TWh and supplied about 5,5 % of the German electricity consumption [1] []. Today, the electrical power generated from wind already covers the total grid load in some grid areas temporarily. According to Federal German Government planning, in the medium-term (15) wind turbines will be erected with a total power of 3 GW on- and offshore which would cover around 15% of the German electricity consumption [3]. This large intermittent generation has growing influence on the security of grids, the operation of other power plants and on the economics of the complete German supply system. In frame of a governmental funded project, a consortium of Transmission System Operators (ENE, VE-T), WT manufacturers (Enercon), meteorological service providers (Deutscher Wetterdienst DWD) and research institutes (ISET, Uni Kassel) investigate solutions for an optimized integration of the expected large offshore wind power into the electrical supply system. Based on a detailed analysis and evaluation of the time behavior of the wind generation for future scenarios, the possibilities of new and innovative operational control concepts for large (offshore) wind farms to provide energy and power control (reserve power, constant power output, profile based generation, energy compliance) as well as reactive power and their effects on system planning and grid operation will be examined. NEW DEMANDS ON WIND POWER INTEGRATION TOOLS.1 DEMANDS ON SHORT-TERM PREDICTION TOOLS One task of transmission system operators (TSO) is the permanent grid balancing within it s control area. The grid load and the feed-in from conventional power plants is available in form of power exchange 1

2 balance group schedules and is calculated with adequate accuracy. The need for balancing power arises; therefore, from the difference in the predicted feed-in from WTs and the actual feed-in values. Therewith, the accuracy of the wind power prediction has direct influence on the amount of control power to be procured. In the last few years ISET has developed a forecast model, which is today in use at all German TSOs. It delivers the temporal course of the expected wind power for control areas, for up to 7 hours in advance []. To achieve this, representative wind farms, or wind farm groups, were determined and equipped with measurement technology. For these locations, the DWD provides forecasted meteorological data in 1- hour intervals for a prediction horizon of 3 days. The corresponding power is calculated with the aid of artificial neural networks (ANN). occurring power for the control areas of ENE, VE-T and RWE currently is about 8% of the installed capacity. The prediction error for the total German grid amounts to 7%. Figure : Online monitoring and prediction In addition to the forecast of the total output of the WTs for the next days (up to 7 hours), short-term high-resolution forecasts of intermittent generation in separate network regions or for wind farms and their clustering are the basis for a secure power system management. Apart from the meteorological values such as wind speed, air pressure, temperature etc., online power measurements of representative sites are an important input for the short time forecasts (15- minutes to 8 hours). Figure 1: ANN layout of prediction module 5 perr nerr Online Forecast The ANNs are trained with predicted meteorological parameters and contemporaneous measured power data from the past, in order to learn physical coherence of wind speed (and additional meteorological parameters) and wind farm power output. This method is superior to other procedures, which calculate the relation between wind speed and power by the use of power curves of individual plants, as the actual relation between wind speed (and other meteorological parameters) and wind farm power output depends on a multitude of local influences and is therefore very complex, i.e. physically difficult to describe. The advantage of artificial neural networks over other calculation procedures is the learning of connections and conjecturing of results, also in the case of incomplete or contradictory input data. Furthermore, the ANN can easily use additional meteorological data like air pressure or temperature to improve the accuracy of the forecasts. The deviation (Normalized Root Mean Square Error NRMSE) between the (day ahead) predicted and actual Power [MW] Figure 3: Measured and predicted wind generation incl. tolerance area In addition to the expected value of power, the forecasting system also provides a tolerance band (i.e. a reliability measure), which is determined from the experiences (prediction errors) of the past and from the data of the meteorological services. Time

3 . DEMANDS ON OPERATIONAL CONTROL OF LARGE WIND FARMS The demands on modern wind turbines are not only high reliability and maximum energy yield, but also - caused by the important role, which wind energy takes in a future power supply structure - the assumption of important functions of conventional power stations, which are partly replaced by large wind farms (wind power stations) [5]. These functions (system services) are in principle reactive power supply, supply of reserve power, the ability of primary control as well as profile based operation for individual wind farms or groups of wind parks by wind farm cluster management. The supply of reactive power can be realized as feed with given power factor or amount of reactive power at the point of mains connection, in addition, in the context of a voltage regulation on the high and/or extra-high tension level. Further demands on wind energy plants concern a driving through of a voltage drop with voltage support during the error by reactive power feed and immediate feed of the available active power after error clarifying as well as the prevention of the immediate disconnection of wind turbines at high wind speeds. This protection concepts will avoid an increasing of the primary reserve. An individual wind turbine is functional as an autonomous system. Figure : Wind farm power output control by Enercon The controlling of the WT supervises and regulates amongst others the start-up at a defined value of wind speed, the power control (and subordinate the control of the number of revolutions and the pitch angle) during operation, the feed of reactive power and if necessary the limitation of active power to a given maximum value. Operating values, warning and error messages are recorded and passed on by a SCADA system (Supervisory Control and DATA Acquisition). The operation of a wind farm consisting of several WTs of one manufacturer and one operator is equipped with one common SCADA system. The controlling of the WTs are linked to the SCADA system by data lines. For active and/or reactive power control of individual wind farms, current and voltage measuring at the grid connection points is necessary. The control variables for active and reactive power are determined by transducers and transferred to the SCADA system, implemented in the wind farm controller who admits the wind turbines with time-variable defaults for maximum active power and/or reactive power. The controller can be parameterised regarding dynamic and limit values. For a control of several wind farms at the grid connection point, the wind farm controller must be implemented as central, common element for all related wind farms. For the processing of external desired values (e.g. production management, default power factor) of a super ordinate operational control unit as well as for the feedback of status information (e.g. technical availability of the WT, available active power, available reactive power, mean wind speed) to a super ordinate operational control unit a wind farm process data interface in combination with a wind farm controller is used. With these features the following functions (features) are already realized as state of the art: - reactive power feed in (default reactive power desired value or default power factor) in the context of the WTs ability of reactive power provision and the wind supply. - generation management (default maximum active power feed) by control or regulation at the grid connection point - process data interface realized by current loops, floating distance contacts or bus protocols - maximum gradients during start-up of wind farms - power output limitation at upper frequency The time constants of the control functions range from 5 to seconds. With the possibilities mentioned, the following functions are realizable today: - schedule setting - voltage control at high and extra-high voltage level based on reactive power provision of the WTs and the wind supply In addition first applications with the following goals exist - fast voltage control in the medium voltage grid - fast supply of reserve power by power reduction of wind farms (delta control) On this basis the following, in the future technically possible control and management strategies can be indicated: - reactive power provision (default reactive power desired value or default power factor) in a conventional-power-station-usual setting range independently of the wind supply - generation management (default maximum active power feed) - schedule setting 3

4 - voltage control on the high and/or extra-high voltage level - control power provision - ability for primary control In spite of these advanced functions, the active contribution of wind farms in grid operational management can, however, no longer be performed alone by individual operational control units of single WTs or wind farms. Large (offshore) wind farms will be controlled by a system operational control center in order to coordinate and control the operation of individual wind turbines with correspondingly defined demands. As the current operational data of large wind farms will be joined together by the grid operator and desired values are given, a type of control center for clusters of large wind farms is established, which increasingly takes on the character of conventional power plants. This, however, requires the rejection of production maximization of the wind farms. New concepts for cluster management will include the aggregation of geographically dispersed wind farms according to various criteria, for the purpose of an optimised network management and optimised (conventional) generation scheduling. The clusters will be operated and controlled like large conventional power plants. Group Cluster Group Single Generator Requirements: Profile Based Operation Mode uninfluenced operation power limitation energy compliance constant power output supply of control energy Requirements: maximum power limitation (dynyamic threshold values) short circuit current emergency cut-off (disconnection) by network outages coordinated start-up and shut-down procedures (gradients limitation) Requirements: safe and reliable operation maximum energy yield TSO control unit Group Cluster Group 1 Group Figure : Wind Farm Cluster Management Group N Gen 1.1 Gen 1. Gen 1.3 Gen.1 Gen. Gen n.n 3.1 PROFILE-BASED OPERATION (WIND GENERATION SCHEDULE) Figure 5: Control Panel of Wind Power Management System operated at RWE The utilization of operation management, to be newly developed, would provide a multitude of new applications to enable simple, flexible and uninterruptible reaction to the demands of participants (wind farm operators, TSOs and electricity traders). These operation control centers must facilitate energy and power control, as well as the provision of reactive power, in order to enable wind farms comparable to conventional power plants. Since 1% accuracy of wind power forecasting is not realisable, the difference between the forecasted and actual supply must be minimised by means of control strategies of Wind farm Cluster Management (WCM) to ensure generation schedule. Power output in this case will be controlled in accordance to the schedule determined by short-term forecasting. This strategy has a large impact on wind parks operation and requires matching of announced and actual generation on a minute-to-minute basis. The schedule execution should be realised within a certain (determined by forecast error) tolerance band. Timevariable set-points should be constantly generated and refreshed for an optimum interaction of wind parks with WCM. A continually updated short-term forecasting for wind farms and cluster regions is assumed for this kind of operation management. The following control strategies will be worked out: - limitation of power output; - energy control; - capacity control; - minimisation of ramp rates. 3 WIND FARM CLUSTER MANAGEMENT The pooling of several large (offshore-) wind farms to clusters in the GW range will make new options feasible for an optimised integration of intermittent generation into electricity supply systems.

5 cluster cluster new wind farm wind farm control turbines to fault currents which requires a need of additional voltage support by conventional power plants. 1 Power [MW] LOSSES REDUCTION, OPTIMIZATION OF ACTIVE AND REACTIVE POWER FLOWS 3:15 1:15 3:15 5:15 7:15 9:15 11:15 13:15 15:15 Time Figure 7: Profile based operation Non-controllable wind farms can be supported by controllable ones of that cluster. So, the strategy allows hybrid clusters to fulfill their requirements. 3. SUPPLY OF BALANCING POWER Partly curtailed wind farm s generation can be kept within a pre-defined capacity band and be available within seconds. Potential control power of a wind park can be determined by means of short-term forecasting and reference capacity measurements and be aggregated to the cluster reserve. 3.3 CONGESTION MANAGEMENT Temporarily wind power generation already today achieves and can exceed maximum permissible thermal ratings of grid components. The situations can be foreseen and avoided by network simulations, based on wind generation forecasting and the limitation of wind power output to a pre-calculated threshold. Moreover, a direct intervention of TSO to the wind farm clusters will be necessary for the cases. Different wind farms in a cluster can be curtailed differently, that gives an opportunity for an economical optimization of the process. 3. REACTIVE POWER SUPPLY AND VOLTAGE CONTROL Wind turbines are already required to support voltage and supply reactive power. In contrary to gridconnected asynchronous generators, which only consume inductive power, new wind turbines are able to supply reactive power in accordance to set values. The demand for and scheduled supply of reactive power can be determined by forecasting and simulations. Post fault recovery of wind turbines can be improved by grid code requirements. Nevertheless there will be always a small contribution of wind Wind power generation is variable not only in time domain, but also geographically that can lead to power flows over large distances with associated losses. Such situation can also be identified beforehand and reduced or even completely prevented by redispatch processes. The implementation of these operating methods will significantly increase the economical use of wind energy. Until now, criticism has often been made that the positive contribution of wind energy in CO reduction would be decreased through the control power reserve in thermal power plants. By the manner of operation outlined above, it would be possible to provide control power by rejection of production maximization. Thereby, the possible/feasible wind farm power in feed would be reduced to the desired control band to provide balancing power by the demand of the grid operator. Therefore, wind farms can be categorized by their forecasting accuracy (e.g. offshore > coastline > onshore). Wind farms on locations with high prediction accuracy can primary used for the supply of balancing power. It is also reasonable to consider the current (i.e. forecasted) output of a wind farm. Wind farms with minimal power output are not expedient applicable for supply of balancing power. Based on innovative wind farm operational control, a control unit between system operators and wind farm clusters, the WCM will enable a profile based generation and manage the following tasks: - consideration of data from online acquisition and prediction - aggregation and distribution of predicted power generation to different clusters - consideration of network restrictions arising from network topology - consideration of restrictions arising from power plant scheduling and electricity trading - scaling of threshold values - allocation of target values to different clusters - allocation of target values to different intermittent generation plants. The combination and adjustment of advanced wind farm control systems for the cluster management will be realised by the WCM. Furthermore, the WCM prepares and administrates power output schedules for the wind farm control systems based on forecasts, operating data, online-acquired power output and defaults from the system operators. If the online 5

6 W(m/sec) Wirkleistung t W(m/sec) W irkleistung t W(m/sec) Windgeschwindigkeit W(m/sec) t measures from the wind farms diverge downward from the forecasted schedules, the WCM will supply balancing power to compensate this difference. If it diverges upwards, the WCM limits the power output of the cluster to the determined maximum. The fast and secure information exchange between energy supply, transmission, distribution and trading will be supported by powerful and reliable information systems and databases. The data processing and administration will cover the following kinds of operation: - providing installation data of intermittent generators - processing of operational data and forecasts - computing cluster production forecasts - processing of preconditions and limits - preparation of production data for TSO - providing network performance data and prediction of critical events During a pilot phase of the project, these new operating methods and control concepts will be verified at large on- and offshore wind farms and the results will be examined in respect of system management and grid control. The results of these investigations will be used to develop long-term strategies for grid connection and expansion as well as for power plant conversion. 3. OFFSHORE GRID CALCULATION TOOL After start-up of the planned large offshore wind farms, the power plant scheduling of system operation control centres of ENE, VE-T and RWE have to consider a feed-in in GW range. The impact of this large intermittent generation on grid control is calculated from weather prediction-basis by a grid calculation tool of WCM (see Figure 8). In case of bottlenecks or unintentional power flows, the TSO is able to re-dispatch wind power feed in. Windspped (m/sec) The TSO can simulate and optimise more different scenarios of power reduction of wind turbine generation, in order not to extravagate equipment capacity, to pay attention the control characteristics of wind turbine generation and not impact an economic demand of different wind farm owners. CONCLUSION In the next few years the partial replacement of conventional power plant capacity by wind power is less in focus, but more important is an improved collaboration between renewable power production and thermal power plants, where operation should be as environmentally friendly as possible. Thereto, fast adjustable power plant units are necessary, which are already increasingly implemented today e.g. gas turbines or combined cycle power plants. From the conventional power plant s viewpoint, besides the dynamic behaviour of wind power is of particular interest. 5 REFERENCES [1] Renewable Energy Information System on Internet (REISI) [] Wind Energy Report Germany, ISET. Kassel, 3,, 1. [3] Dena, Energiewirtschaftliche Planung für die Netzintegration von Windenergie in Deutschland, Berlin 5 [] C. Ensslin, B. Ernst, K. Rohrig, F. Schlögl, Online-Monitoring and Prediction of Wind Power in German Transmission Operation Centres. European Wind Energy Conference 3, Madrid 3 [5] S. Hartge, ENERCON GmbH, Electrical Engineering - R&D, Challenges and Potentials of Wind Energy Converters in Power Supply Systems of the Future, DEWEK, Wilhelmshaven Seegrid 1 Calculation paths Feed-in from windfarm (MW) Input Offshore- Gridcalculationtool Output WCM Feed-in from windfarmcluster (MW) Power Transmission System of Germany System operation control center Figure 8: Grid calculation for WCM