Abstract

Abstract-- 1 This paper explores the available control options for enabling wind power generation plants to participate on the maintenance of system frequency following a major power imbalance. Taking the currently employed control structures for wind generators as the baseline case, possible expansions and additional features have been discussed. The options include voltage or alternatively frequency dependent active power control. The responses of these control schemes vis-à-vis their frequency supporting capability in a power system contingency situation have been simulated and with one another compared. It was found out that at the conceptual level there are indeed a range of options which would place wind generating plants in a position to support system frequency in an emergency situation. Index Terms-Wind power, Wind generator control, Frequency stability, Interconnected system I. INTRODUCTION HE rapid expansion of wind power generation experienced during the past decade is, in all likelihood, set to continue well into the future. Germany, maintaining its precursor role in terms of installed wind capacity, has set itself some ambitious goals. By 2010, the proportion of electricity consumption to be covered by renewable energies is envisaged to rise to at least 12.5%, and further down the line to at least 20% by 2020. Preliminary studies foresee further significant increases in the timeframe extending to the year 2050. Wind being the only renewable energy resource capable of meeting the projected goals in the near to medium term, it is safe to conclude that most of the anticipated increases in renewable energy is going to be wind based. Consequently, in addition to the current large number of onshore sites, many offshore sites are in the planning or implementation stages. The installed capacity of the offshore plants in the North and Baltic seas is expected to reach 2-3 GW by 2010. This figure is projected to rise to 20 -25 GW by 2030. It is already clear that the interconnected European network (UCTE) will, for example, have to accommodate a significant I. Erlich is with the University Duisburg-Essen, 47057 Duisburg, Germany (e-mail: erlich@uni-duisburg.de). K. Rensch is with AREVA T&D Energietechnik GmbH, 45356 Essen Germany (e-mail: kevin.rensch@areva-td.com). F. Shewarega is with the University Duisburg-Essen, 47057 Duisburg, Germany, (e-mail: shewarega@uni-duisburg. de) wind power component in the years to come and wind, in all probability, will feature prominently in the generation mix in many parts of the world. The task at hand, therefore, is the efficient integration of both on-and offshore wind power plants into the interconnected power system by maintaining the current level of system security. Apart from the overall share of wind power, many large off-and onshore wind farms already possess generating capabilities on a scale of conventional power plants, and the number of in-feed points into the network from such plants will continually increase in the years to come. This together with the prominence that wind power has attained in relation to the aggregate system-wide power underscores the need for wind farms to be in a position to undertake tasks necessary for maintaining system frequency within the prescribed tolerance band. This task is currently considered to be the exclusive responsibility of conventional power plants. Moreover, the performance of the wind generating plants during a contingency situation, particularly their fault ride-through capability, and the impact that the wind plants make on the rest of the system is an issue of immense interest. There is a strong need now, for example, for power system operators to carry out stability analysis of the whole power system including dynamic models of the wind farms In earlier papers, the effect of the increased share of wind power on system frequency and voltage profile was presented [2], [3]. This paper builds on those studies and aims to assess the impact of a much larger share of wind power (exceeding 50% of the overall power) on system frequency and voltage profile during a contingency situation. The paper then will dwell on possible control measures to be implemented in wind generating plants to counter the frequency sag arising from a sudden loss of generation or increase in system load. Taking the existing control systems for wind generators as the baseline case, possible expansions and additional features capable of providing frequency support are explored and their performance evaluated. II. DESCRIPTION OF THE TEST SYSTEM Before delving into the detailed discussion of the alternative control options, some background information pertaining to the test system is provided below. A. The test network The test network to be used for this simulation is a 137-bus system containing 16 conventional power plants, 28 transformers and 124 transmission lines. The one-line diagram of the network is given as B. Control of the doubly-fed induction machine (DFIM) The objective of this study, as stated above, is to assess the impact on system frequency of the increasing share of wind power in relation to the overall power during a large power unbalance within the system. For this purpose, some of the conventional plants are to be replaced by wind generation plants in stages, staggered incrementally as follows: The impact of each addition, starting from A (6.3% wind power) to all others up to H (51.9%), has been studied separately. All wind generators replacing these conventional plants are assumed to be of the DFIM type. The DFIM encompasses two control structures, the fast electrical control and the slower mechanical pitch-angle control as shown in C. Configuration of the wind farm As is well known, the output power of the wind plant depends strongly on the wind speed. Due to the random nature of wind speed, a wind plant replacing a conventional power plant and that is expected to deliver the same amount of power needs to have a larger size. The installed capacity of the wind farms in this case is determined to be 1.5 times larger than the capacity of the conventional power plant replacing them. The nominal output power of one of the wind farms, for example, is 1650 MVA. This corresponds to the output of 528 pitch controlled DFIM, each with a nominal capacity of about 3 MVA. The individual units have been represented by a single equivalent model. It is necessary to ensure that the simulation with and without wind power generation has the same initial conditions so that the results can be compared with one another. This implies that during normal operation, the wind farms need to feed the same amount of active and reactive power at the same voltage as the conventional power plants they are replacing. For this purpose an optimization algorithm was implemented in the software package used, which varies the active and reactive power generated by the wind plants until the load flow (with and without wind generation) matched. Because the study essentially focuses on frequency stability following large disturbances, it is necessary to describe voltage and frequency dependency of the loads. The corresponding model is represented by equation (1). The numerical values for voltage and frequency dependency indices used in this study are based on the results of an extensive study carried out on the German network in the late 1980s III. WIND GENERATOR CONTROL OPTIONS AND THEIR IMPACT ON SYSTEM FREQUENCY DURING A DISTURBANCE The centerpiece of this study is to characterize the impact of the loss of generation on system frequency as the share of wind power increases. During the simulation, only the generation allocation is being altered in stages in favor of wind plants, with everything else including the overall generated power remaining unchanged. The disturbance may be the loss of a major tie line carrying a large amount of power or a loss of a significant amount of generation resource. In this work this has been simulated by increasing the load by 1.3 % or alternatively by 2.5 % of the total load (at cosφ = 0.98). At t=2 s the additional load is switched on at various buses concurrently. Four of the 16 power plants are earmarked for primary control. With the primary controlled power plants deploying all the available primary reserve, the frequency deviation for a sustained load increase of 1.3 % will be 200 mHz, which corresponds to the frequency droop in the UCTE system for a similar loss of generation. The objective now is to identify a possible control structure for wind generators that is best suited to provide frequency support during a contingency situation. Of the five control options listed below, the first two are already well-established control schemes for the DFIM. Taking these two as the baseline cases, possible improvements vis-à-vis the frequency supporting capability of the machine will be investigated. The list of the control options are: (1) Constant P and constant Q control (2) Constant P and constant U control (3) Constant P and U control with superimposed static frequency dependent element (4) Constnat P and U control with superimposed static and dynamic frequency dependent element (5) Frequency dependent P and constant Q control A. Baseline case The basic structure for both i. Active and reactive power control In case of an active and reactive power control, i.e. option (1), the control structure is identical to the one in It can thus be concluded that in the later stages of the disturbance, in terms of maintaining constant frequency, the system performs better when it includes a larger wind power component. The reason for this behavior can be explained as follows. The increased frequency drop observed in the immediate aftermath of the disturbance is caused by a smaller inertia support provided by wind generating plants as discussed in the previous section. The presence of voltage controllers maintains the power drawn by the load constant at nearly the pre-disturbance value. The voltage controller, by maintaining the load constant, removes the cushion that the load provides in support of the system frequency following the loss of generation. control This option involves a slight modification of the scheme discussed in the previous section (Section A(ii)). The voltage control channel is augmented by a new proportional term, whose input signal is the frequency deviation. The output is then subtracted from the voltage reference, as shown in Additionally, a 100 mHz or alternatively a 200 mHz dead zone is introduced into the control structure to avoid any unnecessary tampering by the added element with the operation of the voltage controller. C. Static and dynamic frequency dependent voltage control This option involves the expansion of the static frequency dependent voltage control described in the previous section by an additional, dynamic frequency element. The additional controller (also called "booster") is chosen to be a differential element. The structure of the setup is shown in The voltage controller augmented by a superimposed frequency control loop is impacting the system frequency indirectly. Because of the additional frequency dependent signal, the voltage reference value will be reduced. As a result, the voltage level within the grid is also slightly reduced, which leads to less power being drawn by the load. As a result, the power balance is reached faster. This can also be attained by using a simple proportional frequency dependet term for the voltage controller. It should also be mentioned that the method is not limited to wind generators only. On the contrary, conventional power plants and their excitation control system can be included, and this may even be recommendable to avoid local voltage dips within the grid. D. Frequency dependent active power and constant reactive power control This section outlines the control option whereby wind generators implement a frequency dependent P control. As a consequence of the frequency deviation, the controller is designed to reduce or increase the reference value of the real A: 100 mHz dead zone B: 200 mHz dead zone 6 power controller (P ref ) for a brief period using derivative of -∆ f or +∆ f. ∆ f is defined as the actual frequency minus the nominal frequency (50 Hz). WT with frequency dependent P-and Q-control (+) As a result, the frequency dip will be reduced. As can be seen from As stated above, the steady state frequency drop for this amount of loss of generation, with no wind generators connected to the system, would be 1000 mHz. All the results obtained using the various control options should, therefore, be evaluated against this backdrop. The best result is achieved when wind generators employ a real power and static plus frequency dependent voltage control. A similar steady state result is obtained for a real power and static frequency dependent control. If a dead zone of 100 mHz is used, the steady state frequency deviation for the test network chosen in this study turns out to be 164 mHz. A steady state frequency deviation of 669 mHz is obtained when P and Q control is implemented, which in any case is the current control strategy for DFIM. A similar result is obtained for frequency dependent P control, but the initial frequency gradient tends to be steeper. However, it should be emphasized, that the last option requires a strong coordination between power and frequency controller. If this is not the case, system behavior may be adversely affected. The least favorable option in terms of frequency stability seems to be P and constant U control. The final frequency deviation in this case is 1070 mHz, which is well below the set point for the activation of frequency relays. This is due to the fact that the voltage controllers used in this study for wind generators are very strong and therefore, always stabilize the voltage very fast at the reference value. As a consequence, the voltage controllers should be designed not only with the view to maintaining constant voltage but also taking into account the effect of a softer voltage behavior on frequency stability. On the other hand, voltage control of wind generation plants will be necessary in the future due to the fact that all conventional generators to be replaced by wind are voltage controlled. In case of constant U controlled wind generators, the overall reactive power supply (with or without wind generating plants) is nearly constant. For constant Q control, however, the overall reactive power output drops somewhat. This is caused by the reduced voltage profile in the network, especially at buses near the wind generation plants, and the resulting drop in the reactive power absorbed by the load. In terms of reactive power, the largest output reduction by the wind plants occurs in case of frequency dependent U control. 8 lower voltage drops. The frequency dependent U controller causes a voltage drop, which depends on the level of the frequency sag. The bar graph given as V. CONCLUSION This paper explores the available control options for wind generating plants to participate on the maintenance of system frequency commensurate with their share in relation to the overall power. The study has revealed that there are indeed possibilities in terms of involving the wind generating plants by expanding the currently used control structures. A wind farm consisting of P and frequency dependent U controlled wind generators is capable of achieving a significant improvement in terms of frequency stability following a major power imbalance compared to the conventional generation plants. The downside of this approach is that it involves accepting a temporarily reduced voltage profile in the network, especially at buses near wind generation plants. In summary, based on the outcome of this study the following observations can be made: An increasing wind power generation leads to a higher frequency gradient. With increasing wind power the initial frequency drop after a disturbance will be faster. After graduation, he joined the Addis Ababa University, Ethiopia as the member of the academic staff where he served in various capacities. Currently he is a member of the research staff at the University Duisburg -Essen. His research interests are focussed on power system analysis and renewable energy technologies.