IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 24, NO. 4, DECEMBER 2009 935 Clustering of Wind Farms and its Sizing Impact Hadi Banakar, Member, IEEE, and Boon Teck Ooi, Life Fellow, IEEE Abstract—The problem of establishing maximum size of a wind farm (WF) when it joins a WF cluster (WFC) is examined. It is shown that the clustering effect can be quantified by detailed simulation of system dynamics, using appropriate models for wind events, wind facilities, and power system equipment. The paper points out the inevitability of WFC formations and draws attention to their operational impacts. A case study is used to demonstrate how location and layout of a WF can influence system dynamics, and thus, restrict the WF size. Index Terms—Clustering, frequency dynamics, system dynamic response, wind events, wind farm clusters (WFCs), wind power. I. INTRODUCTION R ECENT statistics on installed wind turbine generators (WTGs) indicate a strong trend in increased wind power penetration in both European and American power grids [1], [2]. Given the continued advances in WTGs efficiency and cost/megawatt (MW), as well as political, economical, and en- vironment issues tied to the use of fossil fuels, one can expect this trend to continue well into the future. High wind penetration can lead to the concentration of wind farms (WFs) in relatively small geographical areas where fa- vorable technical and economic factors combine. Such circum- stances exist today in northern part of Germany and in Denmark, along the west coast of Limfjorden. In the United States, a sim- ilar situation is present in West Texas where nearly 1000 MW of wind power capacity has been constructed in Pecos River area [3]. In Canada, Hydro-Quebec is building 2000 MW of wind power capacity in the Gasp´ esie region of Quebec [4]. When multiple wind facilities operate out of a small area, their combined outputs, could give rise to a host of operational challenges. The most obvious one, as is the case in Texas, is con- gestion of the lines that carry their generations to the rest of the power system, under sustained high wind conditions. Also, re- gional depression of the grid voltage may occur as certain WTG technologies absorb large amounts of reactive power at high outputs. Both these problems are manageable when “spilling wind” is an option. That strategy, however, will be of little help when the facilities respond to a wind die-out in unison. WFs within a small geographical area are typically driven by a single wind regime. As such, their outputs are correlated and their responses to a wind speed fall coalesce to form a Manuscript received October 18, 2006; revised April 12, 2007. Current version published November 20, 2009. This work was supported by the Natural Sciences and Engineering Research Council of Canada through a strategic grant on reducing greenhouse gas emission. Paper no. TEC-00481-2006. The authors are with the Department of Electrical and Computer Engi- neering, McGill University, Montreal, QC H3A 2A7, Canada (e-mail: hadi. banakar@mcgill.ca; ooi@ece.mcgill.ca). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TEC.2008.2001454 large aggregate down ramp. When the rate of this ramp is well beyond the combined ramp rate of the online units, the ensuing frequency and voltage excursions could set off the protection system relays and jeopardize continuity of the system operation. The aim of this paper is to draw the attention of wind power practitioners to: 1) the relationship between the internal layout of a WF and power system dynamics; 2) implications of WF clusters (WFCs) for power system planning and operation; and 3) the likelihood that wind power potentials of some power sys- tems have been overestimated by ignoring the clustering effects of their future WFs. The WF sizing problem is chosen as the vehicle for reaching the stated goal. The tendency of WFs to cluster was first noted in [5] where its influences on local weather patterns were studied. The problem of “coincident wind regimes,” reported by Bonneville Power Administration (BPA) [6], also entails concurrent ramps at mul- tiple adjacent wind facilities. As clustering WFs exhibit the same symptoms, the two phenomena may be indistinguishable. This paper is organized as follows. Section II provides back- ground to subsequent sections while Section III details is- sues related to WFs response, clustering, and wind events. In Section IV, modeling of wind events, WFs, and power system equipment are reviewed while the adopted approach and sim- ulation methods are outlined in Section V. In Section VI, a case study is introduced and simulation results are presented. Section VII concludes the paper. II. BACKGROUND A. Sizing of WFs Sizing studies are performed as part of generation expan- sion plans to establish maximum capacities of competing power plants based on their long-term economic performances. When economic factors are mostly favorable, a plant size becomes limited by its power system operational reliability requirements. Among these requirements, the bus voltage limits are often the ones that become active. These limits provide information on the plant reactive power supply requirements under different loadings and grid configurations. Since doubly fed induction generator (DFIG) based WFs have no difficulty in providing their own supply of reactive power, in this case, it is assumed that the plant capacity could be limited by the system frequency dynamics. As such, the sizing prob- lem is examined here only from a system frequency dynamics perspective. B. WF Output Power Components Fig. 1 shows P W (t), a typical WF output profile, and its components P a (t), P t (t), and P r (t) satisfying P W (t)= P a (t)+ P t (t)+ P r (t). (1) 0885-8969/$26.00 © 2009 IEEE