The effect of shaft modelling on the assessment of fault CCT and the power quality of a wind farm

The effect of shaft modelling on the assessment of fault CCT and the power quality of a wind farm
of 5
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
    99 The Effect of Shaft Modelling on the Assessment of fault CCT and the Power Quality of a Wind Farm S.K.Salman, enior Member, I A.L.J.Teo, AMIEE, I.M.Rida, AMIEE School of Eledronic and Electrical Engineering he Rob ordon University Schoolhill, Aberdeen, AB10 FR, Scotland, UK Abstract: Global trend of electricity market deregulation has ed to an increasing numbex of wind farms integrated into utilities’ s network at distri ution level. It has been reported that the incorporation of wind farms mta dishibution networks gives rise to severe voltage recovery problem wt regard to Wind Power Based Embedded Generator WBPEG) following a fault condition on the Bssociated network. This paper presents he rdt of an investigation aiming t detemhing the effect of including the windmill shaft on voltage recovery following a fault on the network It is conducted on a simulated system using EMTP program. This work has shown that the Critical Clearing Time CCT) that defines the stability limit of WBPEG c n be much lower when he windmill shaft is included. It has also shown that Embedded Generator EG) subjected to network fault c n have an aggravated effect on distribution network power quality, which has not previously been identified when a windmill lump model is used In order to improve voltage recovery at the terminals of a windmill and consequently its stability margin reactive power injection technique using Static Shunt Capacitor SSC) has beenPpo4. Keywords Wind farms, voltage recovery, Embedded Generation stabiiity of induction generator, Critical Clearing Time, power quality, Reactive power compensation, EMTP. I. INTRODUCTION The global trend to deregulate electricity markets to achieve economic operation and to also meet the increasing environmental concern has in recent years led to the increase of the number of wind farms integrated into utilities’ network at distribution level which are commonly known as embedded generation EG)). Wind power is considered the most pollution-free and commercially available renewable energy technology for electricity generation. It has been predicted that the annual growth of wind power between 1998 and 2040 would be between 20 30 [l] Consequently, it is expected wind power would supply 20 of the world’s electricity by 2 4 111. Various windmill models have been developed [2 71 for the purpose of investigating the effect of wind farms integration on the stability and power quality of the associated network. The model should include sufficient details to describe as closely as possible the behaviour of a physical windmill system but at the Same time should not be too complex to implement. 0- 803-6499-6/CO/ 10.00 2 5 I S An example of such model is the empirical dynamic windmill model, which is based on a physical windmill system that was reported in [MI his model has been used to investigate the effect of incorporating the windmill shaft in wind Wie induction generator WTIG) model on the voltage re overy following a fault condition at the terminals o the wind fam On the other hand it has recently been reported [8] that a fault on the network to which a wind farm is connected gives rise to severe voltage recovery problem with regard to the wind farm. However in this latter work the effect of modelling the sh ft sepaxately was not included. Instead it was focused on determining the Critical Clearing Time CO hat defines the stability limit of wind fam using lumped model, i. e., the masses of wind turbine, gear box and generator are assumed rigidly connected together. In this context it has been reported in [9] that voltage recovery due to three-phase fault on associated network may be improved using reactive injection technique. This may be achieved using static shunt capacitors SSCS) [lo] The work reported in this paper is concerned with examining the effect of including the shaft model and reactive power injections on the Critical Clearing Time and also on the power quality of the output power of wind fm ollowing fault conditions on the associated network. 11. THE INVESTIGATED SYSTEM The system used as a basis for this investigation is the same as that used in [7].  ig 1  shows that it basically consists of a wind farm qnnected to a distribution network. The latter is connected to the grid, which is represented by a constant voltage source connected in series with the Thevenin’ impedance related to the grid. The system is simulated using EMTP. n Substation 33kV ine a- 4 Fault 2 x 6 kW 4 x 6 kW 1 x 6OOkW Load1 Load2 Fig. 1. Schematic diagram of the considered network  simulation of this work are only difference between the of the windmill mechanical part is done. In tlus work the mechanical part of the windmill is considered to consist of wind turbine, gearbox, high-speed low-speed shaft, and the rotor of electrical generator. ver in c se of [7] ll the above mentioned mechanical parts of the considered windmill are modelled as a lumped element, i.e., one mass representing all of the masses. Therefore each of the windmill mechanical parts in this work, gearbox and generator rotor is the corresponding models are er. The details of modelling consideration are discussed n the next section. In MODELLING CONSIDERATION Practical systems are, in most cases considered as discrete system with finite number of degree of freedom in order to obtain solution in a simpler manner. In reality, some systems, especially those involving continuous elastic members, e.g., shaft in a windmill, have-an infinite number of degree of freedom [ 11 Treating a system as continuous system results in an accurate solution. However, due to the limitations of available analysis methods, most practical systems are considered by treating them as finite lump component [ 11 . This is the assumption normally adopted in conventional windmill modelling. This means that all of the mechanical parts of a windmill such as wind ine, shaft, gearbox and generator rotor are reated as a lump element. In general, more accurate results are obtained by increasing the masses springs, and damper, which are used to rep physical characteristics of the actual windmill system, i.e., increasing the degree of freedom. For example it has been suggested [12] that WTG drive train should have at least two degrees of freedom, i.e., two inertia, to represent the torsional and the electrical modes. Recently it has been suggested [SI that for the purpose of stability studies of a 'bindmill system subjected to faults, the complete physical windmill model, except for the block corresponding to the mechanical eigenswings should be used. The reason for excluding eigenswings when such studies are considered is because they do not affect the stability. On the other hand it has been advised [6] that in c se it is not possible to obtain reliable data for the mechanical system under consideration, the safest approach to be followed is to apply a typical data rather than using a traditional lump dynamic model. Normally the sh ft has finite stiffness. In this context the torsional eigenfrequency ( used in this work is assumed equal to 1.67Hz [5 6]. On the other hand the total inertia constant HT~ f the mechanical part of the 600 kW windmill used in this investigation is 3.7s. This is based on its corresponding manufacturing data sheet. The gearbox also contributes to the shaft stiffness between a windmill and a generator [SI. However it is assumed that the distance between the wind turbine and the gearbox is small and consequently the wind turbine and the gearbox are replaced by a single rotating mass with a moment of inertia equal to the sum of those related to the wind turbine and the gearbox. The inertia constant of the combined turbine and gearbox masses is found equal to 3.2s and that related the generator rotor is 0.5s 95 shaft stiffness can hen be calculated by using the values of the eigenfrquency of the sh ft torsional vibration and inertia constants. Therefore the resultant mech nic l part of the windmill is modelled in this work for the purpose of stability study as two masses connected by a shaft. Using the EMIT software package, wind turbine and generator rotaling masses c n be represented by two capacitors ~o~e~ted o ground. The shaft stiffness on the other hand can be represented by the reciprocal of an inductance connecting the two Capacitors representing the wind turbine and generator rotating masses. In contrast the lump model of the 6 kW windmill system considered in this work is based on a single rotating mass with inertia constant equal to 3.7s. IV EFFECT OF THE SHAFT MODEL ON THE CCT 'Fhe effect of the way the mechanical side of the windmill system is modelled in terms whether using lumped model or a shaft model on the CCT elated to the considered wind f rm has been investigated. s mentioned previously, the lumped model is related to the c se whereby the mechanical part of each windmill system constituting he wind f rm is considered as a single rotating mass. The shaft model, on the other hand, is referred to the c se whereby the mechanical part is considered as two rotating masses connected by a shaft. The investigation is conducted by the application of a 65ms duration three-phase fault on the load feeder at bus 2 see Fig. 1). The corresponding voltage variation at the terminals of the wind fanns, i.e., bus 4 for both models versus time are shown in Fig. 2. It is evident from this figure that the variation of the terminal voltage of the wind fasm for the lump model mover its pre-fault voltage magnitude faster than that related to the shaft model. Fig 3  shows the variation of generator rotor speed versus time for the same fault condition for both models. This figure again shows that the generator speed using lump model regains its pre-fault speed value faster than related to the S model. It c n also be observed from the same figure that the generator rotor speed due to shaft model is subjected to relatively high oscillations while that related to the lumped model recovers its steady state value with almost no oscillation. 1 q Lump Model Shaft Model ;* 0.4 - 0.2 - 8 - 0 6 - I - 0 2.5 3 0.5 1 1.5 Time sec) Fig 2 Variation of wind farm tenninal voltage at bus 4 following a 6Sms three ohase network fault  - 996 1.06 1 os 1.04 1.03 1.02 1 1.01 1 0 99 0 98 ' I 0 0.5 1 1.5 2 2.5 3 lim (sec) Fig. 3 VUi.licm of generaor rotor speed following a 6Jms three-phase network fault The same study is then repeated for the same fault condition described above but with longer fault duration. The stimulation results related to voltage behaviour for a fault duration of 105ms are depicted in Fig. 4.  It c n be seen from the same figure hat the voltage at the terminals of the wind fm sing the lump model regains ts pre-fault terminal voltage value while that due to shaft model shows voltage collapse as well as violate oscillations during the post-fault oeriod. 1 0.8 - 3 o.6 c 0 4 0.2 Shaft Model 0.5 1 1.5 2 2.5 3 llme sec) Fig. 4 Variation of wind farm tmni~l ohage at bus 4 following a 1OJms three pbe network faul 1.34 7 I 1.29 1.24 4 1.19 1 .14 L 1.09 ,o 1.04 0.99 0.94 0 0.5 1 1.5 2 2.5 3 llme (sec) Fig. 5. Variation of generator rotor speed following a 105ms three phase network fault As far as the speed of the rotor of the generator is concemed lump model results in the speed regaining its pre-fault value gradually after the clearance of the fault whereas the same speed resulted from using the shaff model shows cohtinuous increase in its value with large oscillations which means the generator becomes unstable his s illustmted in  Fig. 5.  In order to determine the Critical Clearing Time (CCT) elated to the wind farm under consideration the study described above is repeated with fault duration being gradually increased until the generator terminal voltage in each model c se fail to regain its pre-fault value. The time at which the wind f nn fails to regain its pre-fault nominal value is considered equal to the CCT For the system under consideration it has been found that the CCT for the lump model is 240 ms, while that related to using the shaft model is 101 ms. Therefore it c n be concluded from simulation results that it is impomnt to include modelling of the shaft between wind Wigear box and the generator otherwise large error is in when the CCT is determined. Obviously this is important when the time settings of relays that are nstalled on feeders emanating hm the substation to which the wind farm is connected, is considered. V EFFECT OF SHAFT MODELLING ON THE POWER QUALITY SUPPLIED BY WIND FARM It is now established [8] that due to the random nature of wind speed which constitutes the input of wind farm the corresponding electrical power output of a wind farm normally follows the variation pattern of speed. In order to establish the effect of shaft modelling the developed simulation of the system shown in Fig. 1  is used to detennine the output power due to a randomly generated mechanical input power signal representing the wind speed variation once using the lumped model and then using the shaft model. Fig. 6 shows the randomly generated mechanical signal used for t is c se study while Fig. 7 shows the corresponding output power of the considered wind farm due to lump and shaft models. his figure shows that the power output due to lump model is slightly higher than that obtained using the shaft model. his may be explained that some energy could have been lost due to the torsional torque in c se of the shaft model. I I 0 1 2 3 4 5 6 llme sec) Fig. 6 Randomly generated mechanical input pow- to he wind farm.  2.0 IS - 1 t k 5- ... 0 1 2 3 4 5 6 Rme sec) -- Fig. 7. utput power of the considered wind fann due to lump and shaft models. As far as the power quality is concemed, this shows that both models produce the same power pattem to a certain input signal i.e. power quality at steady state is not affected by the type of model used This nvestigation has been extended by examining the effect of including the shaft model on the quality of output power of the wind farm following fault condition on the associated network. Fig.8. shows the behaviour of the output power of the wind farm following a three-phase fault on load feeder at bus 2 see Fig 1) with 100 ms fault duration, i.e., the fault duration is less th n the CCT Corresponding to both models. This figure illustrates that the inclusion of the shaft model has clear effect on the power quality of he wind farm compared with that resulted from using lump model. 20 I 1 2 3 4 5 ne sec) Fig. 8. The behaviour of output power of the wind farm following a lSOms three-phase fauh at bus 4. VI IMPROVING VOLTAGE RECOVERY USING VAR INJECTION TECHNIQUE In order to improve voltage recovery prospects following fault conditions and consequently improve the CCT of the wind farm, it has been decided to examine the application of var injection technique using static shunt capacitors SSCs) to achieve this objective. These devices have been chosen due to economical consideration 61. It should be pointed out that the SSC banks are applied in addition to the capacitors srcir@lg 7 installed at the termids of the aspcbnous generators o? each windmills for power factor correction purposes. These power factor correction capacitors are designed to supply 138 KVAR to the 600 kW wind turbiie generator used in t is investigation. However, asynchronous generators continue to draw reactive power even after installing the power factor correction capacitors. For example for the c se unk consideration each of the 600 kW generator draws further 153 KVAR under full load condition. It has been decided to consider installing SSC that is capable to increase the CCT from 100 ms for shaft model) to 150 ms n order to achieve this it has been found that such capacitor should have the capability of providing a reactive power of 36 W R during post-fault period. The corresponding capacitor value is 8 mF on 690 V, i.e., at the wind farm tenninals, while its value is only 35 08 pF on the 11 kV side. It assumed that he capacitor is switched on to the system only when a fault is detected and after 150 ms after fault inception. This ime is allowed for the switching operation to take place. The effect of the location of such capacitors is lso considered. It has been decided to consider installing them in turn t the buses 2, 3 and 4. Fig. 9 shows the resulting behaviour of the voltage at the termin ls of the wind farm following a 150 ms. duration three phase fault. It c n be seen that installing SSC at bus 3 or 4 could help the recovery of the voltage at the tenninals of wind farm during post-fault period. However this s not the c se when the SSC is installed at bus 2 where the it c n be seen that the voltage cannot recover. Fig. 9, also shows that the best result is obtained when the SSC s installed t bus 4 where the voltage recover without having large dips compared with that obtained from having he SSC instailed at bus 3. 1.4 , I 1.2 g1 d 0 8 5, 0.6 0.4 0 2 OJ I 0 1 2 3 4 5 %me (sec) Fig. 9. The resulting behaviour of the voltage at th terminals of the wind farm at bus 4 following a SO ms duration he hase fauh due to the effect of compensation capacitors ocated at the buses 2,3 and 4. VII. CONCLUSION A distribution system with integrated wind farm has been simulated using the electromagentic transient program EMTP). The wind farm is assumed to consist of sixteen windmills. The effect of the way the mechanical part of a windmill is modelled on the CCT of the wind farm and the quality of the output power of the same has been considered. In this investigation two types of models have been considered. In first model tvpe the lump model) all rotating;  -998- masses i. e., those related to wind turbine, gearbox and generators, are considered rigidly connected together. In the second model type the shaft model), however, the mechanical part of a windmill is considered as consisting of two rotating masses connected by a shaft. One of the masses s due to the wind turbindgearbox while the second mass is related to the generator rotor. The results obtained from this investigation show that the inclusion of the sh ft in the modelling of the mechanical part leads to number of differences. These include; i) the behaviour of the voltage at the terminals of the wind farm as well as he rotating speed of windmills following a three-phase fault condition on the network behave differently for the lump and shaft models, ii) the critical clearing time resulted from using the sh ft model is considerably less than hat due to the lump model, iii) the quality of the output power from wind farm at steady state condition is not affected by the type of model used However following three-phase fault condition the inclusion of the shaft model shows deterioration in the power quality immediately after the fault and finally this investigation has shown that (iv) var injection using SSC can improve both the CCT and the power quality. The optimal improvement is achieved when the SSC is installed at the tenninals of the wind farm. VIII. ACKNOWLEDGEMENT The authors would like to th nk the Robert Gordon University for provision of facilities. A.L.J.Teo is gratehl to the Robert Gordon University for their financial support to undertake this IFSeXCh. IX. EERENCES l] Thc ~mpcan Wind energy Association (EWEA). “EWEA Publications, 1998,” [WWW] hnp:// 121 AI. Eatanqueiro RM.G.C stro. and J.AG. Saraiva “The used of Stiff and Flexible Rdar Models with Re@ to Wind Turbines Power Quality Assessment,” BWEA ’16 16 Wind Energy Conf., Stirling, UK WC. 1994.p~. 313 312. [3] F.AG.S. Rcis RM.G.Castro Al.LEdanqueiro and J.M. Fmcira de Jesus. “Including a Wind Energy Conversion System Model in ElcNomagnetic Transient Program,” IPST’95 lnt Conf. on Powcr Systems Trnnsienls, Lisbon, Sept. 3 7.1995, pp. 243 247. E. Welfondcr, R. Neifer. and M. Spanner. “Development and Expenmental Identification of Dynamic Models for Wind Turbines and heir Fluctuating Power Generation,” SIF’OWER ’95 Proceedings Vol. Bom the lFAC Syposium, Dec. 6 8.1995. pp. 73 81. V. AkJunatov, H. Khudsen and AH. Niclmn, “Advanced Simulation of Windmill in the Electric Power Supply,” Electrical Power Energy Systems 2000. JEPE 373. [WWW] V. Akhmatov and H. Khudscn “Modelling of Windmill Induction enerators in Dynamic Simulation pprns,” IEEE Power Tech. ’99 Conf. Budapest Hungary, Aug. 29 Sept. 2,1999. BPT99-243-12. S.KSalman and I.M. i4 Impact of ntegration of Wind farms into a Utiliy Network on Relay Setting of he Utility Feeders,” ERA Int Cod Exhit. Quality Security of Supply in Electrical Network. [4] [SI [6] [q London 16 17 Feb., 1999, pp. 2.3.1-8. [8] Salman S K and Rida 1, “Investigating the Electrical integration of wind power based embedded generation into utilities’ network”, Roc. IASTED lnt. Conf., 3“ IASTED Id. cod an Power Energy System Paper No. 01-058.8 -10 November. 1999. Las Vegas USA 191 M. runU. J. Havsager and H. Knudscn Incorporatian of Wid Power in the East Danish Power Syslcm ” lEEE Power Tech ‘99 Conf Budapest, Hungary. Aug. 29 Sept. .1999. BPT99-202-5. [lo] L Mano and N. D. Hatziargyrioy “AlliatiOn of Shunt Capadof Banks as a Countermeasure for Voltage Collapse ” Roccedinga of he 31“ UPEC’96 Cod raklio. Greece, 18 20 Sept 1996. pp. 384-386. S.S. Rao. Mechanical Vibration, Third Edition Addiaion Walcy. 1995, pp. 13. [12] E.N. Hinrinchsen and P.J. Noklan, “Dynamic and Smability of Wind 101 no. 8. Aug 1982. pp. 2640 2648. [13] D. StOjanovic and N. Pclrovic and N. Mitmvic, “Analysis of Tmirmal Torques of Big TurbineGenerator Shafts ” IPST ‘99 Int. National Cod. on Power System Transient, Budapest, Hungary June. 20 24. 1999. Ill] Turbine G~nerat~r~,” EEE TYMS Ch POW- App. and SF Vol. PAS- X BlOGRAPHlES Salmm K Salman: He is a mmk f the IEE and a senior member of the IEEE. was born in Basrahq Iraq 1941. He gradu d hm hc Faculty of Engineering, University of Baghdad in 1964. In 1972 he obtained his MEng in Eledrial and Electronic Engineering from the University of Sheffield. UK and in 1975 he completed his PhD at UMIST, UK In 1975 he joint the Univaaity of Benghrzi, Libya as a lechuer. assistant profam and associa~e professor. His practical eqerien includca testing and commissioning of measuring devices, protedion relays and control Circuits related to 132 kVB3 kV/ll kV systems. He hrs published several papers on voltage control of distribution networks with embedded generation and on the protection of power systems at Mtiond and intcmational conferences and scientificjourmla Heisalso a CO LUlhOC of a book titled “Digital Protection for Power Systems” wbich was published by the IEE in September 1995. Cumntly he is a Rader at the School of Electronic and Electrical Engmeering at The Robcrt Gordon Univcnity, UK Anita L. J. Teo (AMlEE-99) was born in Sabah Eavt Malaysia on October 29.1974. She recnvcd her BEng (Hons) in Electronic and Electrical Engineering Bom he Robcrt Gordon University, Aberdeen, UK in 1999. Cumntly she is studying towards hcr M.Phil PhD. degree at the Robert Gordon University. Ha esearch interests are in the arras of wind power based embedded generation, flexible ac transmission system (FACTS), and voltage control of power system lbrnhlm M. Rldn (AMlEE-93) was born in Ammur, Jordan on July 23. 1960. He received his B.Sc. in ElcNical Engineering om Zaporozhye Machine Building Inslie Wne. in 198s and his M.Sc. in Elactrical Enginming om he Univaaity of Manchester Institute of Science and Technology (UMIST) n 1996. Ibrahim Rida has worked since 1988 a restarch engineer for the National Electric Power Company in Jordan in protection and power system analysis. Currently, he is studying towards a Ph.D. egree a the Roberl Gordon University in Abdeen, UK. His research interests are in the areas of wind power based embedded generation, voltage control and protection of power systems.
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks