24 IEEE Innovative Smart Grid Technologies - Asia (ISGT ASIA) Voltage Profile and Loss Assessment of Distribution Systems ith Fixed Speed Wind Generators M. H. Haque School of Engineering University of South Australia Mason Lakes 595, Australia Abstract Wind generators are increasingly being integrated into high voltage transmission systems as ell as lo or medium voltage distribution systems. The increased penetration of ind generators into poer grid may significantly change the system voltage profile, losses and other operating characteristics. This paper investigates the effects of ind generators on voltage profile and losses of a distribution system. To make the investigation more realistic, time varying system load and historical ind speed data are used. Randomly generated ind speeds through Weibull probability density function are also used. For a given ind speed, the turbine poer is determined through a polynomial obtained from poer data supplied by the manufacturer. The system voltage profile and losses are obtained through repetitive poer flo solutions ith varying load and ind speed data. The exact equivalent circuit of a ind generator is directly incorporated in poer flo calculations. The above technique is then applied to to distribution systems consisting of 33 and 69 buses ith a number of embedded ind generators. The results obtained are carefully analyzed and discussed. Index Terms distributed generation, distributed energy resources, distribution systems, reneable energy systems, system losses, voltage profile, ind generators. I. INTRODUCTION Wind poer is the fastest groing reneable energy source in the orld. A large number of ind turbines or distributed generators have been installed orldide over the past decade []. The increased penetration of distributed generation into poer grid introduces a number of technical and economical challenges that are ell studied in [2, 3]. A ind generating system can be classified into to main categories: fixed speed and variable speed [4, 5]. A fixed speed ind generating system employs a squirrel-cage induction generator, hich is directly connected to the grid through a step-up transformer. A soft starter and a shunt capacitor are also used for smoother connection and reactive poer support. On the other hand, a variable speed ind generating system employs either a doubly fed induction generator ith partial-size converters or a permanent magnet synchronous generator ith full size converters. The ind poer varies cubically ith ind speed. A ind turbine is required to shed poer at higher ind speeds to protect various components of the system. Based on poer shedding techniques, a ind turbine can be classified into pitch controlled and stall controlled [6, 7]. A pitch controlled ind turbine has an adjustable blade angle to shed poer at higher ind speeds. On the other hand, a stall controlled ind turbine has a fixed blade angle but the blades are carefully designed to reduce aerodynamic efficiency at higher ind speeds. Large size ind farms are usually connected to high voltage transmission systems but small size ind farms are normally connected to lo or medium voltage distribution systems. In both cases, the ind farms inject poer into the grid and that cause a change in some operating characteristics including voltage profile and losses. The voltages and losses are usually determined through poer flo solutions. This paper investigates the effects of ind generators on voltage profile and losses of distribution systems. Most of the distribution systems are designed ith a single feeding substation and the structure of the netork is mainly radial. For such a system, poer flos in one direction. Hoever, integration of ind generators into a distribution system may cause to flo poer in either direction. In fact, higher penetration of ind poer may transform a passive distribution netork into an active netork feeding poer into the high voltage system. The demand of a distribution system is not constant but changes ith time. In addition, integration of ind generators adds more uncertainty because of intermittent and sometimes stochastic nature of ind poer. In order to analyze such a system, repetitive poer flo solutions ith varying demand and ind poer are required. This paper investigates the variation of voltage profile and losses of to distributed systems caused by embedded ind generators. The customer loads are modulated to obtained time varying demand. The historical ind speed data is used to obtained turbine poer through manufacturers supplied data. The time varying demand and turbine poer are then used in evaluating the voltage profile and losses of the systems through repetitive poer flo solutions. 978--4799-3-8/4/$3. 24 IEEE 2
24 IEEE Innovative Smart Grid Technologies - Asia (ISGT ASIA) II. TURBINE POWER A ind turbine (WT) converts the kinetic energy in moving air into rotational mechanical energy, hich is then converted to electrical energy using a generator. The mechanical poer P m extracted by a ind turbine can be ritten as [7] 3 P m =.5ρAV Cp( λ, β) () Here ρ is air density, A is turbine blade sept area, V is ind speed, and C p is performance coefficient that depends on tipspeed-ratio λ and blade pitch angle β. The ind speed is not constant but a random variable and can be obtained from field measurements. Alternatively, the ind speed can be modeled by Weibull probability density function (PDF) hich can be expressed as [7] k k V f ( V) = c c k V exp c Here k is called shape parameter and c is called scale parameter. For a given WT, P m depends on ind speed and turbine performance coefficient. Fortunately, most of the manufacturers provide the poer data of the turbines [8, 9]. Table-I: Poer data of GWL 225-kW ind turbine (V in m/s and P m in kw) V P m V P m V P m V P m 4 9 49.6 6 234 22 2 5 9.5 7.4 7 23 23 2 6 32.8 2 89.3 8 229 24 2 7 58.8 3 26 9 225 25 2 8 87.6 4 27 2 25 9 8.7 5 23 2 22 25 2 (2) turbine poer in active operating region (beteen V in and V out ) is estimated by the folloing fifth order polynomial. 2 3 4 5 P = a + av + a2v + a3v + a4v + a5v ; V V V (3) m The coefficients of the polynomial can be obtained through polyfit routine given in MATLAB. Figure shos a comparison of estimated poer obtained through (3) ith the corresponding actual values given in Table-I and is observed to be in very good agreement. The maximum error beteen the actual and estimated values is found as 3.82 kw and that occurred at a ind speed of m/s. Thus, the turbine poer P m can be ritten as P m ; 5 = akv k= ; k V ; V in V < V V in > V out V Note that the turbine poer is zero hen the ind speed is belo the cut-in value or above the cut-out value [7-]. III. GENERATOR MODEL As mentioned, squirrel-cage induction generators (SCIG) are commonly used in fixed speed ind poer applications. The per-phase equivalent circuit of a SCIG is shon in Fig. 2 here R, R 2, X, X 2, R c and X m represent stator resistance, rotor resistance, stator leakage reactance, rotor leakage reactance, core loss resistance and magnetizing reactance, respectively [3]. Note that, for generator operation, the slip s is negative and thus the rightmost resistance R 2 (-s)/s is also negative. In other ords, the resistance delivers a poer of P m hich is obtained from the WT. out R jx R 2 jx 2 in out (4) Wind poer, kw 5 (P g+jq g) R c jx m R 2 (-s)/s P m 5 5 5 2 25 3 Wind speed, m/s Fig. Comparison of estimate and actual poer of a WT estimated value; o actual value Table-I shos the poer at various ind speeds of a stall controlled ind turbine (GWL 225-kW) and are obtained from []. The cut-in ind speed (V in ) and cut-out ind speed (V out ) of the turbine are specified as 4 m/s and 25 m/s, respectively. Some of the previous studies represent the poer (beteen cut-in and rated ind speeds) by a linear function [] or a quadratic function [2]. In this study, the Fig. 2 Per-phase equivalent circuit of a SCIG (P g+jq g) R jx m R 2 jx 2 Fig. 3 Single line diagram of a SCIG R c jx m (-P m+j) In poer flo analysis, a ind generator is normally represented either by a P-Q model or an R-X model [4-7]. The above models are obtained from the equivalent circuit ith some calculations and/or approximations. Hoever, in this study, the equivalent circuit of the generator is directly used in poer flo calculations. Without loss of generality, r 2
24 IEEE Innovative Smart Grid Technologies - Asia (ISGT ASIA) the circuit of Fig. 2 can be represented by a single line diagram as shon in Fig. 3 here the rightmost resistance is replaced by a negative load (-P m + j). The above model (Fig. 3) can easily be incorporated into any poer flo program by simply augmenting the netork by to buses (m and r), to series elements (R +jx and R 2 +jx 2 ), to shunt elements (R c and jx m ) and a load (-P m +j). Such a model is successfully used in [8] to evaluate the poer flo solutions. For a given ind speed V, the turbine poer P m can be obtained from (4) and the corresponding complex poer (P g +jq g ) delivered by the generator can be determined from poer flo solutions. Note that a SCIG alays absorbs reactive poer and thus Q g in Fig. 3 is negative, hich is the sum of reactive poer absorbed by X, X 2 and X m. In this study, the ind generator model of Fig. 3 is used in poer flo calculations. IV. POWER FLOW CALCULATIONS As mentioned, most of the distribution systems have a single feeding substation and the structure of the netork is radial. The above properties of a distribution system are fully exploited in deriving a set of backard and forard recursive equations to solve the poer flo problem. The backard recursive equations are used to determine the poer flo through each branch and the forard recursive equations are used to determine the voltage of each bus. Incorporation of ind generator model of Fig. 3 increases the size of the netork by to buses hile preserving the radial configuration. Thus, integration of ind generators does not change the property of the distribution netork and the same poer flo solution technique can be applied. The detail of poer flo solutions using recursive equations is described in a number of articles [9-22] and thus not given in this paper. V. RESULTS AND DISCUSSIONS The effect of ind generators on system voltage profile and losses is investigated on to distribution systems consisting of 33 and 69 buses. In both system, a number of identical ind turbine (GWL 225-kW) and SCIG (69-V, 225-kW,.9-pf) sets are embedded into the netork though step-up transformers and shunt capacitors. The parameters of the generator are considered as R =.4 pu, R 2 =.8 pu, R c = 5 pu, X =.48 pu, X 2 =.48 pu, and X m = 4. pu. The leakage reactance of the transformer is considered as 8.%. The kvar rating of shunt capacitor is assumed as 3% of the generator kva rating. Because of the lack of time varying load data, it is considered that the hourly load variation of all buses follos the same pattern as described in [23]. The ind speed of all turbines is also considered to be the same as the measured data of a particular site in Australia and is obtained from [24]. The yearly ind speed variation and its discrete histogram are shon in Fig. 4. With time varying demand and historical ind speed data, the results obtained in the above systems are briefly described in the folloing. Hours per year Wins speed, m/s 3 2 2 3 4 5 6 7 8 9 Time, hours 5 5-5 5 5 2 25 Wind speed, m/s Fig. 4 Yearly ind speed variation and discrete histogram A. The 33-Bus System The 2.66-kV, 33-bus distribution system has an average demand of (475+j23) kva. The data of the system is given in [9]. The daily variation of active and reactive poer demand of the system is shon in Fig. 5. Telve ind generators are added to the system at six different buses as shon in Fig. 6. Table-II summarizes the number of generators connected to each bus. Active/reactive poer, kw/kvar 7 6 5 4 3 2 Active Reactive 2 4 6 8 2 4 6 8 2 22 24 Time, hour Fig. 5 Daily variation of active and reactive poer of the 33-bus system First, the poer flo of the original system (ithout ind generators) is evaluated for various operating hours and the corresponding voltage profile of the system is shon in Fig. 7 hich indicates that bus 32 has the loest voltage for all cases (operating hours). The minimum and the maximum losses in the system are found as 24.2 kw and 7.2 kw, respectively. The daily average loss is found as 4.2 kw and that corresponds to a daily energy loss of 9.84 MWh. 22
24 IEEE Innovative Smart Grid Technologies - Asia (ISGT ASIA) 22 25 2 3 4 5 6 function of (2) ith k = 2 and c = 9.27, and that corresponds to an average ind speed of 8 m/s [7]. The poer flo of the system is then evaluated for, sets of random ind speeds. The distribution of ind poer injected into the system is shon in Fig. 9 and it indicates that the ind poer varies beteen 22.3 kw and 2624.7 kw ith an average value of 96.7 kw. In this case, the range of voltage variation of all buses is found to be very similar to that of Fig. 8. The average system loss for this case is found as 296. kw. 24 23 26 7 28 27 29 3 3 32 8 9 2 3 4 5 6 7 ~ 8 2 9 2.2 Fig. 6 Single line diagram of the 33-bus system ith ind generators.98.96 Table-II: Number of generators at various buses of the 33-bus system 2 7 2 2 2 24 3 29 3 The poer flo of the system is then evaluated ith ind generators for a period of one year (365 24 = 876 hours) and the corresponding voltage profile is shon in Fig. 8. Comparison of Figs. 7 and 8 indicates that the ind generators are unable to improve the minimum bus voltage significantly because the turbines remain ideal for a number of hours (94 hours) due to very lo ind speed. Hoever, the ind generators significantly reduce the number of hours the system operates belo a specified voltage. For example, ithout ind generators, bus 32 has a voltage of less than.85 pu for a period of 9 hours per day. With ind generators, the bus has such a lo voltage on average of only 3.45 hours per day. The daily average loss in the system is also reduced from 4.2 kw to 299.9 kw and that corresponds to an average daily energy saving of 2.65 MWh. 7.94.92.9.88.86.84.82 5 5 2 25 3 35 Fig. 8 Yearly voltage profile of the 33-bus system ith ind generators 3 25 2 Wind poer, kw Number of generators.2 5.98 5.96.94.92 2 3 4 5 6 Time, hour 7 8 9 Fig. 9 Distribution of ind poer for, sets of random ind speeds.9.88.86.84.82 5 5 2 25 3 35 Fig. 7 Hourly voltage profile of the 33-bus system ithout ind generator Finally the hourly ind speeds of all six sites are randomly generated through Weibull probability density B. The 69-Bus Systems The 2.66-kV, 69-bus distribution system has a main feeder and seven sub-feeders. The data of the system is given in [25]. The average demand of the system is (7.9+j897.9) kva. The hourly voltage profile of the system is shon in Fig. and it indicates that the system has a minimum voltage of.8799 pu and that occurred at bus 53. The average loss in the system is found as 76.6 kw. Ten identical ind generators are then added to the system at ten different buses as shon in Fig.. The total 23
24 IEEE Innovative Smart Grid Technologies - Asia (ISGT ASIA) capacity of ind generators is 225 kw and is much higher than the average system demand of 7 kw. Thus, the system may become active at higher ind speeds. The yearly voltage profile of the system is shon in Fig. 2 and it indicates that the voltage of some buses increases ell above the root bus or feeding substation voltage of. pu and that occurred mainly at higher penetration of ind poer. It is found that 3.6% of voltages are above the root bus voltage...5.95.2.9.98.85.96 2 3 4 5 6 7 Fig. 2 Yearly voltage profile of the 69-bus system ith ind generators.94.92 2.9 5.86 2 3 4 5 6 Active poer dran, kw.88 7 Fig. Hourly voltage profile of the 69-bus system ithout ind generator 27 28 29 3 3 32 33 34 5-5 - 58 59 6 6 62 63 64 65 66 67 68-5 39 4 2 3 4 5 Time in hour 6 7 8 9 Fig. 3 Distribution of active poer dran from feeding substation 2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9 2 2 22 23 24 25 26 35 56 VI. 57 36 37 38 54 55 4 42 43 44 45 46 47 48 49 5 5 52 53 Fig. Single line diagram of the 69-bus system ith ind generators The average system loss is found as 6.4 kw. Figure 3 shos the distribution of active poer dran by the system at feeding substation. The negative value indicates that the system feeds poer into the grid or the system become active. It can be noticed in Fig. 3 that 34.9% of the time the system feeds poer into the grid. CONCLUSIONS Integration of ind poer into a distribution system may change various operating characteristics of the system, especially at increased penetration of ind poer. This paper investigated the effects of embedded ind generators on voltage profile and losses of to distribution systems consisting of 33 and 69 buses. Time varying system load and historical ind data are used in evaluating the voltage profile and losses through repetitive poer flo solutions. The equivalent circuit of the generator is directly incorporated in poer flo calculations. In the 33-bus system it as found that the ind generators have little effect on the overall system minimum voltage but the number of hours the voltage remains ithin a certain level may change significantly. The ind generators reduce the average system losses and that cause significant annual energy saving. The results are highly dependent on the value of the capacitors used at the generator terminals and the degree of penetration of ind poer. In the 24
24 IEEE Innovative Smart Grid Technologies - Asia (ISGT ASIA) 69-bus system, the voltage of some buses is found to increase significantly because of higher degree of penetration of ind poer and that occurred mainly hen the system becomes active and feeding poer into the grid. REFERENCES [] The Wind Poer: http://.theindpoer.net/index.php. [2] Working Group on Distributed Generation Integration, Summary of distributed resources impact on poer delivery systems, IEEE Trans. on Poer Delivery, Vol. 23, No. 3, 28, pp. 636-644. [3] J. P. Lopes, N. Hatziargyriou, J. Mutale, P. Djapic and N. Jenkins, Integrating distributed generation into electric poer systems: A revie of drivers, challenges and opportunities, Electric Poer Systems Research, Vol. 77, No. 9, 27, pp. 89-23. [4] H. Li and Z. Chen, Overvie of different ind generator systems and their comparisons, IET Reneable Poer Generation, Vol. 2, No. 2, 28, pp. 23-38. [5] IEEE PES Wind Plant Collector System Design Working Group, Characteristics of ind turbine generators for ind poer plants, Proc. 29 IEEE Poer and Energy Society General Meeting, Calgary, Canada, 29. [6] E. Muljadi and C. P. Butterfield, Pitch-controlled variable-speed ind turbine generation, NREL Report No. NREL/CP-5-2743, 2. [7] G. M. Masters, Reneable and efficient electric poer systems, Wiley Inter science, 24. [8] Vestas: http://.vestas.com/en/media/brochures.aspx#!. [9] Siemens: http://.energy.siemens.com/mx/en/reneableenergy/ind-poer/platforms/. [] Freebreeze: http://.freebreeze.com/ind-turbines/gl-225kind-turbine.html. [] W. C. B. Vicente and R. C. N. Hadjsaid, Probabilistic load flo for voltage assessment in radial systems ith ind poer, Electrical Poer and Energy Systems, Vol. 4, 22, pp. 27-33. [2] P. Siano and G. Mokryani, Probabilistic assessment of the impact of ind energy integration into distribution netorks, IEEE Trans. on Poer Systems, Vol. 28, No. 4, 23, pp 429-427. [3] M. G. Simoes and F. A. Farret, Reneable energy systems Design and analysis ith induction generators, CRC 24. [4] A. E. Feijoo and J. Cidras, Modeling of ind farms in the load flo analysis, IEEE Trans. on Poer Systems, Vol. 5, No., 2, pp. -5. [5] U. Eminoglu, Modelling and application of ind turbine generating systems (WTGS) to distribution systems, Reneable Energy, Vol. 34, 29, pp. 2474-2483. [6] K. C. Divya and P. S. N. Rao, Models for ind turbine generating systems and their application in load flo studies, Electric Poer Systems Research, Vol. 76, 26, pp. 844-856. [7] A. Feijoo, On PQ models for asynchronous ind turbines, IEEE Trans. on Poer Systems, Vol. 24, No. 4, 29, pp. 89-89. [8] M. H. Haque, Incorporation of fixed speed ind farms in poer flo analysis, Proc. of the IET Reneable Poer Generation, Beijing, China, 23. [9] M. E. Baran and F. F. Wu, Netork reconfiguration in distribution systems for loss reduction and load balancing, IEEE Trans. on Poer Delivery, Vol. 4, No. 2, 989, pp. 4-47. [2] M. H. Haque, A general load flo method for distribution systems, Electric Poer Systems Research, Vol. 54, 2, pp. 47-54. [2] G. X. Luo and A. Semlyyen, Efficient load flo for large eakly meshed netorks, IEEE Trans. on Poer Systems, Vol. 5, No. 4, 99, pp. 39-36. [22] M. H. Haque, Efficient load flo method for distribution systems ith radial or mesh configuration, IEE Proc.-Gener. Transm. Distrib. Vol. 43, No., 996, pp. 33-38. [23] J. R. Aguero, Distribution system planning in smart grid era, IEEE Poer and Energy Magazine, Vol. 9, No. 5, 2, pp. 82-93. [24] Bureau of Meteorology (BoM), Commonealth of Australia: http://.bom.gov.au/. [25] H. D. Chiang and R. Jean-Jumeau, Optimal netork configurations in distribution systems-part 2: Solution algorithm and numerical results, IEEE Trans. on Poer Delivery, Vol. 5, No. 3, 99, pp. 568-574. 25