986 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007 Theoretical Calculation of Inrush Currents in Three- and Five-Legged Core Transformers Luis Sáinz, Felipe Córcoles, Joaquín Pedra, Member, IEEE, and Luis Guasch Abstract—In this paper, a theoretical study of the three-phase transformer behavior in the presence of a sag is presented by considering that the fault clearing is produced simultaneously in all phases. Analytical expressions of the magnetic flux and the inrush current after voltage recovery are obtained, and the results presented in the literature are analytically justified. The influence of depth, duration, and the initial point-on-wave on the peak value of the inrush current are studied for all sag types. Simple expressions to obtain the value of the current peaks are presented. Index Terms—Inrush current, transformer model, voltage sag. Fig. 1. Electric equivalent circuit of a Wye G-Wye G three-phase transformer. I. INTRODUCTION A VOLTAGE SAG is a short-duration reduction in root-mean-square (rms) voltage. The most severe voltage sags are produced by faults in power systems, and it can lead to malfunctioning equipment [1]. In the case of power transformers, the sudden voltage recovery, when a fault is cleared, can saturate the core transformer. This transformer saturation can involve high inrush currents [2], which are very sensitive to the voltage recovery instant. Therefore, a precise study of the problem must take into account that this instant can only have discrete values, since fault clearing is produced in the natural current zeros [3], [4]. Nevertheless, as in the previous study of [5], the consideration that fault clearing is produced simultaneously in all phases simplifies the study and provides a preliminary estimation of the inrush currents. The current peaks calculated by assuming that voltage recovery can only be produced in the current zeros are lower than those obtained when this assumption is not made. In the studies of [4] and [5], the three-phase transformer model is implemented and simulated in PSpice. In this paper, the inrush currents caused by symmetrical and unsymmetrical voltage sags are analytically studied in detail by considering that fault clearing is produced simultaneously in all of the phases. This simplification allows analytical expressions of the magnetic flux and inrush current after a voltage sag to be Manuscript received February 23, 2006; revised May 5, 2006. This work was supported by Grant DPI2000-0994. Paper no. TPWRD-00081-2006. L. Sáinz, F. Córcoles, and J. Pedra are with the Department of Electrical Engineering, Universitat Politècnica de Catalunya, Barcelona 08028, Spain (e-mail: sainz@ee.upc.es; corcoles@ee.upc.es; pedra@ee.upc.es). L. Guasch is with the Department of Electronic, Electrical, and Automatic Engineering, Universitat Rovira i Virgili, Tarragona 43007, Spain (e-mail: llguasch@etse.urv.es). 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/TPWRD.2006.881428 obtained. Inrush current dependence on sag type, depth, duration, and the initial point on wave is characterized. These expressions allow the transformer behavior presented in [5] to be explained and an approximate value for the peak of the inrush current to be analytically calculated. This can be useful in power system studies, such as relay coordination, transformer stress, etc. II. TRANSFORMER MODEL Two types of three-phase transformers are studied in this paper, namely three- and five-legged core transformers. The model proposed in [4] and [5] is used to study the three-legged transformer, whereas a particular case of this model is used to study the five-legged core transformer. A. Electric- and Magnetic-Circuit Models The electric and magnetic relations of the Wye G-Wye G three-phase transformer are shown in Figs. 1 and 2 (1) In this model, each leg is viewed as a separate magnetic element, and the expression used to represent the core nonlinear that relates the magnetic potential behavior is a function in the leg and the flux through it 0885-8977/$25.00 © 2007 IEEE Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. (2) SÁINZ et al.: THEORETICAL CALCULATION OF INRUSH CURRENTS IN CORE TRANSFORMERS 987 Fig. 2. Magnetic equivalent circuit of a three-legged three-phase transformer. where , , , and are experimental parameters which allow this single-valued function to be fitted to the transformer saturation curve [5]. has been In Fig. 2, it can be observed that the reluctance considered constant because it represents the air path in a threelegged transformer. In order to eliminate the winding turns, the above relations are rewritten according to [4] as follows: Fig. 3. Magnetic fluxes in a symmetrical voltage sag of characteristics . 0.4, t 4.375T , and 1 = (3) is the winding turn ratio and where is the core magnetic flux linked by the primary windings. The relation between the magnetic potential in the leg and the flux through it is rewritten as In this paper, only the primary voltages, currents, and total fluxes are of interest. Thus, these primary variables are renamed as follows for clarity purposes: (4) =0 h = The above considerations are also true for the transformer bank. Thus, the study of the five-legged transformer can be extended to the transformer bank. In all of the cases studied in this paper, the secondary supplies a rated resistive load of 0.8 per phase. In fact, load magnitude and load character (resistive or inductive) have no influence on the transformer behavior [5]. III. ANALYTICAL CHARACTERIZATION OF THE CURRENT PEAKS References [4] and [5] show that the main reason why transformer current peaks appear is found in the dc component of with the total magnetic flux after the sag; for example, in Fig. 3. Transformer saturation occurs when this dc component is not null, causing the current to be high. This section presents an analytical study of the total magnetic flux and and the inrush current when the voltage sag ends with . This study is conducted in a Wye G-Wye G three-phase five-legged transformer (or a transformer bank). B. Transformer Data A. Total Flux Peak Determination The study of the three-legged model has been performed with a 60-kVA, 380/220-V Wye G-Wye G three-phase transformer whose linear and nonlinear parameters have been obtained from experimental measurements [4]. The study of the three-phase five-legged transformer is also performed with the same data as the three-legged model but considering: • reluctance values of legs and are forced to be equal to the values of leg ; has been considered null be• value of the reluctance cause, during transformer saturation, it will not be significant in comparison with the leg reluctances. Considering the transformer model (3), the total magnetic flux linked by the primary windings can be calculated as (5) where the primary total magnetic flux is the voltage drop in the primary winding resistances ( with ) has been neglected. The total magnetic flux is derived from (5) and gives Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. , and (6) 988 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007 TABLE I MAGNITUDE AND ANGLE OF VOLTAGE SAG PHASORS TABLE II AND t TO AVOID OBTAINING FLUX AND CURRENT PEAKS. VALUES OF VALUES OF TO OBTAIN THE MAXIMUM FLUX AND CURRENT PEAKS WHEN t IS FIXED, AND i 1 1 Taking into account that When the supply voltage is with , and the transformer is in steady state, the magnetic flux is expressed as (7) As the mean value of the flux is zero in steady state, the next condition must be true for the initial flux (8) (12) the postsag flux (11) can be rewritten as and the final expression of the flux is (13) (9) The supply voltages assumed when a sag is produced are where the two first terms are identical to the terms of the case without a sag (9). Then, the final expression of the flux when voltage is recovered is . (10) , The pre and postsag voltage is defined as , . The magnitude and angle values of the voltage phasors during and ) are shown in Table I. They are determined the sag ( from the sag expressions of Table II in [4], which were obtained from the sag classification of [1]. Flux continuity implies that flux can be calculated at any instant by means of (6), even if the voltage changes during the sag. The flux after the sag (Fig. 3) can be calculated as (11) where the sag is produced between times and . (14) where (15) This term is constant and its physical meaning is a dc magnetic flux. The Appendix shows the expressions of obtained from (15) for the different sag types. The current peak is produced when there is a flux peak bemagnetic curve (3). Therecause they are related by the fore, the peak values of the total magnetic fluxes must be calculated to determine the current peaks. Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. SÁINZ et al.: THEORETICAL CALCULATION OF INRUSH CURRENTS IN CORE TRANSFORMERS The flux peak produced by the sag depends on the depth , the duration , and the initial point-on-wave . Thus, the following function must be studied: 989 In this case, the magnetic equation of (3) gives (20) This expression allows the relation to be calculated (21) and the use of the nonlinear expression in (3) yields (16) (22) (17) Fig. 4(a) shows the influence of the nonlinear curve parameters on the shape of the nonlinear saturation curve. The above nonlinear relation can be approximated by a piecewise linear saturation curve. If only two pieces are used Fig. 4(a), their parameters can be related easily to the parameters of the nonlinear function (22). The two-piece linear saturation curve is defined as where In the case shown in Fig. 3, the flux peak is produced in phase , and its value is . must be analyzed to study the influence Then, of sag type, depth, duration, and the initial point-on-wave on the maximum and minimum values of the postsag magnetic flux. is obtained from the analytical exThe flux in the Appendix. For example, the dc compressions of ponent of the total magnetic fluxes in a type B sag (Appendix) is (23) where is the saturation flux. The current peak can be easily determined from (23). Fig. 4(b) shows the current peak determination from the sat, , and uration curve for a type B sag of ranging from 0 to . Fig. 4(b) shows that the difference between the use of the nonlinear saturation curve (continuous line) and the two-piece approximation (dashed line) is small. The expression for calculating the current peak when is derived from (23) (24) (18) where and, according to (16), the flux peak gives is the phase with the greatest flux peak. IV. BEHAVIOR STUDY OF THE THREE-PHASE TRANSFORMER (19) Fig. 4(b) shows the curve 0.73 and characteristics for a type B sag of 5.5 obtained from (19). B. Current Peak Determination The transformer total flux and the primary currents can be related if some simplifications are made in the transformer model. • Transformer is at no load (or the secondary currents are of less importance than the primary currents when the trans(or ). former is saturated) is null, as the five-legged transformer is • Reluctance considered (the magnetic potential is also null). • Core-loss resistances are neglected . The expressions of the previous section allow the transformer current peak to be analytically calculated. Thus, the influence of sag type, duration , depth , and the initial point-on-wave on the current peaks can be studied. It must be noted that in the study of the previous section, the source has no internal impedance. This leads to an overestimation of the transformer current peaks because the source impedance damps these peaks [5]. All of the analytical calculations in this paper have been obtained by programming the analytical expressions from Section III and the Appendix in MATLAB [6]. To analyze the duration, initial point-on-wave and depth influence on the current peaks, the following series of sags have been studied for the seven sag types, A to G. • Duration influence: Four series of sags of depth have been studied Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. . 990 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007 Fig. 4. (a) Influence of the parameters on the shape of the saturation curve. (b) Current peak determination from the saturation curve. • Initial point-on-wave influence: Four series of sags of depth have been studied (25) . • Depth influence: One series of sags with the most unfavorable duration and the most unfavorable initial point-onand , obtained from Table II, have been wave studied (26) A. Three-Phase Five-Legged Transformer In order to study this transformer analytically, the data of [4] have been considered imposing that the three legs are identical (the reluctance values of legs and are set equal to the values of leg ) and the air-path linear reluctance is null. Thus, this study can also be extended to the transformer bank. Figs. 5 and 6 show the sag duration and the initial point-onwave influence on the current peak, respectively (series to and to ). By comparing these figures with the corresponding ones in [5], it can be noted that the analytical study predicts the five-legged transformer behavior (and the transformer bank behavior) accurately. Nevertheless, the analytical results are slightly lower than the simulation results in [5] because the resistance voltage drop has been neglected. 1) Duration and Initial Point-on-Wave Influence: The value of and required to avoid obtaining current peaks (which corresponds to minimum flux values) when voltage is recovered can be calculated by imposing in the expressions of the Appendix. These values are shown in Table II. The most unfavorable situation is produced when the maximum flux peak (which corresponds to the maximum current peak) is obtained. The maximum flux peak can be calculated analytically. At this point, the difference between the maximum flux peak (which takes into account all current peaks corresponding to all values of and ), and the maximum flux peak when is fixed (which takes into account only the current peaks corresponding to all values of , but only one value of ) must be considered Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. (27) SÁINZ et al.: THEORETICAL CALCULATION OF INRUSH CURRENTS IN CORE TRANSFORMERS 991 Fig. 5. Three-phase five-legged transformer: analytical results of the sag duration influence on the current peak for the seven sag types. Fig. 6. Three-phase five-legged transformer: analytical results of the initial point-on-wave influence on the current peak for the seven sag types. The value of required to obtain (for example, in series , , , or ) is obtained by imposing in the expressions of Appendix A. These values are shown in Table II for the relation . shown in Table II, only Among all of the values of the highest values produce the most unfavorable situations, that . According to the Appendix, is, the maximum flux peaks these most unfavorable situations are produced for specific Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. 992 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007 Fig. 7. Three-phase five-legged transformer: analytical results of the depth influence on the most unfavorable current peak for the seven sag types. values of with (when i.e., series 7) and for the corresponding relations of Table II. The maximum flux peak can be calculated as (28) where for type A,B,D,E, and F sags for type C sag for type G sag. (29) The maximum current peak can be calculated by the substitution of (28) in (24). It can be noted that type A, B, D, E, and F sags have the highest values. 2) Depth Influence: According to series , Fig. 7 shows the depth influence on the maximum current peak. For a given and , the expressions of the Appendix show that the relation between and the depth is a straight line with negative slope (i.e., there is a linear relation). Then (30) where does not depend on the sag depth . Considering (23), the relation between the current peaks and the depth is also linear with two possible negative slopes depending on whether the total magnetic flux is above or below the saturation flux or . B. Three-Phase Three-Legged Transformer In order to study this transformer analytically, the data of [4] for the three legs have been considered. According to series to and to , Figs. 8 and 9 show the sag duration and the initial point-on-wave influence on the current peaks, respectively. By comparing these figures with the corresponding ones in [5], it can be observed that although the analytical model conis null ( in siders that the value of the reluctance Section III-B), the study predicts the three-legged transformer behavior accurately. This good agreement is achieved because (the magnetic potenthe magnetic potential in the reluctance tial , (3)) is not significant in comparison with the remaining magnetic potentials when the transformer is saturated [ , which is equivalent to considering , as in (20)]. By comparing the five- and three-legged transformer results (Figs. 5 Fig. 8. Three-phase three-legged transformer: analytical results of the sag duration influence on the current peak for the seven sag types. and 6 with Figs. 8 and 9), it can be observed that the main difference is in the symmetry of the waveforms whereas the current peaks do not essentially change. Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. SÁINZ et al.: THEORETICAL CALCULATION OF INRUSH CURRENTS IN CORE TRANSFORMERS 993 The procedure to calculate the maximum current peaks is the same as the procedure of the previous section. However, the three-legged transformer has fewer most unfavorable situations than the five-legged transformer for type A, E, and G sags because the central leg is shorter than the outer legs. 2) Depth Influence: The study of the depth influence leads to the same conclusions as those obtained for the five-legged transformer. V. CONCLUSION The effects on the current peaks of symmetrical and unsymmetrical voltage sags in Wye G-Wye G three-phase transformers have been analytically studied by considering that fault clearing is produced simultaneously in all of the phases. Analytical expressions for the total magnetic flux and the inrush current when the transformer is saturated have been obtained. This saturation is produced when voltage is recovered. This simple model allows the behavior of the current of the five- and the three-legged transformers to be explained when a sag is produced. These analytical expressions also allow the current peak value to be analytically estimated. It can be observed that the obtained current peaks are higher than those obtained when it is considered that voltage recovery can only be produced in the current zeros [4]. Thus, following the analytical process indicated in this paper, this consideration must be taken into account in future studies to obtain more precise results. APPENDIX DC COMPONENT OF THE TRANSFORMER TOTAL MAGNETIC FLUXES The expression of the dc component of the transformer total magnetic fluxes when voltage is recovered can be obtained from and (15) by considering with (31) which can be rewritten as follows: (32) Fig. 9. Three-phase three-legged transformer: analytical results of the initial point-on-wave influence on the current peak for the seven sag types. 1) Duration and Initial Point-on-Wave Influence: As in the case of the five-legged transformer, Table II shows the values and required to avoid obtaining current peaks, and the of to obtain the maximum current peaks when is values of . fixed Equation (32) can be rewritten by using duration and the initial point-on-wave as variables (33) Thus, for the different sag types, the dc component of the total magnetic flux can be obtained from expression (33) and by con- Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. 994 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007 sidering and the values of • Type A and and , of Table I. , Using trigonometric relations, the above expression can be rewritten as (34) (39) • Type B • Type E (35) • Type C (40) • Type F (36) Using trigonometric relations, the above expression can be rewritten as (41) (37) Using trigonometric relations, the above expression can be rewritten as (38) (42) • Type D Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply. SÁINZ et al.: THEORETICAL CALCULATION OF INRUSH CURRENTS IN CORE TRANSFORMERS • Type G 995 [5] L. Guasch, F. Córcoles, J. Pedra, and L. Sáinz, “Effects of symmetrical voltage sags on three-phase three-legged transformers,” IEEE Trans. Power Del., vol. 19, no. 2, pp. 875–883, Apr. 2004. [6] The MathWorks, MATLAB 5.3 and Simulink 3.0. Natick, MA, 1999. Luis Sáinz was born in Barcelona, Spain, in 1965. He received the B.S. degree in industrial engineering and the Ph.D. degree in engineering from the Universitat Politècnica de Catalunya, Barcelona, Spain, in 1990 and 1995, respectively. Since 1991, he has been Professor in the Department of Electrical Engineering, Universitat Politècnica de Catalunya. His main field of research is power system quality. (43) Using trigonometric relations, the above expression can be rewritten as (44) Felipe Córcoles was born in Almansa, Spain, in 1964. He received the B.S. degree in industrial engineering and the Ph.D. degree in engineering from the Universitat Politècnica de Catalunya, Barcelona, Spain, in 1990 and 1998, respectively. Currently, he is a Professor in the Department of Electrical Engineering, Universitat Politècnica de Catalunya, where he has been since 1992. His research interests are electric machines and power system quality. Joaquín Pedra (S’85–M’88) was born in Barcelona, Spain, in 1957. He received the B.S. degree in industrial engineering and the Ph.D. degree in engineering from the Universitat Politècnica de Catalunya, Barcelona, Spain, in 1979 and 1986, respectively. Since 1985, he has been Professor in the Department of Electrical Engineering, Universitat Politècnica de Catalunya. His research interest lies in the areas of power system quality and electric machines. REFERENCES [1] M. H. J. Bollen, Understanding Power Quality Problems: Voltage Sags and Interruptions. Piscataway, NJ: IEEE Press, 2000. [2] E. Styvaktakis, M. H. J. Bollen, and I. Y. H. Gu, “Transformer saturation after a voltage dip,” IEEE Power Eng. Rev., vol. 20, no. 4, pp. 62–64, Apr. 2000. [3] M. H. J. Bollen, “Voltage recovery after unbalanced and balanced voltage dips in three-phase systems,” IEEE Trans. Power Del., vol. 18, no. 4, pp. 1376–1381, Oct. 2003. [4] J. Pedra, L. Sáinz, F. Córcoles, and L. Guasch, “Symmetrical and unsymmetrical voltage sag effects on three-phase transformers,” IEEE Trans. Power Del., vol. 20, no. 2, pt. 2, pp. 1683–1691, Apr. 2005. Luis Guasch was born in Tarragona, Spain, in 1964. He received the B.S. degree in industrial engineering and the Ph.D. degree in engineering from the Universitat Politècnica de Catalunya, Barcelona, Spain, in 1996 and 2006, respectively. Currently, he is a Professor in the Department of Electronic, Electrical, and Automatic Engineering, Universitat Rovira i Virgili, Tarragona, where he has been since 1990. His research interests are electric machines and power system quality. Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2010 at 14:50:41 UTC from IEEE Xplore. Restrictions apply.