See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/335428520 Measurement and FEM analysis of DC/GIC effects on transformer magnetization parameters Conference Paper · June 2019 DOI: 10.1109/PTC.2019.8810423 CITATIONS READS 17 679 3 authors: Hilary Chisepo C.T. Gaunt University of Cape Town University of Cape Town 18 PUBLICATIONS 90 CITATIONS 162 PUBLICATIONS 2,394 CITATIONS SEE PROFILE Les Borrill SSE plc 9 PUBLICATIONS 53 CITATIONS SEE PROFILE All content following this page was uploaded by Les Borrill on 29 September 2019. The user has requested enhancement of the downloaded file. SEE PROFILE Measurement and FEM analysis of DC/GIC effects on transformer magnetization parameters Hilary K. Chisepo and C.T. Gaunt Leslie D. Borrill Department of Electrical Engineering University of Cape Town South Africa chshil001@myuct.ac.za; ct.gaunt@uct.ac.za Koeberg Nuclear Power Station Eskom Holdings SOC Cape Town, South Africa BorrilLD@eskom.co.za Abstract— Topologically derived equivalent circuit models for transformers can be improved by more accurate parameterization of the magnetization characteristics. We examined the changes in transformer magnetization parameters by measuring the effects of dc on a scaled down model singlephase four-limb (1p4L) transformer and confirmed them with FEM modelling. The results define the non-linear character of the transformer inductance when partial saturation occurs, such as in over-excited transformers or with half-wave saturation caused by leakage dc or geomagnetically induced currents. The analysis clarifies the difference between the constant air inductance of a winding (Lair) and the instantaneously changing saturation inductance (Lterminal) used in low-frequency transformer equivalent circuit models and shows how to choose a representative value for the equivalent circuit parameter. The tests lead to a more accurate graphical depiction of related parameters during transformer half-cycle saturation than usually presented. Index Terms-- flux, inductance, GIC, FEM, measurements I. INTRODUCTION Geomagnetically induced currents (GIC) or other sources of dc excitation in the presence of ac energization can disturb the normal operation of power transformers. GIC/dc components of current cause transformers to operate under half-cycle saturation and the response depends on the core structure and magnetization characteristics. Half-cycle saturation generates several unwanted conditions, including the generation of even and odd harmonics, overheating, draw of reactive/non-active power, and audible noise [1-5]. In severe cases, it may lead to power transformer damage, power system instability, and even blackouts [6], [7]. The derivation of representative electromagnetic transients (EMT) transformer models has been widely researched with the aim of improving the accuracy for mid- to low-frequency studies [8-13]. Several studies depict the transformer halfcycle saturation diagrammatically by relating various parameters to the flux B and rms magnetic field H [14-18]. Amidst the numerous efforts to enhance transformer This work was supported in part by Eskom Holdings under the EPPEI program, Royal Smit Transformatoren and a grant from the Open Philanthropy Project. Mentor, a Siemens Business provided the Simcenter MAGNETTM academic license for the FEM simulations. modelling for GIC (and inrush currents and ferroresonance [9]), the understanding of the complex and different transformer responses is incomplete [10] or contradictory. A simplified illustration [12] suggests the terminal inductance of a transformer in deep saturation, Lterminal, “may or may not” differ from the inductance of the same transformer’s windings without its core, Lair. (The two inductances are considered equal in some GIC studies [9], [1921].) The uncertainty arises from the difficulty of measuring Lterminal in an industrial testing facility/laboratory, and because manufacturers do not readily provide the data. Despite the uncertainty, Lterminal (or Lair) is probably the most important parameter for accurate transformer modelling for GIC and other studies involving deep core saturation [22]. By measurement and FEM simulation, this study shows a clear difference between Lterminal and Lair thus consolidating the understanding of the suggested differences in [12]. This leads to a new composite depiction of the important magnetization parameters during half-cycle saturation derived from real BH properties of power transformer core steel. II. THEORY AND MEASUREMENTS GICs are variable very low frequency currents (mHz range, so quasi-dc relative to the power frequency) driven by an induced earth surface potential between the neutrals of widely separated transmission transformers. An offset in the magnetic flux B(t) is introduced in the transformer core and, depending on the size of the GIC, may shift B(t) into the saturation region in one half of the ac cycle (usually depicted on a B-H curve). At the same time, a corresponding Imag is drawn by the transformer to meet the partially saturated core’s H requirements. (Since the Imag lags the applied voltage by 90°, the reactive/non-active power demand in the transformer also increases.) Fig. 1, based on widely presented illustrations, shows the interacting magnetization parameters. A significant question is “what is the inductance between ‘a’ and ‘b’ in Fig 1”? Ideal BH curve of core steel b a Developed mean flux ‘Normal’ permeability H Imag(t) TIME B(t) TIME Fig. 1. Transformer core half-cycle saturation due to GIC/dc, adapted from various reports [14-18, 23]. A method proposed by de León et al. [22] to measure Lterminal was applied to a model single-phase four-limb (1p4L) transformer (at the time of this preliminary work [22] it was common belief that a saturated transformer was in “air core inductance” operation, however the actual measurements here represent Lterminal). The model was fabricated with two parallelconnected winding assemblies (each consisting of two windings with subtractive polarity) and the same grainoriented electrical core steel as an actual 1p4L generator transformer. The model had no other materials of electromagnetic significance, so Lterminal is a function of the core and windings only. The rating of the model transformer was defined at the limit of linear magnetization (no overexcitation) to be 110/206.2 V, 4.4 kVA. Each pair of primary (outer) windings had 80 turns and secondary (inner) windings had 150 turns, referred to as 80t and 150t hereafter. A. Description of Lterminal De León’s three-phase measurement circuit [22] was adapted as shown in Fig. 2 to measure Lterminal using a singlephase circuit (only a single-phase rectifier was available). This circuit used supplemental batteries to drive the transformer into deep saturation with sufficient dc Ampère-turns and a small ac ripple supplied by the rectifier sufficient to determine the transformer inductance in saturation). One phase of a 0380 V, 60 A, 50 Hz three-phase variac supplied the circuit and all measurements were recorded with an IEC 76-1 (1993) compliant WT1800 Yokogawa Power Analyzer. the incremental flux while removing winding and source resistances from the calculations, thus eliminating possible error caused by temperature-dependent changes in resistance. Fast Fourier Transform values of the voltages and currents were then substituted in (1) [22]: !"#$%&'() = *+,-_. (1) /01. '234_. where k is the harmonic order, Vout_k is the amplitude of the secondary voltage of the dominant harmonic, fk is the dominant harmonic frequency, n is the transformer ratio and I in_k is the amplitude of the dominant harmonic of the primary current. While other harmonics were considered and documented, separately, yielding good correlation suggested by [22] using a three-phase circuit, the dominant saturating harmonic of the distorted input current (see Fig. 3) was chosen in this study using a single-phase circuit to calculate Lterminal with increasing dc up to its final value. 90 Voltages (Volts), Input Current (Amps) B Result of dc component 40 -10 0 10 20 30 40 50 -60 -110 -160 V out V in I in Time (ms) Fig. 3. Measured input and output voltages (black) and resultant input current with a dc component of 60 A dc. Input current has a dominant 2nd harmonic. B. Description of Lair measurements After completing the Lterminal measurements, the core was removed without disturbing its winding positions, which was possible with the construction method of the transformer. Fig. 4 shows the “air core” transformer for measuring Lair. Winding assemblies Fig. 2. Schematic circuit for measuring the saturation inductance of a 1p4L transformer by supplementing the dc offset with batteries. The 80t and 150t windings were investigated separately, controlling the current in the test winding and measuring the voltage across the other open circuited windings. This captured Fig. 4. 1p4L test transformer without the core for the Lair measurements relative to the removed core (indicated by the white dotted lines). Lair was measured with ac only, using a 0-240 V, 20 A single-phase variac in series with a current limiting variable resistor. We have not found any description of measuring L air with which to compare this approach. The linear or constant characteristic of an “air core” was verified by injecting currents ranging from 3-10 A rms and then calculating using (2): !(&$ = 5 /012 6 runs of parametric analyses) and maintain accuracy, a oneeighth symmetric mirror model was used, illustrated in Fig. 5. After each simulation, the full model solution was recalculated. (2) / (2) is derived from the expression78 = 9 :(&$ , where 8 is the reactive power, I is the line current, f is the supply frequency of 50 Hz and :(&$ is the air core reactance. These two measurements of Lterminal and Lair were then compared with the results of FEM simulations. III. FEM SIMULATION PROTOCOL A FEM model of the 1p4L transformer had already been validated in a parallel study [24] using FEM models and physical measurement with ac-dc excitation. The core nonlinear characteristic used in the FEM is detailed in that study of the core joints. The purpose of the further FEM simulation here is to investigate Lterminal and Lair beyond the limits of measurement of the physical model. A. FEM calculations for Lterminal The FEM model included core nonlinearity effects on the magnetizing currents and leakage/stray flux with simultaneous ac and dc excitation in the transient analysis domain. The winding excitations were performed using a field-circuit coupled approach with an external circuit adapted from the schematic in Fig. 2. Unlike the laboratory procedure in Chapter II that relied on the dominant saturating harmonics to determine a final value for Lterminal, post-processing of the global quantities in multiple FEM solutions measured the change of Lterminal with current. From the FEM solution, Lterminal is determined from the magnetic stored energy W when a current i flows in the windings, given by (3): /; (3) !"#$%&'() = 6 < & W is the integral over the whole 3D volume of the magnetic energy density derived from Poynting’s theorem [25] and is expressed as (4): > = 7 ?* @ AB (4) where w is the corresponding magnetic energy volume density relating to the area under the BH curve. When the magnetic flux density is B1 the magnetic energy volume is given by (5): H @ = ?H I CDEF G AE (5) J where H(B) is the magnetic field intensity corresponding to a given B for the non-linear core. A 3D model with stranded winding cylinders was required for the Lterminal calculation to capture the magnetic spatial distributions and inductances in the third dimension. Although 2D models solve more quickly, accuracy is sacrificed. Instead, to shorten simulation time (which was necessary for multiple Fig. 5. A 1/8th symmetric 3D model with explicitly modelled laminations close to the core surface at normal ac only excitation For the voltage driven coupled field-circuit model, the current density J for the full FEM model is defined by: L* (6) K= < MN where A is the cross-sectional area of the windings as a function of the winding height, V is the applied voltage, N is the number of turns, R is the resistance of the windings given by )M6 O= (7) PL and l and s are the windings’ sweep distance and conductivity, respectively. Keeping J constant for the symmetric and full models, it can be shown that the symmetric model requires an applied ac voltage Vsym that is a quarter of the equivalent full model’s voltage (8). BPQ% = B. TU RSG 6 G V W X 6 = SMN YL * = < Y (8) FEM calculations Lair A FEM geometry for the windings was based on the actual “air core” transformer of Fig. 4. Because excluding the core avoided long calculations of the nonlinearities and core joint details, the problem became linear and could be solved in the magnetostatic domain with dc only. The primary and secondary windings separately, with the other side open circuited. For each pair of windings (which are connected in parallel with opposing polarity in the actual laboratory transformer) a dc current of 0.5x A is injected into Winding Assembly A and a reverse dc current of -0.5x A is simultaneously injected in the Winding Assembly B. This means that if x=1 then the total winding current for substitution into (3) is 1 A. Since it is a linear problem, the calculated Lair is independent of the magnitude of the current. This approach was repeated with ac in a time-harmonic domain (50 Hz) to check for consistency with the global solutions derived from the preceding magnetostatic solutions. IV. RESULTS A. Measured and simulated results In the presence of an applied voltage of 60 V rms and with increasing levels of dc, the 2nd harmonic component of the input current was found to be dominant for both the 150t and 80t windings. The measured Lterminal appeared to converge at 798 μH for the 150t windings, and 275 μH for the 80t windings. The measured Lair was constant over the full range of the applied ac voltage, as expected, and yielded 694 μH and 355 μH for 150t and 80t, respectively. (The analytical calculation L=N2μoA/l using the geometry of the windings yielded 664 μH and 353 μH for Lair, corresponding closely to the measurements.) The FEM simulation protocol for Lterminal differed from the practical measurements in the sense that saturation was not achieved by injecting harmonics and a dc component with a non-ideal rectifier. Instead, as is allowable in the FEM domain, increasing levels of dc (much higher than experimental dc levels) were added to an applied voltage of 60 V rms. The simulated Lterminal initially appeared to converge at 722 μH and 380 μH for the 150t and 80t windings, respectively. The simulated Lair was a constant 654 μH (150t windings) and 380 μH (80t windings). B. Multiple parameter simulations for Lterminal The FEM simulation experiments were extended to investigate why the measured 80t Lterminal (275 μH) was lower than the measured and simulated Lair, (355 μH and 340 μH). In Table I, experiments (Exp) 1 and 4 are the results corresponding to the physical model in subsection A. Without the constraint of the physical model, Exp 2 and 5 investigated the extent to which mutual inductances between the windings affected the simulated Lterminal by removing from the simulation the winding not under investigation. FEM Exp. 3 and 6 represent the effect of swopping the positions of the 80t and 150t windings while keeping the winding not under investigation open circuited. In a further experiment, not shown in the table, the effect of drastically reducing the applied ac (to approximately 5 V) in the presence of the same high dc was shown to have no effect on the value of Lterminal. Exp 1 and 2 show the effect of mutual inductance Lm between the 80t and 150t windings (an increase of 16 μH when the 150t winding is removed) at Lterminal. Similarly, Exp 4 and 5 show that removing the outer 80t winding reduces the 150t Lterminal by Lm =20 µH. Changing the positions of the windings, the outer 80t winding (Exp 1) to the inner position (Exp 3), or the inner 150t (Exp 4) to the outer position (Exp 6), returns a higher inductance for the larger diameter winding, as expected. The measured and FEM modelled Lterminal of Exp 4 are comparable (10% difference). By contrast, the measured Lterminal of Exp 1 is much lower than the simulated Lterminal of Exp 1 and 2 and the measured and simulated Lair, especially considering the FEM model’s consistency with other measured parameters. C. Further Lterminal, Lair comparisons in the FEM with high dc Consideration was given to the possibility of measurement error for the 80t inner windings. Looking at the final dc levels measured in the laboratory to derive Lterminal it became clear that the very large dc input needed to reach deep core saturation in the 80t windings exceeded the limits of accuracy of the Yokogawa power meter. Lterminal for the 150t windings was reached at a dc input of 71 A, whereas the fewer turns of the 80t winding required 117 A dc. Even using shunt resistors to measure high dc, the interpretation of the power meter resulted in errors at high dc levels. Therefore, the Lterminal and Lair parameters were compared directly in the FEM analysis with very high dc up to 10 kA. Lterminal with a dc of 10 kA was reached at 341 μH and 657 μH for the 80t and 150t respectively, shown in Fig. 6. These values are only a few µH higher than the FEMair core solutions (340 μH and 654 μH, respectively), and suggest that the FEM model’s core is very close to (complete) “air core” saturation. It should be noted, though, that the unrealistic dc in the kA range far exceeds maximum real GIC [~100 A/phase]. It was used only to test how much dc would be needed to reach an Lair of 340 μH and show that the result is much higher than the erroneous measurement value of 275 μH. 2000 TABLE I. FURTHER SIMULATIONS PERFORMED TO INVESTIGATE DISCREPANCY IN MEASURED 80 TURN TERMINAL SATURATION INDUCTANCE Exp 1 Exp 2 Exp 3 Exp 4 Exp 5 Exp 6 Description of FEM Exp 80t outer, 60 V ac + dc, 150t inner open circuit 80t outer, 60 Vac + dc, without 150t inner winding 80t inner, 60 V ac + dc, 150t outer open circuit 150t inner, 60 Vac + dc, 80t outer open circuit 150t inner, 60 V ac + dc, without 80t outer winding 150t outer, 60 V ac + dc, 80t inner open circuit FEM Lterminal (μH) Measured Lterminal (μH) 380 275 396 - 270 - 722 798 702 1133 L air 150t inner 1600 L (μH) FEM Exp L terminal 150t inner 1800 1400 L terminal 80t outer 1200 L air 80t outer 1000 800 600 400 200 0 0 2 4 6 dc current (kA) Fig. 6. Further FEM simulations with extremely high dc 8 10 In this study, then, the difference between measured and modelled Lterminal for the 80t windings is attributed to limitations in the resolution of measurement equipment at very high distorted currents and should be replaced with the FEM derived value of 380 μH. V. PARAMETRIC HALF-CYCLE SATURATION DEPICTION This study’s results lead to an improved depiction of halfcycle saturation. Fig. 7 shows some of the multi-parameter relationships under steady state GIC or dc offset of the ac. The widely used [14-18] half-cycle saturation depiction of the BH curve is adapted. The ac flux B(t) and dc component interaction is represented by the black waveform on the left of Fig. 7. The dc offset creates a ‘displaced zero’ axis above the mid-point of the linear part of the BH curve. A simplified piecewise inductance model, as used in EMTP software [26], defines the knee-point at the flux density Bn at rated voltage, with two linearized portions at 1.15B n and 1.25Bn between the linear ideal operating BH curve and a linear air core characteristic, but such models often lack accuracy in saturation studies. When the transformer is driven beyond the knee of the BH curve, the relatively small (exciting) magnetizing current, shown by the green line in the bottom right part of Fig. 7, changes with dc offset to the Imag (solid red) curve for a singlephase transformer. (The shape and displacement of the red Imag curve is derived from separate FEM simulations with the same typical core steel [27] used for power transformers.) McLyman [23] defined the knee point of the BH curve where the tangent from the origin identifies the maximum ‘normal’ permeability (Z7 B/H), also called amplitude permeability [27]. L=k1μr, B(t) B [H B(t) with dc μr=k2 [\ Real BH curve of core steel 1.8 1.6 1.4 Developed mean flux 1.2 B =1 pu 1 Instantaneous Lterminal due to dc 0.8 L, μr 0.6 0.4 Lair 0.2 0 0 100 200 TIME H "∞ B(t) with ac only Imag(t) Imag(t) with dc Nominal Imag(t) TIME ‘Displaced zero’ axis caused by dc Fig. 7. Parametric relationships during half-cycle saturation showing a real BH curve, relative permeability and transformer inductance. The permeability μ and relative permeability μ r as defined by the local gradient (dB/dH) at any ‘point’ on the BH curve have a different maximum magnitude and location from that identified by the amplitude permeability. The instantaneous (or incremental or differential) inductance L Z7 dφ/di is directly proportional to μ and μr. Therefore, L, μ and μr follow the same shape with different scaling constants. μr tends to 1 and L tends to Lair as H tends to infinity, but these values are not reached even with high practical values of dc or GIC. The incremental or differential μr is directly related to the BH curve, requiring complete consistency between the measurements of L and BH core steel data. The real BH curve (shown in blue in Fig. 7) was determined by linearizing the portion of the BH curve [27] below 0.4 T and compensating for the corresponding H to incorporate hysteresis, and then performing a ‘moving’ calculation for dB/dH for the entire BH curve to yield μr The shape and slope of μr and L were verified using several grades of power transformer electrical core steel. As a result, the real BH curve shows the saturation transition beyond the knee in the region of a-b of Fig. 1. The instantaneous Lterminal in Fig. 7 corresponds to the peak of the Imag as a result of the dc component. Further, the calculated instantaneous inductance and ‘point’ permeability before the knee are shown without the straight line simplification used in other models [14]. VI. CONCLUSIONS This paper identified the need for more representative EMT transformer models for GIC and related studies. The most important parameter for improved accuracy requires the interpretation of the saturation inductance in topological models. The measurements and FEM simulations have clarified the difference between Lterminal and Lair for the 1p4L laboratory transformer without a tank and these results support the suggestions from the literature [22]. The main improvement of Fig. 7 over Fig. 1 is the avoidance of misleading simplification of the BH curve, enabling a consistent depiction of real BH data and the instantaneous L and μr and Imag with dc. Also, it clarifies that the maximum instantaneous permeability is not coincident with the knee point of the BH curve. The Fig. 7 (time domain) model answers the question about the shape and slope of the inductance and relative permeability between ‘a’ and ‘b’ in Fig 1. The instantaneous inductance L falls sharply as the gradient of the BH curve changes, and Lterminal tends towards Lair but Lair is not reached even with the highest measured GICs. In the context of GIC, the suitable parameter for input into any single-phase topological model is Lterminal and it does not have a single value because it is a function of the GIC or dc as shown in Fig. 6. The implications of the improved model are better accuracy in power system modelling with GIC and dc, and even to partial saturation caused by over-excitation. It is seldom reported that the transformer Q-GIC characteristic is not strictly linear moving from partial to deep saturation [28]. Figs. 6 and 7 from this paper could be used to further understand the non-linearity of L, its effect on Q-GIC, and implications for power system stability studies. The results have been derived and tested using single-phase transformers. Ongoing studies are extending the investigation to transformers with tanks and with three-phase core structures. REFERENCES [1] J. G. Kappenman and V. D. Albertson, "Bracing for the geomagnetic storms," IEEE Spectrum, vol. 27, no. 3, pp. 27-33, March 1990. [2] V. D. Albertson, B. Bozoki, W. E. Feero, J. G. Kappenman, E. V. Larsen, D. E. Nordell, J. Ponder, F. S. Prabhakara, K. Thompson and R. Walling, "Geomagnetic Disturbance Effects on Power Systems," IEEE Transactions on Power Delivery, vol. 8, no. 3, pp. 1206-1216, July 1993. [3] C. T. Gaunt and G. Coetzee, "Transformer failures in regions incorrectly considered to have low," Lausanne, Switzerland, 2007. [4] J. Berge, "Impact of Geomagnetically Induced Currents on Power Transformers," PhD Thesis University of Western Ontario, Ontario, 2011. [5] R. Gong, J. Ruan, J. Chen, Y. Quan, J. Wang and S. Jin, "A 3-D Coupled Magneto-Fluid-Thermal Analysis of a 220 kV Three-Phase Three-Limb Transformer under DC bias," Energies, vol. 10, no. 4, pp. 422-431, 23 March 2017. [6] P. Czech, S. Chano, H. Huynh and A. Dutil, "The Hydro-Québec System Blackout of March 1989: System Response to Geomagnetic Disturbance," in In Proc. EPRI TR-100450: Geomagnetically Induced Currents Conference, Carlifornia, 1992. Electromagnetic Transients - Part II: Complementary Modeling Guidelines," IEEE Transactions on Power Delivery, vol. 31, no. 5, pp. 2420-2430, 2016. [13] S. E. Zirka, Y. I. Moroz, J. Elovaara, M. Lahtinen, R. A. Walling, H. K. Høidalen, D. Bonmann, C. M. Arturi and N. Chiesa, "Simplified models of three-phase, five-limb transformer for studying GIC effects," Internation Journal of Electrical Power & Energy Systems, vol. 103, no. 2018, pp. 168-175, 2018. [14] S. Calabrò, F. Coppadoro and S. Crepaz, "The Measurement of the Magnetization Characteristics of Large Power Transformers and Reactors," IEEE Transactions on Power Delivery, Vols. PWRD-1, no. 4, pp. 224-234, 1986. [15] D. Tousignant, L. Bolduc and A. Dutil, "A method for the indication of power transformer saturation," Electric Power Systems Research, vol. 37, no. 2, pp. 115-120, 1996. [16] L. Bolduc, A. Gaudreau and A. Dutil, "Saturation time of transformers under dc excitation," Electric Power Systems Research, vol. 56, pp. 95-102, 2000. [17] M. Lahtinen and E. Jarmo, "GIC Occurrences and GIC Test for 400 kV System Transformer," IEEE Transactions on power delivery, vol. 17, no. 2, pp. 555-561, 2002. [18] R. Girgis and K. Vedante, "Effects of Geomagnetically Induced Currents on Power Transformers and Power Systems," in PES T&D, Orlando, 2012. [19] H. Tay and G. W. Swift, "A Novel Method of Detecting Asymmetrical Transformer Core Saturation due to GIC," IEEE Transactions of Power Apparatus and Systems, Vols. PAS-103, no. 1, pp. 183-189, Jan 1984. [20] P. Picher, L. Bolduc, A. Dutil and V. Q. Pham, "Study of the acceptable dc current limit in core-form power transformers," IEEE Transactions on Power Delivery , vol. 12, no. 1, pp. 257-265, 1997. [21] M. A. S. Masoum and P. S. Moses, "Influence of Geomagnetically Induced currents on Three-Phase Power Transformers," Sydney, 2008. [7] M. Wik, R. Pirjola, H. Lundstedt, A. Viljanen, P. Wintoft and A. Pulkkinen, "Space weather events in July 1982 and October 2003 and the effects of geomagnetically induced currents on Swedish technical systems," Annales Geophysicae, vol. 27, no. 4, pp. 1775-1787, 2009. [22] F. de León, S. Jazebi and A. Farazmand, "Accurate measurement of the Air-Core Inductance of Iron-Core Transformers with a Non-ideal Low Power Rectifier," IEEE Transactions on Power Delivery, vol. 29, no. 1, pp. 294-296, 2014. [8] B. A. Mork, F. Gonzalez and D. Ischenko, "Parameter estimation and advancements in trasnformermodels for EMTP simulations.," Task MTU-7: Model performance and sensitivity analysis, 2004. [23] C. W. McLyman, Transformer and Inductor Design Handbook, Third ed., New York: Marcel Dekker, 2004. [9] S. Jazebi, F. de León and D. D. Farazmand, "Dual Reversible Transformer Model for the Calculation of Low-Frequency Transients," IEEE Transactions on Power Delivery, vol. 28, no. 4, pp. 2509-2517, 2013. [10] A. Rezaei-Zare, "Enhanced transformer model for low- and midfrequency transients - Part I: Model Development," IEEE Transaction on Power Delivery, vol. 30, no. 1, pp. 307-315, January 2015. [11] S. Jazebi, S. E. Zirka, M. Lambert, A. Rezaei-Zare, N. Chiesa, Y. Moroz, X. Chen, M. Martinez-Duro, C. M. Arturi, E. P. Dick, A. Narang, R. A. Walling, J. Mahseredjian, J. A. Martinez and F. de León, "Duality Derived Transformer Models for Low-Frequency Electromagnetic Transients—Part I: Topological Models," IEEE Transactions on Power Delivery, vol. 31, no. 5, pp. 2410-2419, 2016. [12] S. Jazebi, A. Rezaei-Zare, L. Lambert, S. E. Zirka, N. Chiesa, Y. I. Moroz, X. Chen, M. Martinez-Duro, C. M. Arturi, E. P. Dick, A. Narang, R. A. Walling, J. Mahseredjian, J. A. Martinez and F. de León, "Duality-Derived Transformer Models for Low-Frequency View publication stats [24] H. K. Chisepo, L. D. Borrill and C. T. Gaunt, "Measurements show need for transformer core joint details in finite element modelling of GIC and dc effects," The International Journal for Computation and Mathematics in Electrical and Electronic Engineering, vol. 37, no. 3, https://doi.org/10.1108/COMPEL-11-2016-0511, pp. 1011-1028, 2018. [25] MIT Press, Magnetic circuits and transformers, Second ed., New York: John Wiley & Sons, 1944, p. 124. [26] EMTDC/PSCAD, "Classical Approach, Saturation Characteristic," HVDC Research Centre, Manotiba, 2005. [27] Thyssenkrupp, PowerCore® C 110-23 grain oriented electrical core steel, Duisburg, Germany, 2014. [28] A. Rezaei-Zare, "Enhanced transformer model for low- and midfrequency transients - Part II: Validation and simulation results," IEEE Transaction on Power Delivery, vol. 30, no. 1, pp. 316-325, January 2015b.