Contents lists available at ScienceDirect Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes Importance of core joints in GIC/dc studies with scaled down laboratory transformers Leslie D. Borrill a , Hilary K. Chisepo b, ⁎ , C.T. Gaunt b a Eskom Holdings SOC, Koeberg Nuclear Power Station, Cape Town, South Africa b Department of Electrical Engineering, University of Cape Town, South Africa ARTICLE INFO Keywords: Core joints Equivalent circuit FEM Saturation Stray fux Transformer inductance ABSTRACT The response of large power transformers to geomagnetically induced currents (GICs) cannot be tested fully in factories because of the signifcant reactive power required. While carrying out physical tests on scaled-down models of a power transformer, we discovered that models with high-grade electrical core steel and even a transformer tank, as widely used in investigating the bias efects of GICs and leakage dc, cannot replicate adequately the actual performance if they are built without relevant construction details. Laboratory testing and FEM simulation of three single-phase four-limb transformers show that transformer models need multi-step lap mitered joints similar to those in full-scale power transformers to reproduce correctly the saturation response to dc components. FEM simulation also shows the change of transformer inductance when driven into deep sa- turation, as required in equivalent circuit models. The approach of achieving consistency between equivalent circuit, FEM and physical models even in the presence of saturable core joints is a novel contribution which provides guidance for future research in slow transient phenomenon studies. 1. Introduction Geomagnetically induced currents (GICs) are low frequency tran- sient currents (1 mHz to 100 mHz) that can be treated as direct current in some power frequency simulation tests. Both GICs and stray dc (such as from HVDC and traction systems) can drive transformers into part- wave saturation with potential to disrupt high voltage power networks [1–3]. Spurred by concerns related to GICs [4,5] and to better under- stand the risk of damage, various approaches have been followed in investigations of the low frequency transient response of transformers and their magnetic cores. Full-size and scale model tests injecting dc into transformers with various winding and core structures [6–9] demonstrate responses to part-wave saturation, including the generation of harmonics and re- active power, or non-active power (as defned by IEEE Std. 1459:2010 [10]). DC injection tests in power transformers, however, are not rou- tinely performed by manufacturers because they may not have the re- quired reactive power capacity for the tests with high levels of dc, or for fear of damaging a transformer [11]. The large physical size and weight of generator step up transformers of more than about 900 MVA rating make them difcult to transport. Single phase transformers ofer a more convenient alternative, but are described as being the most susceptible to dc injection [7]. Undesirable phenomena, such as waveform distortion by a transformer driven into part-wave saturation by dc or GIC, could potentially interfere with the safe operation of power stations and justify rigorous investigation consistent with that required by nuclear regulation [12]. Accordingly, this research set out to develop a duality-derived to- pology-based transformer model of a single phase four-limb (1p4L) transformer for which no model existed. Such models are used ex- tensively for low frequency electromagnetic transient modelling [13–15] where the duality theorem [16] allows. The parameters needed for GIC and dc bias models include the core magnetizing inductance, core loss resistance and piecewise linear hys- teresis curve with two slopes; and the bulk leakage, and winding re- sistances apportioned according to their dc resistances. Fig. 1 illustrates the derived equivalent circuit of a 1p4L model transformer for which some parameters can be measured using suitable test circuits and others determined from the core dimensions. Subsequent to our research being completed we received a paper [17] with a magnetic circuit similar to Fig. 1 (left), but without an electrical equivalent circuit. The physical core parameters of most power transformer are not available and ‘gray box’ models are useful for deriving slow transient transformer models. Test circuit requirements for determining the ‘gray box’ terminal saturation inductance of a large single-phase power transformer [19,20] are arduous and, naturally, utilities are unwilling https://doi.org/10.1016/j.ijepes.2020.105974 Received 28 August 2019; Received in revised form 24 January 2020; Accepted 28 February 2020 ⁎ Corresponding author. E-mail address: hilary.chisepo@uct.ac.za (H.K. Chisepo). Electrical Power and Energy Systems 120 (2020) 105974 0142-0615/ © 2020 Elsevier Ltd. All rights reserved. T