b'Engineer Innovation | ElectronicsHaving tested FEM transformer-GIC models in Simcenter MAGNET and achieved satisfactory correlation with measured recordings (apart from the inexplicable underestimated no load magnetizing current when compared against measurements), the next phase of the research involved the modeling of practical transformers relevant to the manufacturing and utility industries. The main issue with FEM modeling in transformer research with AC-DC excitation is that Unfortunately, no measurement results are available to validate the [3D FEM] analysis, [6]. This work, therefore, involved exhaustive measurements on larger scale single-Figure 4. Primary current waveforms for 1p3L and 3p3L test transformers with 550 mA DC andphase four limb (1p4L) transformers to 160 mA DC per phase respectively investigate key parameters for saturated transformers operating undertopological and FEM modeling. the influence of GIC or simultaneous AC-DC excitation. Figure 4 shows theExplicitly modeling individualmeasured and simulated current(0.25-0.3 mm) laminations of waveforms for the single-phase andtransformer cores often results in FEM three-phase three limb transformers. Ascalculations failing to converge expected, the three-phase three-limbbecause of excessively large mesh (3p3L) bench-scale transformer (withoutstructures [7]. In the context of AC and a tank) did not generate harmonics withDC, this problem was overcome using various levels of DC, because a zeroan approach that models laminations sequence path to a tank is necessary forexplicitly only close to the core surface Figure 5. DC time response of the 3p5L modelDC flux to flow in and out of thiswith the rest of the core treated as transformer with 160 mA in the neutral particular core structure, compared withsolid but without sacrificing the air gap the single phase and the five limb coredetails at the transformer joints with types that have free return limbs tothe appropriate boundary conditions complete the DC flux return path [3, 4].[8]. Figures 6 and 7 (1/8th symmetric The single-phase three limb (1p3L) FEMmodel of a 1p4L test transformer) show model generated multiple even and oddthe difference in flux distribution when harmonics matching the measuredmore detail is added to the core distorted current waveform quite closely.surface and core joints, compared with a solid core. The application of these An adjacent study introduced themethods that include air gap details at concept of transformer time response inthe joints resulted in more accurate the calculations of GIC in the Southernestimations of the no load magnetizing African networks [5]. The time responsecurrents and also a better FEM of a FEM model of a three-phase fiveinterpretation of the flux distributions limb (3p5L) transformer with DC waswith AC only and with AC and DC. The also tested against laboratorymodel shown in Figure 6 (right) and observations, yielding consistent results.zoomed in for Figure 7 gave the results Figure 5 shows that three phase linethat are closest to all the physical currents are not equal in all the phasesmeasurement data, including reactive during the transient period upon DCpower, terminal saturation injection and that they reach steadyinductance, and even the leakage flux state after 12 s. This is consistent, notdistribution measured with search coils only with the times recorded in thein the air spaces. The terminal laboratory, but also confirms the ideasaturation inductance is a key that the transformer time constant is aparameter in testing topological function of the magnitude of the GICtransformer models for GIC and slow [5]. The FEM analysis, therefore,transients, but it cannot be measured validates the improved GIC calculationreadily in any transformer factory. approach that includes the timeWithout the validated FEM or special response of power transformers.measurement protocols, this parameter 38'