Remanence detection and low-frequency degaussing for power transformers based on modified J-A model

Abstract

I . INTRODUCTIONInrush currents are high magnitude currents generated typically during energization of a transformer, being a large transient caused by saturation of the magnetic core of the transformer . The magnitudes of the inrush currents can be initially several times the rated load current which will cause error actions of the relay protection equipment [1] . When a transformer is put into operation at the optimal angle, the residual flux becomes the main factor leading to inrush currents . Measuring methods on residual flux before often had to saturate transformers during the detection and were complex and not accurate enough [2] . Besides, in past AC-degaussing methods, high-power supplies were used but hard to carry [3] . In this paper, based on the modified Jiles-Atherton model, the effects of eddy current and temperature were taken into account to study the hysteresis characteristic of any point on B-H loops . Then, as long as the initial demagnetization point is known from voltage and current waveforms in practice, the terminal of demagnetization, or the residual flux, can be determined . A low-frequency degaussing method is introduced to reduce the source power, but it'll take more time when compared with the degaussing under the normal frequency . Accordingly, the multi objective programming approach was adopted to determine an optimal frequency trading off between power demands and degaussing time . What's more, to assess the effectiveness of both residual flux measuring and degaussing methods, an experiment platform was built eventually . II . ONE REMANENCE DETECTION WAY BASED ON JILES-ATHERTON MODELAfter the microcosmic magnetization process and dimensional consistency were analyzed, the modified Jiles-Atherton model can be derived as follows [4] . dM/dH=(1-c)(M an -M irr )/[δk/(1-c)-α(M an -M)+cdM an /dH] (1) Then, revise the model to incorporate the effects of classical eddy current and excess losses caused by the temperature shift as below [5] . H = H hyst + H cls + H exc (2) A steel sample with 3% carbon (M s =1 .7049×10 6 A/m, a=987 A/m, α=1 .027×10 -4 , k=1835 A/m, c=0 .0953) was used to obtain a family of M-H loops in Fig . 1(a) . The source waveform was a 50 Hz sine wave . As shown in the graph, the residual fluxes are almost the same when demagnetizing from points in saturated zone . The most important reason causing the residual flux in a transformer is DC tests before its re-operation . Accordingly, certain modified square waves with different amplitudes were applied to the input of J-A model in Simulink, thus B-H curve can be obtained in Fig . 1(b) . When the magnitude of H varies from 1000 A/m to 2000 A/m the residual flux increases sharply . However, with the magnitude of H rising sequentially, the residual flux grows slowly and even barely budges in the end . In general, demagnetization curves can help easily locate the residual flux, as long as the initial demagnetization point is acquired . The initial demagnetization point is determined by the current magnitude used in DC t...