Paul Jarman

With the growing interest in monitoring and diagnostic techniques for assessing transformer condition, dissolved gas analysis (DGA) stands out as a crucial diagnostic tool to detect electrical and thermal faults in liquid‐immersed transformers. Laboratory‐based DGA studies have primarily focused on detecting gas concentrations and patterns under simulated fault conditions in different types of insulating liquids. To understand the significance of dissolved gas values it is invaluable to establish the quantitative relationship between the gas generation amount and the electrical energy associated with the fault. This study aims to systematically investigate the correlation between gas generation characteristics and the electrical energy associated with either partial discharges (PD) or breakdown sparks (BD) in a mineral oil (MO) and a synthetic ester (SE). An in‐house designed experimental setup was developed, and various influencing factors were explored. The results of DGA‐PD tests indicated that PD magnitude influences the fault gas volume per unit energy (GVPUE, μL/J) generated in both liquids. The GVPUE in MO was found to be higher than that in SE for the same PD magnitude. The results of DGA‐BD tests showed that larger breakdown gap distances lead to a higher GVPUE and a lower H ₂ /C ₂ H ₂ gas ratio. The findings bring new insights on fault gas generation mechanisms in transformer liquids.

An increasing number of power transformers operating in the developed countries do not have OEM technical support. There is an urgent need to obtain transformer design information such as winding types through interpretation of measured FRA data. Winding types may be manifested as different features in FRA traces within specific frequency ranges. The objective of this paper is to identify the winding type via FRA measurement data in order to help manage transformer assets. Support Vector Machines (SVM) are supervised learning models for classification analysis in machine learning. Given a set of training FRA examples with known winding categories, an SVM algorithm is built to assign new FRA traces to a most possible winding category. A group of 400/275 kV transformers are tested in this study with excellent classification results of winding types, indicating a promising “digital twin” technological approach to transformer asset management.

Currently there is no standard FRA interpretation methods and not all the FRA practitioners are capable to model transformer windings using equivalent circuits. A low-frequency transformer model of a three-limb core-type 1000 MVA 400/275/13 kV auto-transformer is established to analyze the effects of transformer electric parameters on the FRA plots. The equivalent electrical circuit of the magnetic core is derived based on the Principle of Duality. Air-core inductance and winding capacitances are included in the model. FRA simulation results are validated by measurement results, the suitability of transformer model is discussed and the dominant factors in different frequency regions are concluded through sensitivity studies.

Frequency response analysis (FRA) is an effective tool to detect winding deformation. If variation occurs to the FRA trace, a fault may be indicated. A new method to estimate the transfer function from FRA measurement results on large bandwidth is proposed. A MATLAB program is developed to auto-process the measured data and to obtain the transfer function expression. Measurement results from large frequency ranges are divided into several ranges, and key information is abstracted from each range to form a new transfer function. Afterwards, correction is still needed to eliminate the difference between estimated transfer function and measured data. This method is verified as valid for estimation of transfer functions of 48 groups of FRA data of a 400/275 kV auto-transformer.

Experiments were carried out on transformer oil contaminated by cellulose fibre. After the high voltage was applied to the electrodes, the pressboard particles started to accumulate charge from one electrode and discharge to the other electrode. We have developed a mathematical model to simulate the charging and discharging dynamics. The dielectrophoretic, coulomb and drag forces are considered for simulation of the particles motion. The predicted particle movement path and the velocity between the electrodes showed similar trend to the experimental results. The model is a successful means of predicting particle movement phenomenon compared with the experimental data.

Oil-flow distribution in the winding has a direct impact on cooling performance. In addition, static pressure drop over the winding determines oil split among windings connected hydraulically in parallel. In this paper, experimental verifications are provided to support computational fluid dynamics (CFD) simulations for disc-type windings in oil-forced and directed (OD) cooling modes. Oil-flow distribution in and pressure drop over disc-type winding models are measured using a particle image velocimetry system and a differential pressure instrument, respectively. Dimensional analysis is adopted to analyze the relationship between flow distribution, or pressure drop, and the controlling parameters. CFD parametric sweeps of the dimensionless parameters obtained from the dimensional analysis are conducted and the CFD results are then correlated with the dimensionless parameters. The comparisons between measured results and corresponding results obtained from the correlations demonstrate constant consistency, proving the validity of both the method of dimensional analysis and the correlations. Finally, comparisons of experimental results from isothermal and nonisothermal conditions in OD cooling modes are executed, which show that the isothermal conclusions can be extended to nonisothermal cases because the effects of buoyancy force and hot-streak dynamics prove to be negligible.

This paper introduces a network transient simulation study for the UK South-West Peninsula transmission system using the transient software ATP/EMTP. Assume that the system suffers a geomagnetic disturbance with a constant 5V/km Southwest-Northeast electric field, severities of the GIC in terms of steady state DC current magnitude, extra reactive power losses experienced and time to reach steady state by transformers in the system were firstly quantified. Using typical field waveform in IEEE guide [1], real-time GIC effects on transformers were simulated to illustrate the importance of transient study, which gave more realistic GIC current magnitudes and extra reactive power losses. The system simulation also revealed that at the same substation the 5-limb transformers are more vulnerable to GIC than 3-limb transformers; and the tertiary delta winding delays the core saturation and results in a lower reactive power loss. It is validated that the GIC level suffered by individual transformer is determined by multiple factors including the location of the substation; the characteristic of the load; the direction, magnitude and waveform of the quasi-DC electric field.

Dissolved Gas Analysis (DGA) method is one of the most effective techniques to monitor the condition of transformers. In this method, free gas leakage condition is required as any leakage of dissolved gases might distort the DGA results and mislead the diagnosis of transformer fault. In this paper, a mass transfer model based on two-film theory is built to simulate the gas leakage process from transformer oil when it is in contact with air. The mass transfer coefficients of dissolved fault gases in mineral oil have been calculated at different temperatures based on an experiment. The mass transfer coefficient at certain temperature can be predicted by the Arrhenius equation. According to the ascertained mass transfer coefficient, oil surface and volume, gas leakage from a sampling rate can be estimated. For practical application, the gas leakage from oil sampled by bottle is quantified.

In this paper dimensional analyses of flow distribution in and pressure drop over disc-type transformer windings in an oil forced and directed (OD) cooling mode are presented. Computational fluid dynamics (CFD) parametric sweeps were conducted to quantify the dimensionless relationships obtained from the dimensional analyses to form predictive correlation equation sets. With the help of the correlation equation sets, the influences of coolant types including a mineral oil, a gas-to-liquid (GTL) oil and a synthetic ester on flow distribution in and pressure drop over a winding model were investigated. It was found that the synthetic ester, which has a higher viscosity than the others, could be advantageous in terms of suppressing reverse flow and getting a relatively more uniform flow distribution at the cost of a higher static pressure drop.