The investigation of frequency response analysis for power transformers winding condition

In modern electrical transmission and distribution systems, power transformers are critical components within the network. In the event that a failure occurs in service, the impact can be far reaching. The majority of mechanical deformation within power transformers is due to short circuit faults. The ageing transformer population increases the likelihood of failure so a reliable diagnostic tools required to determine the remaining life of these assets. Although, frequency response analysis (FRA) has been recently recognized as the most reliable detection tool for mechanical deformation in transformers, in the event of minor fault, the current FRA interpretation approach may not able to detect any variations between healthy and faulty FRA signatures. This paper focuses on FRA tests for power transformer and interpretation of obtained FRA signatures. One of the most important problems of transformer maintenance is the monitoring system and technical diagnosis development. More than 40 % of transformers have exceeded their design life (25y, established by standard papers [1]). Nowadays, there are several diagnostic techniques for power transformers, for example, vibration analysis, ultrasonic contact fault detection, dielectric loss angle test and partial discharge [2]. The diagnosis of power transformer can be carried out offline, online and partly. Parameters used for online transformer tests are temperature, pressure, leakage currents, vibration amplitude, noise level, velocity vibration and acceleration vibration. They must have properties, such sensitivity, accuracy of measurement, stability, depth of investigation and simplicity of diagnosis. The transformer defects occur due to unsuitable means of transporting short circuit currents, earthquakes, oil ignition in tank etc [3]. Due to CIGRE working group A 2.26 the most common cause of power transformers fault is inter-disk faults [4]. One of the most effective techniques of condition monitoring for power transformers are Sweep Frequency Response Analysis (SFRA) and Impulse Frequency Response Analysis (IFRA) [5]. Lots of investigations are held to improve this technique at the moment, but from 1978 there is no general opinion how to make a conclusion about technical condition [6].The main advantage of this technique is its high sensitivity: even minor local changes of winding elements(turns, disks) result in dramatic changes on diagnosis signature [7]. Different types of deformations lead to changes in FRA signature in different range. Conventional techniques of condition monitoring are not so sensitive. That’s why FRA is the best technique for inter-disk fault detection. Frequency Response Analysis has two directions of use: sweep and impulse [3]. IFRA is provided by low voltage impulse given on winding (The impulse form can be rectangular or conventional or exponential). Current or voltage on bushing terminals or frequency current transformers of other windings is measured. The transition process in windings is recorded with the help of analog-digital converters. After that the response is transformed to frequency area using Fourier algorithm. Finally, transfer function as a ratio of injected and measured signal is founded. SFRA is used in frequency area directly. In this case the sine 10 V signal from sweep generator is injected on the bushing terminal. It’s changes from several Hz to several MHz. The response is measured from other terminals as a function of the variable frequency [3]. Each way has its advantages and disadvantages. For example, SFRA is more sensitive on low frequencies and quite simple maintenance as much as in interpretation results [8]. However the process takes long time. A foreign power transformer was selected as an object of study. Its characteristics are given in table 1. This transformer was chosen because of its initial data, which were used for modelling. Table 1. The power transformer parameters No Classification Value 1 Rated power 40 MVA

Abstract. In modern electrical transmission and distribution systems, power transformers are critical components within the network. In the event that a failure occurs in service, the impact can be far reaching. The majority of mechanical deformation within power transformers is due to short circuit faults. The ageing transformer population increases the likelihood of failure so a reliable diagnostic tools required to determine the remaining life of these assets. Although, frequency response analysis (FRA) has been recently recognized as the most reliable detection tool for mechanical deformation in transformers, in the event of minor fault, the current FRA interpretation approach may not able to detect any variations between healthy and faulty FRA signatures. This paper focuses on FRA tests for power transformer and interpretation of obtained FRA signatures.
One of the most important problems of transformer maintenance is the monitoring system and technical diagnosis development. More than 40 % of transformers have exceeded their design life (25y, established by standard papers [1]). Nowadays, there are several diagnostic techniques for power transformers, for example, vibration analysis, ultrasonic contact fault detection, dielectric loss angle test and partial discharge [2]. The diagnosis of power transformer can be carried out offline, online and partly. Parameters used for online transformer tests are temperature, pressure, leakage currents, vibration amplitude, noise level, velocity vibration and acceleration vibration. They must have properties, such sensitivity, accuracy of measurement, stability, depth of investigation and simplicity of diagnosis. The transformer defects occur due to unsuitable means of transporting short circuit currents, earthquakes, oil ignition in tank etc [3]. Due to CIGRE working group A 2.26 the most common cause of power transformers fault is inter-disk faults [4].
One of the most effective techniques of condition monitoring for power transformers are Sweep Frequency Response Analysis (SFRA) and Impulse Frequency Response Analysis (IFRA) [5]. Lots of investigations are held to improve this technique at the moment, but from 1978 there is no general opinion how to make a conclusion about technical condition [6].The main advantage of this technique is its high sensitivity: even minor local changes of winding elements(turns, disks) result in dramatic changes on diagnosis signature [7]. Different types of deformations lead to changes in FRA signature in different range. Conventional techniques of condition monitoring are not so sensitive. That's why FRA is the best technique for inter-disk fault detection.
Frequency Response Analysis has two directions of use: sweep and impulse [3].
IFRA is provided by low voltage impulse given on winding (The impulse form can be rectangular or conventional or exponential). Current or voltage on bushing terminals or frequency current transformers of other windings is measured. The transition process in windings is recorded with the help of analog-digital converters. After that the response is transformed to frequency area using Fourier algorithm. Finally, transfer function as a ratio of injected and measured signal is founded.
SFRA is used in frequency area directly. In this case the sine 10 V signal from sweep generator is injected on the bushing terminal. It's changes from several Hz to several MHz. The response is measured from other terminals as a function of the variable frequency [3]. Each way has its advantages and disadvantages. For example, SFRA is more sensitive on low frequencies and quite simple maintenance as much as in interpretation results [8]. However the process takes long time.
A foreign power transformer was selected as an object of study. Its characteristics are given in table 1. This transformer was chosen because of its initial data, which were used for modelling. The parameters of the frequency characteristics with the parameters of the winding could be correlated because the power transformer winding is a multi-frequency oscillatory circuit [9]. It consists of series resistance, series inductance and series capacitances. A transformer winding equivalent circuit was compiled in the Multisim software package to carry out the experiments ( fig.1).Transformer parameters are:  series inductance and series resistance of HV and LV windings (Ls, Rs);  shunt capacitance between HV and LV windings (Csh);  series capacitances between HV / LV windings and the earthed tank / core (Co);  series capacitances between the HV and LV windings (CHL).  The equivalent circuit of the first HV winding disk is shown in Fig. 2.   Fig. 2. The equivalent circuit of the first HV winding disk C1, C6 (Co) -capacitance between first HV winding disk and the earthed tank C2 (Csh) -capacitance between first and second HV winding disk. C3, C8 (CHL) -capacitance between first HV winding disk and LV winding. L1 (Ls) -inductance of the first HV winding disk. R1 (Rs) -resistance of the first HV winding disk.
To conduct tests, the elements implemented in Multisim in the form of blocks were used. They are:  The generator of sine signal, which allows to change the amplitude, frequency and duration of the signal. The amplitude is10 V.  BodePlotter, used to get FRA signature.  Load resistance, which consists of the cable resistance and measuring instrument resistance, etc. The view of the BodePlotter block is shown in Fig.  3.

Fig. 3. BodePlotter block
The view of the generator of sine signal block is shown in Fig. 4.  During FRA test the transfer function -frequency response was found. This function is the ratio of injected signal on the top of the winding to the measured signal from the winding end. It's a plot, defined by frequency Each of 10 HV winding disks was short-circuited in this test.
Under the test capacitance Csh (С2) was enlarged in 100 times. It was made to ensure accurate results. At one plot damaged transformer FRA signature and healthy transformer FRA signature were placed. (fig. 6, fig.7 and  fig. 8). Fig. 6 shows: healthy transformer FRA signature, the first, the second and the third short-circuited winding disk FRA signature. Fig. 7 shows: healthy transformer FRA signature, the fourth, the fith, the sixth and the seventh short-circuited winding disk FRA signature.  windings condition, 4 -fourth disk winding short-circuited, 5 -fifth disk winding short-circuited, 6 -sixth disk winding shortcircuited, 7 -seventh disk winding short-circuited. To compare damaged and healthy FRA signatures it is noticed:  before 2 MHz changes in curves are noticeable; The first resonance frequency on heathy winding is about ~320 кГц, and the first antiresonanse is ~800 кГц; When frequency changes from 1,5 MHz to 4 MHz there are lots of resonanses and antiresonanses. When different disk winding short-circuited the next thing should be marked:  fig.6 shows the growth of resonance amplitude at displacement from 1 disk to 3 disk on ~680 kHz;  fig. 7 shows the growth both resonance and antiresonanse displacement amplitude at the middle of the winding( from 4 disk to 7 disk);  The decrease of resonance amplitude is observed at the end of the winding (from 8 to 10 disk);  The first resonance frequency varies from 500 kHz to 600 kHz, depending of the fault place. The changes above were observed at ~680 kHz. The amplitude varies from 18 to -43 dB. After that, the signal is decreasing.