Embedded sensors for industrial transformers ... - IEEE Xplore

Embedded sensors for industrial transformers maintenance Francisco Poza, Perfecto Mariño, Santiago Otero, Vicente Pastoriza Electronic Technology Department University of Vigo 36200 Vigo, Spain E-mail:{fpoza,pmarino,jsotero,vpastoriza}@uvigo.es

Abstract— The Condition Monitoring of On-Load Tap Changers is very important because they have proved to be the elements with noticeable failures in a power transformer. This article describes the development of a portable virtual instrument for monitoring this kind of elements. The monitoring task is based in the measurement and analysis of the vibrations that a tap change produces. In contrast with other methods that can be used, this one has the advantage of being able to do continuous monitoring because the transformer can be operated on line.

I. I NTRODUCTION To be able to react to the different electric energy demands produced in an electric power distribution substation, it is necessary to make use of power transformers with on-load tap changers (OLTCs). The OLTCs have shown to be elements that need a more intensive maintenance and that cause more failures, from all the ones concerned in a power transformer [1][2]. This is because OLTC is the only mechanical mobile element that is present in a power transformer. To carry out a diagnosis of the OLTC, the information obtained measuring the contact resistance of the power transformer winding during the switching process can be used. This method is valid to detect possible damages in the switch contacts as well as in the changer resistors. Nevertheless this method is not valid to fulfill the continuous monitoring of the OLTC, since the power transformer must be out of service. Furthermore the OLTC has other mechanical elements that are fundamental for its correct operation such as the driving motor, springs, transmission systems, etc., that cannot be assessed using the previously commented method. There are other monitoring techniques that do not exhibit these drawbacks, for example the estimation of the power developed by the drive motor of the OLTC, and the measurement of vibrations generated during the change from one tap to another. This last one method seems to be useful for the condition assessment of OLTCs from different manufacturers [3][4][5]. The work presented in this paper is a part of an industry funded project, in which the development of a portable OLTC condition monitoring system for power transformers is carried out. The project, which is under development, includes several phases. The first one is the development of a portable virtual instrument that allows to carry out, in an easy way, the tap

changers operation tests in different OLTCs and to store the waveforms obtained from sensors for its subsequent analysis. The second one is the execution of a field measurements campaign in different OLTCs from several power transformers. The third one is the offline analysis of the measurements carried out to obtain the features that identify the OLTC correct operation. For this purpose digital signal processing techniques will be used (FFT, JTFA, wavelets, etc.) [4][5]. Finally the virtual instrument developed will be adapted to perform the condition monitoring of OLTCs making measures from time to time, and comparing the results with the patterns obtained in the previous phase. At present the project is in the second phase. This article focuses in the description of the hardware and software architecture of the portable virtual instrument, developed to carry out the acquisition and data storage and shows some initial digital signal analysis results. II. P ROPOSED APPROACH This project is not intended to develop a condition monitoring system that is found physically attached to only one power transformer to fulfill the continuous monitoring of its OLTC. On the contrary, one of the main requirements of the project is that the condition monitoring system should be portable, so that it could carry out the assessment of different OLTCs in power transformers located in electric power transformation substations distributed in different geographical areas. As it will be commented further on, the requirements for processing power and storage capacity of the portable equipment should be relatively high, therefore the developed virtual instrument is based on a notebook computer with a 1.7GHz Centrino processor, 1GB of RAM memory and 60GB of hard disk space. The used operating system is Microsoft’s Windows XP. A. Data acquisition The data acquisition part of the condition monitoring system is in charge of providing the measurement of different parameters, relative to the operation of the OLTC. The hardware used for data acquisition includes a set of sensors, for the measurement of physical phenomenon, conditioning equipments and a data acquisition card. The most

1318 0-7803-9701-0/06/$20.00 ©2006 IEEE

Fig. 1.

GUI aspect of the developed virtual instrument.

important characteristics of each one of them are commented next. 1) Sensors: The virtual instrument uses 8 analog sensors. Four Hall Effect sensors are used for measuring the voltage and current in two phases of the OLTC drive motor. To measure the vibrations that are produced when a tape changer operation is carried out, four piezoelectric accelerometers are attached in different locations close to the OLTC in the power transformers tank. The sensibility of the used accelerometers range from 1.006 to 1.043 pC/ms-2. 2) Conditioning: All the previously commented analog sensors have specific conditioning equipment, which adjusts the type and level of the output signal to values that match the input range of the data acquisition card and also isolate the sensor signal. The Hall Effect sensors used for measuring the intensity that run into the OLTC drive motor, are located in two current probes model PR30 from LEM. Those probes supply an instantaneous voltage that is proportional to the instantaneous intensity that flows through them. The sensibility of these elements is 100mV/A, with an input range of ±30A and a dynamic bandwidth ranging from DC to 100KHz.

The Hall Effect sensors used for measuring the voltage in the OLTC drive motor were mounted in a PCB card specifically designed for this application. This card supplies an instantaneous voltage that is proportional to the instantaneous voltage applied at its inputs. Its sensibility is 1/60. The charge signal provided by the accelerometers is conditioned with the commercial NEXUS equipment from Brüel&Kjær. It provides an instantaneous voltage ranging from -3.16 to +3.16V. B. Acquisition The data acquisition card used is the DAQCard-6062E from National Instruments. This card has PCMCIA architecture and presents the following technical characteristics: 16 common mode (8 differential mode) analog input channels, 12 bits resolution, 500KS/s and variable input range from ±0,05V to ±10V. This card has been chosen because it allows the acquisition of each one of the 8 differential analog input channels using a sampling rate of 50KS/s. With this sample rate it is possible to analyze harmonics from vibration signals up to 25KHz. For avoiding the missing of data that can be useful for doing an offline analysis, the duration of the data acquisition has been

2006 IEEE International Conference on Industrial Informatics

1319

Fig. 2.

Voltage and intensity signals from the drive motor (upper graphic). Vibration signal (lower graphic).

extended to all the time that the OLTC drive motor remains in operation. This time is in the range of 5 or 6 seconds. Keeping in mind that the data supplied by the acquisition card drivers (NI-DAQ) are double precision floats (8 bytes), this means that the size of the data acquired for all the 8 sensors are about 20MB for each one of the tap changes. C. Data Storage The data storage requirements are high because, as it was commented in the previous section, the data files are relatively large. Also, to be able to obtain reliable data for each one of the tape changes, it is necessary to perform various tests of each one. Assuming an OLTC with 20 contacts, from which 5 tests are taken both from contact 1 to 20 and vice versa, the total amount of data that must be stored is about 3,5GB for only one OLTC. In order to give the storage of acquired data several alternatives were analyzed. The option that used only files for storing all kind of data was rejected, because problems arose when the number of tests increase making difficult to found a particular test. Also the option of storing all the data in a database was tested. This option has the drawback that the size of the database grows very quickly with few tests, limiting the system efficiency.

1320

Finally it was decided to use a combination of both methods that can be able, on one hand, to easily carry out the search of a particular test, but without having a database with an excessively large size. For this reason a database with three tables was used: one to store the data of different power transformers, another for holding the data from different OLTCs, and finally another to retain the data concerning tests, such us: power transformer identification, test date-hour, sampling frequency, number of samples, start and end tap, number of tests and data file. This last field stores the path to the binary data file where really the acquired data are kept. Using, in this way, an adequate directory and file structure, the management of the tests is simple and fast. D. Virtual Instrument Software The virtual instrument software was developed using the LabVIEW 7.1 graphic programming environment from National Instruments [6][7][8]. This software provides the acquisition and data storage tasks. The user interface of the virtual instrument is formed by several tabs (Figure 1). Next it is briefly described how to carry out the tests. The power transformer to test is selected in the tab “General Data” using a list box where all the power transformers stored


Fig. 3.

Vibration signal(upper graphic), envelope (middle graphic) and CWT ridge plot of the envelope signature.

in database appear. Once it is selected all the OLTC data is shown, as well as the data from different tests executed in this OLTC. In this tab the user also can configure the necessary data to make a test: initial tap, end tap and number of tests to carry out in each one of the tap changes. Once the user starts the test, and so that he can verify that really the data for the current tap change is correct, in the tab “Test Information” the values for the start and end tape as well as the corresponding test number are shown. Also the system shows the user that it is waiting to start the data acquisition. Next, the user starts the tap change driving by hand the corresponding push-button situated in the cabinet that controls de OLTC. The acquisition starts and finishes when the conditions that the user has configured in the tab “Acquisition” are fulfilled. It is in this tab where the user can also configure other data acquisition characteristics such us: sampling rate, scales for the measurements and so on. Finally the virtual instrument stores in the database the corresponding test characteristics and creates a file in the suitable directory structure that stores the acquired data. This process is repeated until all taps from the initial one to the end one, and vice versa, have been tested the desired number of times. III. R ESULTS As it was already stated the industrial project is in progress and until now the portable virtual instrument, to carry out the acquisition and store of data for doing field tests, has been developed. Currently we are doing a field measurements campaign for obtaining enough data sets for offline analysis.

The figure 2 shows how the virtual instrument displays the different acquired signals. In the upper graphic the voltage and intensity signals from R phase of the OLTC drive motor can be seen. The lower graphic shows the vibration signal acquired by one of the accelerometers located in the lateral side of the transformer’s tank. As it can be seen this vibration signature is composed by various burst, each of which is produced by a specific contact movement of the OLTC. The figure shows that the real tap change duration is about 100ms. These signals were obtained moving the OLTC from tap 1 to tap 2. The virtual instrument developed is portable but at the same time it is very powerful and allows storing a lot of information for its subsequent analysis. To make easy the execution of the field tests, the virtual instrument provides the user the capability of defining different conditions for automatic starting and ending of acquisition tasks (analog rising or falling edge, entering or leaving window, etc.). For the acquisition starting the virtual instrument takes advantage of the hardware start of acquisition detection that is available in the data acquisition card. Nevertheless the detection of the acquisition end condition is executed by software, verifying all the values measured until they satisfy the condition. It is possible to start the acquisition based in the data obtained from one sensor and stop it based in the data obtained in another. This possibility allows a great degree of flexibility for establishing complex acquisition conditions. With the correct configuration of the acquisition start and end, the virtual instrument can acquire the data automatically, without user operation. In this way the tests are made easier because the user only must configure the initial and final tap number, optionally, the number of times that each tap change must be performed and


1321

push the correct button in the control cabinet. Currently we are analyzing some of the data acquired in various power transformers. The process we are doing with the vibration’s data includes a Hilbert transform for obtaining the envelope of the signal, after the envelope is normalized and finally the continuous wavelet transform (CWT) using the fourth order Symlets wavelet [9][10] is applied, for detecting the singularities in the vibration signal. Figure 3 shows this process applied to the vibration signal obtained when the tap changer operate from tap 3 to 4. As the figure shows, the vibration signature of this type of OLTC contains six dominant transient bursts. The upper graphic shows the vibration signal, the middle one depicts the envelop detected using the Hilbert transform and finally, the lower one provides the CWT ridge plot of the envelope signature. R EFERENCES

[2] Kang P, Birtwhistle D, Daly J and McCulloch D, “Non-invasive on-line condition monitoring of on-load tap-changers”, Proceedings of IEEE Power Engineering Society Winter Meeting, 2000, Singapore. [3] Bengtsson T, Kols H, Foata M and Léonard F, “Monitoring Tap Changer Operations”, Cigre 1998, paper 12-209. [4] Kang P, Bitwhistle D, “Condition Assessment of Power Transformer On-Load Tap-Changers Using Wavelet Analysis”, IEEE Transactions on Power Delivery, Vol. 16, No 3, July 2001. [5] Kang P, Bitwhistle D, “Condition Assessment of Power Transformer OnLoad Tap-Changers Using Wavelet Analysis and Self-Organizing Map: Field Evaluation”, IEEE Transactions on Power Delivery, Vol. 18, No 1, January 2003. [6] LabVIEW 7 Express. Measurements Manual, April 2003 ed., National Instruments Corporation, 2003, part Number: 322661B-01. [7] LabVIEW 7 Express. User Manual, April 2003 ed., National Instruments Corporation, 2003, part Number: 320999E-01. [8] B. Mihura, LabVIEW for data acquisition, National Instruments virtual instrumentation series. Upper Saddle River, NJ 07458: Prentice Hall, 2001. [9] C. K. Chui, An introduction to wavelets, Academic Press, 1992. [10] M. Misiti, Y. Misiti, G. Oppenheim, J.M. Poggi, Wavelet Toolbox User’s Guide, The MathWorks, 2005.

[1] CIGRE SC 12 WG 12.05, “An international survey on failures in large power transformers in service”, ELECTRA, no 88, pp. 21-47, 1983.

1322
