skip to Main Content
Smart Solutions for Smart Grids


The importance of transformers in the SS Margem Direita of Itaipu for supply of power to Paraguay led to the installation of a pilot project for on-line monitoring in November 2006, which included the monitoring of regulator transformers R1 and R4, and three 500kV pedestal CTs. This work presents the monitoring functions applied to each piece of equipment, as well as the architecture used, solutions employed in the installation and results achieved


Treetech Sistemas Digitais Ltda. Marcos E. G. Alves


Itaipu Hydroelectric Power Plant is a bi-national undertaking between Paraguay and Brazil, and is in charge of the supply of approximately 90% of the power consumed in Paraguay, which is made via Substation Margem Direita – SEMD. It is located within the Itaipu premises, where there are four transformer sets, each one comprised by one 3-phase auto-transformer, 550/245kV, and one 3-phase Regulator Transformer 245/245kV 375MVA. By considering the great importance of these transformers for the Paraguayan power system and the work conditions which they are submitted to, such as high ambient temperatures, normal dielectric demands from the operation and high loads, especially at peak hours, it has been considered of great importance the on-line monitoring of these pieces of equipment. The purpose is to reduce the risk of failures, as the monitoring has the purpose of detecting failures in an incipient stage, which is not always possible with preventive maintenance only.


The deployment of on-line monitoring of transformers at the Margem Direita substation started with the installation of a pilot system in Regulator Transformer R1, shown in figure 1, followed by an expansion of the same system for monitoring of the gas dissolved in oil of Regulator R4.

To do that, it was considered the previous experience of Itaipu with the maintenance of such equipment, as well as the transformer subsystems that present the greatest impacts for operation in case of failures, as detailed below.


Fig. 1 – Regulator transformer R1

2.1 - Bushing Subsystem - Monitoring the Capacitance and Tangent Delta

During normal operation of a transformer, several external phenomena can originate transient overvoltages, such as switching surges and atmospheric impulses. Due to the interface function they perform in this equipment, bushings are naturally the first ones to receive these stresses on their dielectric.

Associated to the normal operational voltages that they are permanently subject to, this fact makes that bushings are pointed out in the statistics as one of the major failure points in power transformers. An international survey carried out by Cigré revealed that failures started by the bushings represented 12.3% of all the forced and scheduled disconnections on transformers equipped with on-load tap changer at substations [1].

By the other side, the failure of dielectric in a bushing is a situation that poses extreme risk to people occasionally close to the equipment, in addition to cause severe damages to the transformer where it is installed on, which may cause its total loss in extreme cases.

Due to these reasons, the on-line monitoring of bushing conditions was one of the subsystems selected for regulator transformers at the Margem Direita substation. Since the high-voltage bushings have capacitive type construction, with several insulating layers interleaved with conductive layers, as shown in figure 2, the on-line measurement of capacitance and tangent delta changes in the phase-earth insulation enables the early detection of most of the defects [2].

corte da bucha – artigo eletronorte -eng

Fig. 2 – Representation of the radial section of a capacitive bushing

In case of regulator transformers, a special feature – the existence of Bushing Potential Devices (BPD) connected to the load-side bushing taps – demanded the application of a bushing monitoring system specially adapted.

In normal applications, where there is no BPD, the monitoring system is directly connected to the bushing tap in order to measuring the dielectric’s fault current  and thus monitoring its capacitance and tangent delta changes. This application is shown in figure 3, where one of the Supply side bushings is illustrated.


Fig. 3 – Direct connection of the monitoring system to the bushing tap of the Supply side

By their turn, in bushings of the Load side, where BPDs are connected to, the tap cannot be used for monitoring purposes. For these applications, the monitoring system manufacturer developed a special solution, which uses the BPD outputs (nominal voltage 115V) for measuring purposes. In this case, only the capacitances are monitored, as tangent delta is very sensitive to inaccuracies introduced by the internal circuits of BPDs.

In case of evolution of the bushing defects, the monitoring system provides several alarm levels, which provide to the maintenance engineering elements for decision-making in terms of the corrective measures to be adopted:

  • Alarm by tangent delta evolution trend, with indication of the expected time, in days, for occurrence of the two next alarms.
  • Alarm by high tangent delta.
  • Alarm by very high tangent delta.
  • Alarm by high capacitance evolution trend, with indication of the expected time, in days, for occurrence of the two next alarms.
  • Alarm by high capacitance.
  • Alarm by very high capacitance.
  • Auto-diagnostic alarms of the monitoring system, including loss of signal from the bushing tap.

2.2 - Subsystem Active Part - Hydrogen Monitoring

The measurement and analysis of gases dissolved in oil of power transformers has been used for decades for diagnostic of internal defects in the equipment; such analysis can detect failures related to overheating, partial discharges, internal arcs, among others. Traditionally, this analysis is performed by removal of oils samples, which are sent to laboratory for determining the gases dissolved in oil by gas chromatography test, which measures separately the concentration of each fuel gas. Among the several gases generated by these faults, hydrogen highlights as a key gas, as it is generated in almost all the internal defects in a transformer.

By leveraging this characteristic, and due to the high cost for deploying on-line monitoring with individual measurement of each gas, the monitoring system frequently use an on-line hydrogen sensor dissolved in oil.  Once detected an increase of the hydrogen content, the monitoring system generates an alarm that indicates to maintenance engineering the need for removing an oil sample to be submitted to chromatography in the laboratory. The results from these tests are included in the Chromatography Engineering Module of the system, which performs the off-line analysis of gases and issues a report based on the analysis criteria most accepted in the market, such as IEC60599 standard and Duval method, among others chosen by the user.

In the case of the pilot project for regulator transformers, the on-line monitoring of hydrogen in oil has been proven by using an existing sensor in Regulator R4, shown in figure 4, which was integrated to the system by using the existing infrastructure for the Regulator R1. By doing that, the modularity and scalability of the decentralized architecture could be also proven for the open monitoring system used, which also enables the integration of third-party sensors.


Fig. 4 – Integration of the existing gas monitor in the regulator R4 to the monitoring system. (a) Hydrogen sensor; (b) Data acquisition module for integration

2.3 - Subsystem Active Part - Thermal Aging Monitoring

The major material used for insulation of the windings of power transformers is paper, which cellulose (pulp) fibers are comprised by long chains of glucose rings, as shown in figure 5. The number of rings in this chain is the Polymerization Grade (GP), which is about 1000 to 1400 in new insulations. The material degradation causes the gradual reduction of GP, by usually considering that the lifetime end was reached when this number is below 200. Although the dielectric characteristics of the insulation do not change substantially when this occurs, its mechanical properties weaken in a significant way, making the transformer susceptible to failure upon the occurrence of mechanical stresses, for example, in case of short-circuit in a transmission line.


Fig. 5 – Glucose rings constituents of the cellulose molecule

Three major mechanisms can contribute for the cellulose degradation: pyrolysis, hydrolysis and oxidation [3], which are intensified, respectively, with temperature raise, water content and oxygen content. With oxygen and moisture under controlled conditions, pyrolysis then becomes the predominant factor for shortening of the insulation lifetime.

As the temperature is not evenly distributed on a transformer, the monitoring system uses the temperature in the hottest spot of the winding to calculate the lifetime shortening on-line, as this is the place that will suffer the greatest degradation. To do that, a Temperature Monitor has been installed in Regulator R1, which calculates the winding temperature from measurements of temperature on the oil top and load current, as shown in figure 6.

In addition to the remaining lifetime percentage, the system also determines the remaining time to reach the theoretical lifetime end, based on the mean lifetime loss rate within a time period selectable by the user. When the remaining time is less than a threshold adjusted, the monitoring system generates an alarm informing the maintenance engineering on the need for corrective actions.


Fig. 6 – Temperature measurement. (a) Temperature sensor on the oil top; (b) Temperature monitor for oil and in the winding hot spot; (c) Ambient temperature sensor

2.4 - Subsystem Active Part - Aging Monitoring by Water on Paper

As mentioned in item 2.3, one of the phenomena that causes deterioration of the insulating paper of windings is the moisture, in a so-called hydrolysis phenomenon. Fabre and Pichon [4] proved that hydrolysis acts as an accelerator factor for loss of the insulation thermal lifetime (pyrolysis).

Therefore, the water content on the insulating paper is an important parameter for monitoring the insulation lifetime loss, and its determination is made from an Engineering Module in the monitoring system software, a mathematical model that uses the oil moisture and the transformer’s oil and winding temperatures as input variables. Moisture is measured by a sensor immersed in oil, installed on a valve, as shown in figure 7, and the temperatures are achieved from the Temperature Monitor, as already shown in figure 6.

In case of detecting high water content in oil, in order to cause accelerated lifetime loss of the insulation, the monitoring system generates and alarm, informing the maintenance engineering on the need for corrective actions.


Fig. 7 – Moisture sensor on the draining valve of regulator R1

2.5 - Subsystem Active Part - Bubble Formation Monitoring

In addition to accelerated aging of the insulation, excess water on paper poses the risk of bubble formation in the presence of high temperatures, for example, due to emergency overloads. As regions with low dielectric strength are created, the evolution of bubbles from the water present on paper represents a risk of dielectric failure of the equipment, due to the very intense electric fields between coils.

Based on the water content on paper, calculated in the Engineering Module mentioned in item 2.4, the system calculates the temperature from which there is risk of bubble formation [5]. For greater safety, considering the natural errors of the mathematical model used, a safety margin is used in order to generate an alarm when the hot spot temperature just approaches the bubble formation temperature.

In addition to monitoring the current temperatures, by checking whether they reach the safety margin for bubble formation, the monitoring system also checks whether, under the current load conditions and ambient temperature, the winding temperature will evolve to reach these risk values.

2.6 - Subsystem Conservation Tank – Membrane Burst Monitoring

As mentioned in item 2.3, oxidation is the third factor for deterioration of the cellulose used for winding insulation. To maintain the oxygen concentration in oil at safe levels, without occurring accelerated aging of the insulation, the transformers are equipped with a sealing system, which maintains the oil insulated from the atmosphere, while enabling its expansion and contraction due to temperature changes. This is achieved via rubber bag inside the expansion tank, which prevents the contact of oil, on its outer face, with air in its outer side.

Therefore, to make this system effective, it is necessary to assure the bag integrity, as its burst will enable contact of oil with air, contaminating it with oxygen and moisture. To do that, a sensor was installed inside the bag of regulator R1, as shown in figure 8, which detects the presence of liquid, thus indicating its burst. In this case, the monitoring system generates an alarm to the maintenance engineering.


Fig. 8 – Installation of membrane burst sensor of the expansion tank. (a) Sensor on the membrane, inside the tank; (b) Connection box on the tank top; (c) Sensor cable connection

2.7 - Subsystem Insulating Oil - Moisture Monitoring

In addition to acting as an aging accelerator and posing risk of bubble formation at high temperatures, the presence of excess water in the insulation poses the additional risk of free water formation in oil in case of low temperatures.

This risk is related to the fact that the reduction of the oil temperature also decreases its capacity of absorbing water (water solubility in oil). Thus, as the temperature decreases, the percentage of water saturation in oil increases, even that the quantity of water present in oil does not change. If the temperature decreases in such a way that saturation reaches or exceeds 50%, the water cannot more remain dissolved in oil, thus occurring its separation, that is, formation of free water.

For the current water saturation informed by the on-line moisture sensor (figure 7), the monitoring system checks continuously at which temperature the free water formation would occur, generating an alarm when the measured oil temperature approaches the calculated value. To do that, a safety margin is established in a similar way as for the bubble formation monitoring.

2.8 - Subsystem Cooling - Efficiency Monitoring

The proper cooling of a transformer is essential for its safe operation, with no accelerated lifetime loss of the insulation, in the presence of high loads. Therefore, it is important that the cooling operates in a proper way, removing the generated heat efficiently.

For this reason, the monitoring system is provided with the Engineering Module to check the Efficiency of the natural and mechanical cooling systems. This module operates by comparing the temperature measured on the oil top, achieved by the Temperature Monitor, with its expected value, calculated in function of the ambient temperature, load current and cooling stage in operation (figure 6). When the temperature measured is above the expected one, a low system efficiency warning is issued.

2.9 - Subsystem Cooling - Fan Maintenance

Fans play an essential role for operation of the regulator transformers, as their failure would limit the maximum load that the transformer can be submitted to. For this reason, the monitoring system has the Engineering Module for Mechanical Ventilation Maintenance, which provides useful alerts to assist the maintenance by measuring the fan operation times, such as service times of every ventilation group, mean daily times of operation and expected times to reach the number of hours for inspection or maintenance.

2.10 - Load Simulation Module

By considering the importance of regulator transformers for the Paraguayan electric system, and the possibility of operating them in overload condition in case of contingencies, the monitoring system has an Engineering Module for load simulations, where the user can can check the consequences of hypothetical situations in terms of temperatures reached, lifetime loss and bubble formation risk. In all the simulations, the user can change the cooling command mode (automatic or manual) and the temperatures for cooling activation and hysteresis.


By using the existing infrastructure for the regulator transformers, and thanks to the modularity and scalability characteristics of the decentralized architecture of the monitoring system installed, it has been expanded with integration of the on-line capacitance and tangent delta monitoring for the insulation of three 500kV current transformers at the same substation, shown in figure 9a.

By using the existing infrastructure for the regulator transformers, and thanks to the modularity and scalability characteristics of the decentralized architecture of the monitoring system installed, it has been expanded with integration of the on-line capacitance and tangent delta monitoring for the insulation of three 500kV current transformers at the same substation, shown in figure 9a.


Fig. 9 – CT Monitoring. (a) 500kV pedestal CT; (b) Measurement module in the junction box for monitoring of capacitance and tangent delta of the insulation


The monitoring system installed at the Margem Direita SS is provided with modular and decentralized architecture, which are features that enabled expanding the pilot project to Regulator R4 and pedestal 500kV CTs, as shown in the diagram of figure 10.


Fig. 10 – Architecture of the monitoring system installed at the Margem Direita SS

The monitoring server located in the maintenance room is connected to the Intranet of Itaipu, in order that the access to data, information, diagnostics and prognostics can be performed from any computer in the company network.

To prevent the need for ongoing system tracking, which would consume a great mount of time of the maintenance team, the monitoring system was equipped with an automatic e-mail submission mechanism in case of any abnormality,as the transformers will remain under normal operation conditions most of the time.

Once commissioned, in November 2006, the monitoring system started to save the measurements in a database, as exemplified in figure 11, where we see in a graphical chart format, the evolution of ambient, oil and winding temperatures and lifetime loss for a period of 20 days, in the first months of system operation.


Fig. 11 – Examples of measurements saved in the history database


The evaluation of the monitoring system operation, made by Itaipu since its installation in November 2006 to the end of 2008, enabled to verify the applicability of this system to transformers at the Margem Direita SS, with positive results in terms of the benefits achieved, which include the reduction of catastrophic failure risks and reduction of downtimes for preventive maintenance, for example, for bushing tests, with consequent increase of the equipment availability to the electric system, among others.

As usual in this type of application, adaptations were required during installation and throughout the project, as in the case of monitoring of bushings equipped with BPD, for which a special monitoring technique has been developed. A proof of sensitivity and operation of this solution was achieved upon the occurrence of poor contact in the internal wiring of a BPD, which was correctly indicated by the monitoring system.


[1]   Revista ELECTRA, Ref. no. 88, “An International Survey on Failures in Large Power Transformers in Service”. Paris: CIGRE, 1983.

[2]   Melo, Marcos A. C., Alves, Marcos, “Experiência com Monitoração On-Line de Capacitância e Tangente Delta de Buchas Condensivas”, XIX SNPTEE – Seminário Nacional de Produção e Transmissão de Energia Elétrica. Rio de Janeiro, Brasil, 2007.

[3]   McNutt, W. J., “Insulation Thermal Life Considerations for Transformer Loading Guides”, IEEE Transaction on Power Delivery, vol. 7, No. 1, pp. 392-401, January 1992.

[4] Fabre, J., Pichon, A., “Deteriorating Processes and Products of Paper in Oil. Application to Transformers”, CIGRE Paper 137, 1960.

[5]   Oommen, T. V., Petrie, E. M., Lindgren, S. R., “Bubble Generation in Transformer Windings Under Overload Conditions”, Doble Client Conference, Boston, 1995.

[6]   Albuquerque, Roberto, Alves, Marcos, “Monitoração On-Line de um Banco de Autotransformadores 345-138/13,8kV 150MVA com Comutação Sob Carga”, XIX SNPTEE – Seminário Nacional de Produção e Transmissão de Energia Elétrica. Rio de Janeiro, Brasil, 2007.

[7]       Alves, Marcos, Araújo, Daniel C. P., Martins, Alvaro J. A. L., Costa, Marcelo A., “Monitoração e Diagnóstico On-Line de Transformador de Potência com Óleo Vegetal”, V Workspot – Workshop on Power Transformers, Belém, Brasil, 2008.

Back To Top
×Close search