Benefits of On-Line Condition Monitoring Systems for High-Voltage Circuit Breakers

By M. Landry; G. Diagneault; J-F Theoret; A. Mercier; R. Bennet; R. Schwabe and S. Zelingher

On-line monitoring systems are becoming more attractive as they help utilities reduce operation and maintenance costs, increase breaker reliability, and acts as an effective tool for planning maintenance. It can be anticipated that such monitoring systems will allow utilities to shift from preventive maintenance carried out at fixed time intervals to predictive maintenance based on the equipment's condition.

In the fall of 1990, a joint project was initiated by three major electrical utilities: New York Power Authority (NYPA), Consolidated Company of New York (Con Edison), Hydro-Quebec, and the Empire State Electric Energy Research Corporation (ESEERCO). In the spring of 1993, the first monitoring-system prototype was successfully installed on an SFA SF6 double-pressure circuit breaker [1]. This new monitoring system was named MONITEQ (MONIT stands for monitor or monitoring and EQ for equipment, since the monitoring system can be adapted to other strategic high-voltage substation equipment). Eleven systems are now in service in North America: six are on 345-kV and 800 kV SFA SF6 double pressure circuit breakers and five on PK-type air-blast circuit breakers with voltage ranging from 245 to 800kv.

This paper will report about the benefits of the MONITEQ system thus far. The first selection will review the monitoring system architecture while the second section will deal with the selection of breaker parameters and the corresponding sensors.

The third section will present three case studies of benefits obtained by the installation of MONITEQ on PK circuit breakers. These cases address the following:

1. a defective mechanical link between the horizontal operating rods on an 800kV air blast PKV type reactor circuit breaker at NYPA's Massena substation,

2. an incomplete main breaker contact travel due to defective accordion contacts on a 400kV air blast PK type backup circuit breaker at IREQ's high power laboratory, and

3. an event where catastrophic failure was not prevented on a 245 kV air-blast, PK type capacitor bank circuit breaker at Hydro-Quebec's Levis substation, but the stored data in the monitoring system was used to quickly identify the faulty part in the pneumatic operating mechanism.

Finally, the last section will discuss future projects related to adapting MONITEQ on other circuit-breaker types and improving data analysis.

Review of the monitoring system architecture

The system architecture (fig.1) comprises:

non-intrusive sensors, incorporated into the monitored equipment to evaluate its major operating parameters;

a local controller (data acquisition unit), the heart of the monitoring system, mounted in an EMI shielded cabinet at the e base of the apparatus. It measures and records the mechanical and electrical signals from sensors monitoring the parameters as defined in [1,2,3];

a central controller (PC-based computer) in the substation building, which handles long-term storage of trend and event data, diagnosis and alarm processing, and notifies control and maintenance personnel of the occurrence and magnitude of an alarm. Up to 16 local controllers can be connected through a bi-directional fiber optic link to a single central controller;

user PC's

For maintenance or diagnostics, information coming from the monitoring system can be accessed at the central controller directly in the control room or remotely by modem or at a PC connection directly at the breaker local controller.

Monitored parameters and selected sensors on PK breakers

The parameters to be monitored must meet established criteria . Their definitions must be clear to people in the field, and be measured reliably. The corresponding sensors must be non-intrusive to the equipment, and they should be strategic or critical to the behavior of the monitored equipment and have the ability to detect incipient problems before they occur.

The following electrical and mechanical parameters were selected for the PK breaker:

dynamic air-pressure drop at the base of each column, which provides the signature of the breaker operating mechanism and synchronism between breaker heads;

air-pressure in each single-pole tank;

moisture content of the compressed air in each single-pole tank;

current circulating through the closing and tripping coils and electrical continuity of coils;

interrupted current and sum of Joule integrals (·I2t) and evidence of arc recognition or restrike;

control cabinet temperature (one sensor for each single-pole and three-pole cabinet);

AC and DC auxiliary power supply voltages

Benefit case studies

At the 1996 CIGRE session [3], it was reported that MONITEQ had detected two abnormal conditions:

first, the system saved a HV PK circuit breaker from a major failure at IREQ's high-power laboratory by detecting a mechanical defect;

and second, it helped electricians trace a thermal insulation problem in a high-pressure tank of an SFA breaker at NYPA's Massena substation.

Three new case studies are presented below.

Defective mechanical link on an 800-kV air-blast reactor circuit breaker

On June 17, 1996, Moniteq once again proved its effectiveness by identifying, during an opening operation, a severe mechanical defect on an 800 kV PKV type reactor at NYPA's Massena substation.

The schematic in Figure 3.1a provides an understanding of the reduction in pressure of the compressed air when the breaker operates. When the beaker is closed, the entire system is in equilibrium and P1 = P2 = P3, but when blast valves VS1 and VS2 open, the volume V1 is soon exposed to atmospheric pressure, with the result that the corresponding pressure P1 suddenly decreases.

This pressure reduction is felt immediately at the bottom of the column (P3). The pressure sensors are judiciously positioned to measure the drop in pressure (P3), which is correlated to a signature of the mechanical operation of the blast valves of each breaking head. FOr a breaker's electrical performance to be adequate, the valves must function flawlessly and be synchronized.

Figure 3.1b shows the air pressure drop (P3) in the four breaker heads of phase C of the 800-kV PKV reactor at Massena substation. It depicts the head no. 4 has almost no pressure drop compared to the other three normal breaker heads, which are very well synchronized. Inspection by the maintenance team reveled that a missing locking pin at pole No. 3 (Fig. 3.2) allowed the operating arm between breaker heads 3 and 4 to drop out. As a result, the main blast valves (VS1 and VS2) of breaker head No. 4 did not operate. The small pressure reduction that can be observed in Figure 3.1b is only the result of an air pressure decrease in the tank following the operation of the three normal breaker heads.

Clearly, if the breaker had to interrupt at full rated short-circuit current and voltage, it would not have been capable of doing so with only three of its four heads operational and it would have failed. It is estimated that MONITEQ saved NYPA $400,000 in avoided repair costs as well as the costs associated with the replacement of adjacent apparatus that may have been damaged if the circuit breaker had actually exploded.

Moreover, it should be pointed out that this type of mechanical defect would have been detected only during timing tests, which are normally performed every 6 years. This successful detection once again demonstrated MONITEQ's capability to detect mechanical problems on PK-type air-blast circuit breakers and confirmed that the pressure sensor at the base of each supporting column is a vital sensor for this type of breaker.

Incomplete main breaker contact travel on a 400-kV air-blast circuit breaker

In the fall of 1993, MONITEQ was installed on a 400-kV air-blast PK-type breaker at IREQ's high -power laboratory. This installation was ideal since the breaker was prone to frequent failures, due to its acting as a back-up breaker during short-circuit breaking tests on power circuit breakers, exposed to much more difficult conditions than those found in usual high-voltage substations. It would also provide a remarkable test bench for the whole monitoring system.

On August 21, 1996, MONITEQ triggered an alarm indicating that a critical operating parameter was out of bounds. A pressure signal originating from the breaker head no.2 of phase B indicated that the operation was abnormal. As illustrated in Figure 3.3, the pressure drop is less than the normal level. The breaker was then immediately de-energized. The inspection revealed that the tubular contact assembly was jammed in a semi-open position.

Upon investigation, it was noted that one of the extremities of the accordion contacts was pointing inwards and was in direct contact with the fixed contact tube, as shown in Figure 3.4. During the last maintenance, the moving contact assembly had been refitted with these inappropriate accordion contacts. Friction of the contact edge with the fixed tube created debris, which in turn increased the friction of the moving parts to the point that they were impeded by their movement.

Under such conditions, the circuit breaker would not be able to interrupt the rated short-circuit current. It is estimated that the monitoring system saved IREQ up to $225,000 adjacent apparatus that would have been damaged if the circuit breaker had actually failed. Again, MONITEQ proved its effectiveness by identifying beforehand a critical problem which could not have been detected otherwise.

Catastrophic failure on a 245-kV capacitor bank circuit breaker

Up to now, MONITEQ has proven to be effective in identifying defects prior to failure of the monitored equipment. On August 6, 1996, it was used to identify the root cause of a catastrophic breaker failure by analyzing recorded data. It is demonstrated that the breaker anomaly, presented below, would not have been detected with any complement of sensors.

In the fall of 1995, MONITEQ was installed on a two-head 245 kV PK-type capacitor-bank circuit breaker. Until then, the monitoring was uneventful and the breaker showed no signs of premature degradation. However, during a breaker opening under normal capacitive lad current, phase C of the air blast breaker was destroyed. During the event, MONITEQ recorded the different functional parameters, allowing the research team to retrieve the events that had occurred, in the same manner as they would with a black box after an airplane crash.

To identify the cause of the breaker failure, various critical signals were analyzed:

each nominal phase was normally interrupted

one single tripping opening order was given and currents circulating in the tripping and closing coils were also normal

the position of the auxiliary contacts confirmed the signals applied to the tripping and closing coils;

a major anomaly was detected on one of the pressure signals.

Pressure-drop signals in phases A and B were normal for the opening of the breaker whereas the phase C signal was very different. Analysis of the recorded characteristic "pressure drop" at the base of the two columns of phase C showed repeated pulsing intervals in the pressure signal of phase C Fig 3.6a). Both heads of phase C reacted in the same manner as shown in Fig.3.6b for a time span of 300ms.

Analysis of these signals leads us to conclude that the problem was located either in the pneumatic drive mechanism common to the two breaker heads or in the horizontal driving rods. Each pressure drop indicated that the breaker had somehow received multiple or opening orders, and the main blast valves operated each time. However, no closing or opening electrical order had been transmitted to the pneumatic drive mechanism after the initial one.

This suggests that the problem was located in the pneumatic drive mechanism. After close inspection, it was noted that the Rislan seals on the position (Fig. 3.6c) in the opening electro-valve were scratched and the piston assembly was not moving as freely as it should have been.

The internal temperature and humidity data of the single-pole cabinet, recorded by MONITEQ, showed that the breaker operated at 44.7 degrees Celsius which is much higher than the usual average of 20 degrees Celsius. Test on the Rislan O-ring seals revealed that over time they became prone to heat induced dilatation. This was pointed out as the cause of the breaker failure.: the piston remained stuck in a semi-open position, causing the pilot valve to oscillate and thus sending successive opening orders to the breaker's main blast valves.

The chronology of events was easily reconstructed to be the following:

1. An opening order was transmitted to the breaker to isolate the capacitor bank from the electrical network.

2. An electrical order was transmitted to the opening electro-valve through its tripping coil

3. Due to heat dilatation, the opening electro-valve piston jammed in a semi-open position. The breaker opened normally, but the jammed electro-valve piston forced a pumping motion of the pilot valve.

Cyclical operation of the pilot valve caused the repetitive operation of blasting valves VS1 and VS2 (Fig. 3.1), which in turn gradually decreased the air pressure in the tank.

5. Loss of air pressure reduced the breaker's inter-contact dielectric strength, resulting in an inter-contact restrike on both heads and subsequent arcing and destruction of the insulating porcelains of the interrupting chambers.

MONITEQ helped identify the where and why of the failure in situations where it was not possible to detect an anomaly prior to breaker failure. Without the recorded air-pressure drop signals, this piston problem would not have been detected. It also helped determine why the part was defective, thanks to the event log and the control-cabinet temperature data. Although the catastrophic failure was not prevented, recommendation were issued for future maintenance plans to prevent any reoccurrence of such an event. It fact, corrective maintenance work allowed defective pistons to be detected on two other 800 kV air-blast circuit breakers at Hydro-Quebec high-voltage substations.\

Summary and conclusion

The monitoring system discussed in this paper has shown tangible benefits in both operation and maintenance (O&M) and provided significant cost savings. With the first prototypes installed in high-voltage substations in Canada and the U.S. in 1993, the eleven monitoring systems installed so far have brought estimated savings in excess of $1-million by detecting four functional breaker anomalies and providing crucial data in the case of the catastrophic breaker failure. These case studies highlight two major functions of breaker monitoring systems, namely:

an O&M tool to identify equipment malfunctions, prevent catastrophic failures and facilitate planning of maintenance tasks;

an Information System (black-box data recorder) by providing crucial data to quickly identify the root causes of the equipment failure, speeding up maintenance actions and minimizing outage time.

Hydro-Quebec and NYPA are planning to expand the ongoing program and implement MONITEQ on strategic power circuit breakers at their high-voltage substations. Furthermore, R&D initiatives are under way to adapt the monitoring system to other breaker technologies at different voltage levels: other air-blast circuit breaker designs, live-tank and dead-tank single pressure SF6 circuit breakers, generator circuit breakers, etc. To this end, new sensor technology , always non-intrusive and commercially available, is being evaluated and implemented.

To improve the predictive capability of the monitoring system, use of export systems, artificial intelligence and fuzzy logic systems is being considered. In addition, these software tools will provide substation maintenance personal with accurate breaker failure diagnoses and facilitate the planning of maintenance tasks.

Because of its capabilities, MONITEQ can be used to monitor any type of equipment. For instance, it has already been adapted by Hydro-Quebec to synchronously monitor fault indicators in an underground distribution system for which conventional data acquisition systems are unsuitable because of the large distances between manholes. ET

M. Landry, G. Daigneault, J-F Theoret, and A. Mercier are with Hydro-Quebec. R. Bennet, R. Scwabe and S. Zelingher are with the New York Power Authority.