On-Line Condition Monitoring Systems

For High Voltage Circuit Breakers

A Collaborative Research Project 1997 - 2001
Project Participants:
ABB, Alstom, BC Hydro, Doble, ESKOM, Manitoba Hydro, Siemens, SPI PowerNet, TransAlta
Utilities, Manitoba HVDC Research Centre

Abstract: A three year field implementation of on-line monitoring technology was applied to a 240 kV
SF6 circuit breaker. The main focus of the project was to evaluate the state of the art in on-line monitoring
technology by first hand experience, thus providing an understanding of the benefits and the practical
issues of implementing this technology. The project implemented a comprehensive range of monitoring for
electrical, energy, mechanism and SF6 gas systems. Transducers of various types were installed. The
breaker was operated over 1000 times during the test period with ambient temperature conditions ranging
from –35C to +30C. The on-line monitoring was supplemented with off-line testing performed 4 times
during weeklong outages. The application of on-line monitoring has produced many valuable results and
enhanced the knowledge base for the apparatus under test. Clearly on-line monitoring of HV circuit
breakers has potential, however the installation of on-line monitoring systems must be considered carefully.
Monitoring systems can provide improvement in the understanding of the operation of a breaker and
provide input into RCM programs. However, the monitoring systems themselves require maintenance and
attention. The application area of data interpretation and/or conversion of this data into information
requires continued research and development.

Introduction: The objective of this project was to

review and evaluate On-line Condition Monitoring
Technology for High Voltage Circuit Breakers. An
ABB ELF 240 kV SF6 breaker located at Dorsey
Converter station was selected for this project by
convenience but the type of breaker was really
independent of the research being performed. The
ELF breaker is a three independent pole design
allowing for three separate monitoring systems to
be installed and evaluated. The project installed a
different type of monitoring system on each phase
and incorporated a large variety of transducers. The
240 kV breaker was operated over 700 times at
rated voltage and an additional 300 times during
Fig 1: ABB ELF 240 kV Breaker
maintenance over a three year monitoring period.

Monitoring Systems Installed: Each of monitoring systems installed measured the same basic parameters,
including contact travel, “a” and “b” auxiliary contacts, phase currents, coil currents, heater and pump
current, SF6/CF4 pressure and temperature. Vibration transducers were installed on each mechanism.
Different styles, makes and costs of transducer were used whenever possible. There were primarily two
separate classes of monitoring systems. One class as installed in one phase can be described as a protective
relay or stand-alone system. This type of system has all of the on-line intelligence located at the device
itself (Fig. 2). The monitoring algorithms processes the data and generate alarms based on preset threshold
values not unlike a modern digital protection relay. There is limited storage of previous event data and no
long term trending. Remote communications were implemented using a telephone modem. An automated
dial-up database system designed in Microsoft Access was able to provide trending information for this
system configuration.

Breaker On-line Monitoring Page 1

 Fig. 2: Stand Alone System

The second class of system as installed in the other two phases is a distributed intelligent system (Fig. 3).
The equipment located at the breaker is primarily data acquisition. The data is transmitted to central data
server over a network. The on-line analysis is performed and the alarms from the system are generated at
the central computer server. Since the data is stored, long term trending is possible. Remote communication
can be accomplished through the server. Two separate systems of this configuration were installed on two
phases of the test breaker.

Fig. 3: Network or Server Based System

Installation:
The monitoring systems were retrofitted to an existing breaker.
Approximately one week of field effort was required to install and
commission each system. The installation of the various
transducers required space in the existing control panels. The phase
current signals are not routed via the breaker cabinets and a new
cable route from the zone box, which contains the current
transformers cabling, to the breaker control cabinets was required.
A decision was made to implement as many different transducers
types as possible. Table 1 lists the parameters and transducers
implemented in each phase.
Fig 4: Various Phase Current CTs

Breaker On-line Monitoring Page 2

Experience and Insight in On-Line Monitoring:
• Data Acquisition Equipment: Data acquisition and the transducer components of the systems worked
very well. A wide variety of transducers were utilized to monitor an equally wide variety of breakers
parameters. Phase B system experience showed that, while the integration of standard components into
a working system is not a trivial exercise, it is certainly possible to collect data using many standard
devices and communication protocols. Transducer failures did occur during the project, including two
separate pressure transducers and one travel transducer, but defects in this equipment were relatively
minor and straightforward to troubleshoot and repair.

• Computer and Computer Peripheral Equipment: All of the systems installed have had software,
operating system, and computer hardware or communication hardware problems. These problems
tended to be less than easily resolved and several attempts or steps are required before problems are
isolated and repaired. Examples include the following:
• Windows NT server and other software that intermittently hangs up.
• CPUs that lose communication and will not restart until a master reset is performed.
• Defective modems, phone switchers etc.
• Defective tape drives.
• Failed CPU cooling fans.
• Excessively long communication and download times.
• Difficulty in upgrading operating systems and/or application software.

The following example illustrates one computer related problem. A monitoring system would fail to
respond to remote attempts to interrogate and download data over phone modems. Personnel was
dispatched to site and determine the server application program was not running. The server
application would be restarted and the system would again operate properly for several days to a week.
The process would repeat. After about 3 or 4 weeks it was finally determined that the cooling fan on
the server processor had failed, thus overheating the processor, which would automatically power
down. Several hours later the processor would cool down and server computer would restart. However
the automatic start routines did not include all of the applications necessary for the on-line monitoring
server to function. A new CPU cooling fan corrected the problem and generated a review of the
automatic startup procedures for the server computer. Although this problem was not particularly
serious, the troubleshooting process was longer and more frustrating then one would have hoped.

• System Set Up or Initialization: All of the systems required some degree of set up or initialization by
the system providers. The project demonstrated the importance of supporting documentation on the
customization of parameters, set points, rule sets and so on. It is possible to have a system
“operational” but to have incorrect, invalid or non-existent data programmed into the system, thereby
making the results questionable and erroneous. As the monitoring systems continue to mature this
issue should be resolved and therefore become less important. However the implications of this issue
are serious. Is a monitoring system that provides no alarms really working? Is the apparatus
monitored healthy or does the absence of alarms simply mean that the monitoring system is not
functioning?

• Skill sets and competencies required to operate, maintain and upgrade monitoring systems need to be
assessed by each utility. While this comment relates to the Manitoba Hydro situation, it can also be
applied very generally to the utility industry. The technicians responsible for high voltage apparatus
are highly skilled but generally do not have the computer trouble shooting or installations skills that
are required to keep these systems operational. As a result large scale implementation of on-line
monitoring would require either re-training and/or new personnel with the necessary skills.
Instructions, even to experienced staff, must be documented, complete and sufficiently detailed.

• Detailed monitoring and investigation for a particular breaker can reveal important information with
respect to that breaker. This knowledge base can be applied to determine the scope, and benefits of a
monitoring program for the breaker. One of the benefits of this project is the decision by Manitoba

Breaker On-line Monitoring Page 3

 Hydro to modify its maintenance practice for ABB ELF breakers and to determine the extent of on-
line monitoring systems for ELF breakers.

Table 1: Transducers installed in each phase

Off-Line Testing Program:

Off-line testing was performed during the monitoring period, including timing, vibration and dynamic
resistance measurements. Operating conditions of the breaker were adjusted in order to observe the impact
of these parameters on breaker timing and to determine the response of the on-line monitoring systems to
any changes that occurred in the breaker. Tests were performed at different ambient temperatures, control
DC voltage, SF6/CF4 gas pressure and hydraulic energy levels. Variations of up to 10% were observed in
the timing performance of the breaker. The on-line monitoring systems did provide alarms for the some of
the conditions. For example a “Long Closing Coil time” alarm would be issued. The alarm set point was 50
and the measured values ranged from 58 to 63. The difficulty from the breaker maintenance point of view
is in deciding “what action do you take when you receive an alarm”. Table 2 indicates the variation due to
specified parameter changes. Please note the parameter changes are all within the “normal” operating range
for the breaker.

Energy Storage
Temperature N-2 Close All
Operation 20°C to –20°C N-1 Open SF6 Pressure Energy SF6 Parameters
Close 2% 5% 1% 5-6% 8-10%
Open 2-3% 2-3% 1-1.5% 4-5% 5-6%

Table 2: Variation in close and open times

Breaker On-line Monitoring Page 4

SF6 Gas Monitoring Detailed Results:
The intensive monitoring and investigation did achieve some interesting results. SF6 gas density is
generally calculated by using a temperature compensated method. This method requires both a
measurement of gas pressure and gas temperature to be collected. A total of five SF6/CF4 gas systems were
installed and each system generated erroneous results when the ambient temperature was below –20°C. For
this particular breaker there is gasket heater, which is turned on at –20°C. The location of this heater is
physically close to the RTD for the SF6 gas temperature. As shown in Fig. 5, the SF6 gas temperature
recorded for ambient temperature below –20°C is notably disturbed, while the pressure remains linear with
temperature. Of course the SF6 gas density or compensated pressure readings are corrupted below an
ambient temperature of –20°C. This systemic error resulted in false alarms from all monitoring systems.
One system had a difficult time in dealing with oscillatory alarms of this nature.

80 40C

70 30

60 20

50 10

40 0

30 -10

20 -20

10 -30

0 -40
-40 -30 -20 -10 0 10 20 30 40 T Amb
T Diff (SF6-Amb) SF6 P SF6 P Comp
SF6 T Amb T Linear (SF6 P)

Fig. 5: SF6 Gas Temperature and Pressure vs. Ambient Temperature

Fig. 6 illustrates the trending of SF6 gas pressure over a three-year period. The following observations can
be made. Notice the oscillations in SF6 density when the SF6 temperature is near –20°C. This is due to the
problem described above. The absence of data is a result of failure of a transducer. Notice the slight
negative slope of the density line. This indicates a very small SF6 leak which was undetectable by reading
the gas pressure gauge. Further investigation located the leak at the connection to the gauge, which was
installed as part of the on-line monitoring. In this example both the strengths and weaknesses of on-line
monitoring can be illustrated. The on-line system did identify a problem, however, the system user had to
interpret the data and the problem itself was a result of the on-line monitoring being installed.

SF6 Daily Average

700 120
600 100
500 80
400 60
300 40
200 20
100 0
0 -20

Pressure kPaG Comp Pressure kPaG Temperature Density %

Fig. 6: SM6 Consolidated Density Results (Sept 97- Sept 2000)

Breaker On-line Monitoring Page 5

Vibration Analysis:
One of the monitoring systems incorporated an
on-line vibration measurement using an
accelerometer as the sensor and a signal-sampling
rate of 40 kHz. The purpose in doing the
measurement was two-fold: firstly to determine if
it was possible to get consistent signals over a
long period and secondly, assuming a consistent
signal, to derive a method whereby the signal was
machine-readable.

The results of the measurement were very positive

in that the vibration signals were remarkably
reproducible. In fact the signals exhibited a
repeatable seasonal variation showing very
evident differences between summer and winter Figure 7: Vibration Correlation Matrix
and about equal characteristics in the spring and autumn. The signals were analyzed using a standard
signal processing method and comparison between the processed signals was done using a cross correlation
method. Fig 7 shows an example of a 2-dimensional bicubic interpolation of a correlation matrix over a
period of one year. The repeatable difference between summer and winter is evident, the peaks being
winter and the valleys summer with spring and autumn in between.

Conclusions:
The project provided important conclusions and insights into the application of condition monitoring on
circuit breakers.

Long-term on-line monitoring systems can provide valuable information regarding the performance of a
HV breaker. Data analysis performed by qualified personnel is still required in order to determine the
action plan resulting from the monitoring results. This can be minimized if an expert system is
incorporated.

Monitoring systems are subject to failure. While data acquisition and transducers related faults are
relatively easy to detect and repair, computer systems, data servers, and software problems can be
significantly more time intensive and require computer skill sets to solve effectively. Utilities exploring this
technology must be aware of the skill sets required to implement and maintain these on-line monitoring
systems. This situation will improve, as utilities become more IT oriented.

Monitoring systems will require some degree of maintenance. The maintenance program should include a
periodic review that the data acquisition systems are collecting sensible data. This maintenance program
will also include any software and firmware upgrades, updates or revisions. A maintenance system for
tracking software, firmware, and operating system revisions must be in place. Future systems will be
expected to be self-diagnostic to at least some degree.

Monitoring system require a careful observation period to ensure the system is set up and responding
appropriately. Two main pitfalls of on-line monitoring that must be avoided are the absence of alarms
resulting from the monitoring systems not setup properly and/or disabled and generation of a high number
of nuisance or false alarms.

Each piece of apparatus to be monitored requires study or careful review to customize the monitoring for
each application. At this point there is “not a one size fits all” monitoring system. The findings from this
intensive general monitoring program were used as inputs to an RCM program for this breaker. As a result
changes to the maintenance program for this breaker have been implemented and these findings have
clarified the monitoring requirements for this breaker. It is difficult to extrapolate the findings for this
breaker to a breaker incorporating other designs.

Breaker On-line Monitoring Page 6

In our project the breaker did not experience any failures, so there can be no discussion on a monitoring
systems ability to detect and determine any trend towards failure. One of the major unanswered questions
is: “Can a monitoring system detect and predict failure, maintenance requirements or end of life?” This
prediction must take into account the normal variation of measured and calculated parameters and must not
generate trivial or false indications.

Future:
The goal of on-line condition monitoring systems has always been to provide information to the user of the
systems. This information will form the basis of a maintenance, repair or refurbishment program for the
apparatus. On-line monitoring and off-line monitoring techniques are keys for allowing maintenance
personnel to make intelligent decisions.

A number of areas require continued development and improvement. Our experience demonstrated that the
reliability of the HV breaker tested is a least an order of magnitude higher then the reliability of the
monitoring equipment installed. The monitoring systems have to be at least as reliable as the breaker or
apparatus monitored. The reliability of the data acquisition portion is nearing this goal but continued efforts
to increase the reliability of the computer based server technology while maintaining reasonable costs is
required. The interface to the end users of the technology can be improved. On-line monitoring systems can
generate large amounts of data. Data handling, storage and conversion of this data into meaningful
information require further and continued development.

The present approach to equipment monitoring is unlikely to continue. Utilities cannot tolerate having to
deal with a multitude of equipment monitoring systems, as would be the case in very large stations.
Utilities are moving, however slowly, towards the concept of ITEC (Information Technology Electronics
Communications). With ITEC, electronic systems, super IEDs, Plant information or PI servers will capture
data and information. Communication systems will deliver the data and information to the stakeholders in
the required timeframe and with the required quality and security; and IT systems will process the data and
information, make decisions and to an ever-increasing degree execute the decisions. More information on
this subject can be found in Canadian Electric Power Technology Roadmap: Forecast, March 2000
available from Industry Canada.

How will ITEC effect equipment condition monitoring? The most probable answer to this question is that
equipment will have to be supplied compete with specified sensors having specified outputs and using a
specified communications protocol. Data capture, processing, analysis, etc will then be performed by the
utility IT system. B.C. Hydro, in fact, has initiated a major project to demonstrate the value of the concept
and to date has developed a six-layer conceptual data model for the entire transmission system that will run
on a GIS platform. A pilot application is in-progress for a large 500/230 kV transmission station and its
associated transmission lines.

Reference:

CIGRE Brochure No. 167 User Guide for the Application of Monitoring and Diagnostic Techniques for
Switching Equipment for Rated Voltages of 72.5 kV and Above.

B. Rushford, "Justifying On-Line Equipment Monitoring and Associated Data Acquisition and Processing".
CEA Electricity '98 Conference, April 1998, Toronto, Canada.

Breaker On-line Monitoring Page 7