CIRED

18th International Conference on Electricity Distribution

Turin, 6-9 June 2005

TRIP COIL SIGNATURE ANALYSIS AND INTERPRETATION FOR DISTRIBUTION CIRCUIT

BREAKER CONDITION ASSESSMENT AND DIAGNOSIS
Scott STRACHAN, Dr. Stephen McARTHUR,
Prof. James McDONALD
University of Strathclyde - UK
scott.strachan@strath.ac.uk

INTRODUCTION
A circuit breaker may be considered an active link, used to
make and break an electrical circuit under conditions of
varying severity. In the event of a circuit fault the circuit
breaker must respond within tens of milliseconds in order to
prevent damage to connected plant and safeguard personnel.
As such, a circuit breaker will remain idle for the bulk of its
lifetime, only being called into action periodically to
interrupt load or fault current. This makes the circuit breaker
one of the most safety critical, yet most unpredictable plant
items on the power network. In addition, circuit breaker maloperations account for a significant proportion of Customer
Minutes Lost (CMLs), Customer Interruptions (CIs), plant
replacement costs and safety issues. Conflicting regulatory
and economic pressures to improve the performance,
reliability and longevity of these, and other major plant items
while reducing maintenance costs have forced many utilities
to reassess their asset management and maintenance
strategies [1]. This has led to significant investment in novel
condition monitoring technologies, which in turn precipitate
the need for new methods of analysing and interpreting new
forms and increasing volumes of plant data.
This paper centres on a data-driven approach to the analysis
of captured trip coil current signatures to establish the
operational boundaries characterising circuit breaker
condition. Presenting the trip coil signature data in a visually
appreciable format to maintenance experts facilitates the
definition of these thresholds and subsequent condition
assessment rules. It is envisaged that a separate diagnostic
knowledge-base will further enhance the current prototype,
providing maintenance staff with an explanation and
approximate location of any deterioration associated with
breakers undergoing routine trip testing.
The paper describes the implementation of a prototype system
for the condition assessment of distribution circuit breakers,
which has recently become the subject of a pilot study within
SP Power Systems.
CIRCUIT BREAKER OPERATING MECHANISM
AND TRIP COIL SIGNATURE
Operating Mechanism
Circuit breakers are comprised of a control unit which
actuates the movement of the breakers operating mechanism,
and which in turn actuates the make or break of the main
CIRED2005
Session No 1

William LEGGAT, Angus CAMPBELL

SP Power Systems - UK

contacts. This paper focuses attention on 11kV spring

operated distribution circuit breakers where the trip energy is
stored in a latched spring mechanism [2].
Control Unit Trip Coil Current Signature
The control or trip unit of the circuit breaker transforms the
trip signal into the physical operation of the breaker
mechanism. The trip coil consists of a conductor wrapped
around a movable iron plunger forming an electromagnetic
actuator. The trip coil becomes energised following the
initiation of a trip command.
The trip coil plunger motion can be characterised by the
current flowing through the actuating trip coil. The correlation
between the breaker operation and condition, and the trip coil
current signature suggests the use of the trip coil as a readymade non-invasive condition monitoring sensor [3]. The
resulting trip coil current signature provides a chronological
record of the operating sequence and timings associated with
the various breaker components during a trip operation.
The current flowing through the breaker trip coil generates an
electric field affecting a force upon the plunger, moving it
towards the breaker latch mechanism. At the same time, the
motion of the iron plunger induces an e.m.f. in the coil. Figure
1 illustrates the operation of the control units trip coil
plunger mechanism and its effect on the trip coil current
signature [4].
When the trip command is initiated the coil is energised and
the current rises causing a magnetic field to apply a force on
the iron plunger (Figure 1(a)). When the force on the plunger
exceeds that of a retaining force and stiction the plunger
begins to move (Figure 1(b) and point A on trip signature).
The motion of the iron plunger induces an e.m.f. in the coil
effectively reducing the current flowing through it, as
indicated by the first current peak (Ipk1).
As the plunger continues to accelerate through the coil the
current flowing through the trip coil continues to fall (Figure
1 points B to C), until the plunger eventually strikes the latch
mechanism (Figure 1(c)), where a sudden reduction in the
velocity of the plunger occurs resulting in a corner in the
current signature (Figure 1 point C to D). The combined mass
of the plunger and the latch continue to move at a reduced
velocity causing a further reduction in the coil current (Figure

CIRED

18th International Conference on Electricity Distribution

1 points D to E) until it hits a buffer bringing it to rest (Figure

1(d) and point E). With the plunger at rest the current
increases to the maximum rating of the coil (Figure 1 point F).
Meanwhile, the latch unlocks the spring operating
mechanism, opening the main contacts (Figure 1(e)).
Typically, after a short delay the auxiliary contacts open,
disconnecting the trip coil from the substation d.c. battery
supply (Figure 1 point G). As the coil is de-energised the trip
coil current decays quickly to zero in accordance with the coil
inductance (Figure 1 point H), causing the plunger to return to
its initial position (Figure 1(f)).

Turin, 6-9 June 2005

condition assessment and diagnosis [7].

SP Power Systems has approximately 600 MV circuit
breakers currently in service within their distribution network,
presenting a substantial maintenance burden. The trip
signature recorder shown in Figure 2 was trialed as a
supporting technology for the implementation of a conditionbased approach to future circuit breaker maintenance.

Figure 2 Trip signature recorder connection diagram (taken from [4])

Prior to the use of the portable trip signature recorder it was

common practice to remove the breaker from service before
assessing its condition and readiness to trip. This approach
ignored the conditions affecting the timings associated with
the initial breaker trip required to remove it from service. As
part of ongoing maintenance and asset management
responsibilities, an extensive data gathering exercise was
undertaken to capture trip signatures from the population of
in-service MV circuit breakers.
Figure 1 Control unit and trip coil current signature (taken from [4])

CIRCUIT BREAKER TRIP TESTING

Deterioration in a circuit breakers condition may result from
its inertia [5]. The effects of stiction on a breaker having
experienced a long period of inactivity may be removed by
simply exercising the circuit breakers operating mechanism.
However, the underlying problem/s responsible for this slow
operation often remain disguised and subsequently neglected.
An innovative method of circuit breaker condition monitoring
exists in the measurement of its control circuit characteristic
(i.e. trip coil current signature), and provides a useful insight
into the operation of the circuit breaker and its condition [6].
In addition to its primary function of controlling the breaker
operation, the control circuit trip coil performs a secondary
role as a non-invasive condition monitoring sensor from
which the trip coil current signature from the all-important
first trip may be recorded and analysed for off-line breaker
CIRED2005
Session No 1

A lack of trip signature interpretation expertise resulted in the

use of a simple heuristic to determine whether a circuit
breaker required maintenance (i.e. IF main contact time
100ms, THEN breaker requires maintenance).
The main contact operating time is of obvious concern to
asset owners and maintenance personnel when assessing the
performance of a circuit breaker. However, while the 100ms
threshold may be considered a satisfactory disconnection time
required to adequately protect a faulted circuit, this is not
necessarily indicative of the circuit breaker health. In practice,
operating times vary with the breaker model and
manufacturer, where 100ms may be considered satisfactory
for one breaker model but excessive for another faster model.
Therefore applying this simple heuristic may prove
misleading. In addition, other aspects of the circuit breakers
operation characterised by its trip signature provide a more
detailed insight into the circuit breaker condition and
developing faults.

CIRED

18th International Conference on Electricity Distribution

DATA ANALYSIS OF BREAKER TRIP SIGNATURES

With the limited availability of trip signature interpretation
knowledge or reference trip signatures the research challenge
presented here involved the identification and diagnostic
explanation of the various breaker condition states made
apparent from the analysis of historical trip signature data.
The outcomes of this data analysis then formed the basis for
the elicitation and implementation of circuit breaker condition
assessment thresholds and rules within a standalone prototype
system, providing decision support during routine trip testing
and circuit breaker maintenance.
The main tasks of the data analysis and prototype
implementation involved:
Identification of specific trip signature shapes indicative
of specific breaker conditions using clustering
techniques.
Establishing the operational thresholds from the trip
signature characteristics that define the different breaker
conditions identified through data visualisation.
Elicitation of diagnostic explanations describing the
different breaker conditions identified, and observable
signs of deterioration.
Implementation of these thresholds and diagnostic
explanations in a working prototype offering decision
support to maintenance staff and asset managers in the
future prioritisation, repair and maintenance of
distribution circuit breakers.
Data Preparation
The feature vector used to characterise the trip signature and
discriminate between different aspects of breaker operation
and condition is illustrated in Figure 3. The temporal features
reflect the condition of the breaker operating mechanism,
while the current features provide an indication of the trip coil
condition itself and the integrity of the battery supply.

Turin, 6-9 June 2005

represented in six-dimensional (i.e. six features) Euclidean

space, however the Sammon map approximates the inherent
data structure transforming it into a more visually appreciable
two-dimensional representation [8].
Figure 4 also shows the outcome from a K-Means clustering
algorithm [9] mapped onto the initial Sammon map, where the
different K-Means clusters are represented by points of
different shapes. This mapping enables the K-Means
perception of trip coil signature similarity to be visualised
graphically and illustrates how closely this corresponds to the
clusters observed from the Sammon distribution.

Figure 4 Sammon map of trip signature feature vectors

While the clustering techniques are used to identify signatures

sharing similar characteristics, it is necessary to identify
which characteristics make these signatures similar.
A parallel co-ordinate chart provides another means of
mapping n-dimensional space onto two-dimensions, using n
equidistant and parallel axes. These charts are particularly
useful for visualising and analysing time series data [10].
Each axis corresponds to a separate feature of the trip
signature feature vector. Each trip signature feature vector is
then represented as a plotted line traversing the parallel axes.
The chart illustrated in Figure 5 depicts the operational
envelopes corresponding to the clusters identified by the
Sammon mapping and K-Means clustering illustrated in
Figure 4. This envelope provides a more quantitative
description of the cluster, from which the feature thresholds
characterising this cluster may be observed.

Figure 3 Trip coil signature feature extraction

Data Visualisation and Analysis

The Sammon map distribution of feature vectors illustrated in
Figure 4 represents a population of trip coil signatures for one
specific circuit breaker model and manufacturer, offering a
notional indication of the (dis)similarity between trip
signatures and consequently different forms of breaker
behaviour/condition. These feature vectors may be
CIRED2005
Session No 1

Figure 5 Parallel coordinate chart of trip signature feature vectors

The densely populated envelope illustrated in Figure 5

corresponds to the largest cluster of Figure 4, and to the peaks
of separate frequency distributions describing each feature.
Based on the population of data observed it is likely that this

CIRED

18th International Conference on Electricity Distribution

envelope (or cluster) represents that of normal circuit breaker

operation.
A box plot is another means of plotting groups of
observations [11], effectively providing a shorthand
representation of a plotted distribution and enabling
comparisons between several groups of observations. A box
plot is comprised of an upper and lower quartile, an interquartile range and a median. The lower quartile is the value
below which 25% of observations lie, and the upper quartile
is the value above which 25% of observations lie. The
population median bisects the inter-quartile range which
accounts for 50% of observations and which separates the
upper and lower quartiles. The box plot provides an
indication of the shape and skew of the frequency distribution,
and subsequently where across a range of values the most
frequently occurring observations lie.
An appreciation of the most frequent (typical) operating times
may be identified from frequency distributions and box plots
describing each feature. This extends the principle of the
parallel co-ordinate chart to present a more quantitative
description of the operational envelopes representing different
circuit breaker conditions and the temporal thresholds that
define them. These distributions may be overlaid with
subsequent trip signature feature vectors to visualise where
and to what extent these trip signatures deviate from normal
operation (Figure 6).
Figure 6 compares the frequency distribution describing the
timings associated with each stage of a particular breaker
models operation. The dotted line represents the feature
vector plot of a normal trip signature captured from a
particular circuit breaker, where each feature value falls
comfortably within the inter-quartile range of its
corresponding box plot (or region of typical breaker operating
times). The solid line indicates a subsequent trip signature
recorded from the same circuit breaker at a later date. The
Mcon feature shows a significant deviation from the previous
signature, showing an outlying Mcon time in the 4rth quartile.

Figure 6 Box plots and frequency distributions describing breaker operation

CIRED2005
Session No 1

Turin, 6-9 June 2005

Expert judgment is required to set the feature thresholds that

define normal and abnormal breaker behaviour. Traditional
rule-based systems often rely solely on the expertise of
domain experts to define the rule thresholds, sometimes with
empirical corroboration [5]. However, outcomes from the
analysis of existing data, via the process outlined in this
paper, support experts tasked with setting these thresholds,
and subsequently improves the reliability and classification
performance of the condition assessment and diagnostic rules
finally implemented.
Visualisation and analysis of the trip signature data in the
manner described allows the analyst (typically an experienced
maintenance engineer) to accurately identify the timing
thresholds that define the observed envelopes of normal and
abnormal circuit breaker operation. Assuming the majority of
breakers in service operate satisfactorily, the most frequently
observed feature times may be considered typical of normal
operation. Therefore, for each stage (aspect/feature) of the
circuit breakers operation the timing thresholds defining
normal operation are formed by the upper and lower bounds
of the region along the frequency distribution exhibiting a
sufficiently high frequency of observations, that it may be
considered typical of normal operation.
While the definition of a sufficiently high frequency of
observations and the ensuing feature thresholds remain at the
discretion of the expert/analyst, presenting the data in this
graphical format informs this process and provides some
qualitative justification of their judgment.

Figure 7 Timeline describing breaker operation

The developed prototype provides a traffic light condition

CIRED

18th International Conference on Electricity Distribution

assessment of the breakers condition as characterised by its

trip signature features, where:
RED suggests immediate maintenance is required.
AMBER suggests signs of deterioration, requiring
maintenance at the earliest opportunity.
GREEN suggests no maintenance action is required
at present (i.e. until the next planned maintenance).
Subsequent breaker trip signatures can then be represented
and assessed as a timeline of breaker operation and
condition as illustrated in Figure 7. The breakers operational
timeline indicates its condition at each stage of its operation,
as defined by the thresholds set by the analyst. In addition, the
position of the feature value within the region indicates the
extent of the problem. Figure 7 shows the Start to Latch
time of a captured trip signature firmly placed in the middle
of the Red region and the Buffer to Mcon time marginally
entering the Amber region, while all other features of the
circuit breaker operation are assessed as normal (i.e. Green).
PROTOTYPE IMPLEMENTATION
As described, the developed prototype facilitates the analyst
in the definition of operational thresholds that characterise
the condition of different aspects of a particular breaker
models operation. These thresholds in turn form the rule
premise used to determine the condition of circuit breakers
subjected to future trip testing. The condition assessment and
diagnostic output from these rules then support
recommendations and prioritisation of future maintenance.
The condition assessment rule-base takes the form of an
eXtensible Stylesheet Language Transformation (XSLT)
document, while the user-defined timing thresholds are stored
in an eXtensible Markup Language (XML) document [12].
The prototype architecture comprises two modules:
The Trip Signature Analysis and Thresholds Definition
Module is used by the expert/analyst to visualise and
analyse the trip coil signature data, supporting the analyst
in the assignment of suitable thresholds defining Red,
Amber and Green circuit breaker condition.
The Circuit Breaker Condition Assessment and
Diagnostic Field Module is used by maintenance field
engineers conducting routine trip tests, to select the
appropriate pre-defined thresholds for the circuit breaker
model currently undergoing trip testing. The circuit
breaker condition assessment output is then displayed to
the user as the timeline of Figure 7, and operated on a
separate XSLT diagnostic rule-base.
CONCLUSION
The condition assessment and diagnostic output by this
system is intended to support the strategic decisions required
of maintenance and asset management personnel when
targeting and prioritising circuit breaker maintenance.
Other methods of trip signature analysis have relied upon
CIRED2005
Session No 1

Turin, 6-9 June 2005

comparisons between monitored trip signatures with stored

reference fingerprints captured from dedicated bench tests
or during circuit breaker commissioning. However, where
obsolete or vintage circuit breaker models remain in service
and are no longer installed as new, or the availability of
experimental resources is limited, the acquisition of such
reference data may prove impracticable. In addition,
manufacturer data does not typically extend to the provision
of reference trip coil signatures. The analysis of historical trip
signature data is presented as an alternative means of defining
suitable operating thresholds capable of discriminating
between normal and abnormal breaker condition.
This paper presents a data driven approach to the condition
assessment of distribution circuit breakers, operating on trip
signature data captured from a population of in-service circuit
breakers during routine trip testing. This allows analysts and
maintenance experts to identify and establish correlations
between signature shapes and specific breaker conditions.
Furthermore, using appropriate data visualization techniques
the operational features and thresholds characterising these
breaker conditions may be defined and implemented in a rulebased system for future circuit breaker condition assessment.
REFERENCES
[1] OFGEM, 2000, Information and Incentives Project
Review of PES Measurement Systems, pp26,
www.ofgem.gov.uk/temp/ofgem/cache/cmsattach/
214_21jul00.pdf.
[2] R. T. Lythall, 1972, JSP Switchgear Book 7th Edition,
Newnes-Butterworths, London UK.
[3] W.R. Speed, 2000, Circuit Breaker Operator Signature
Analysis, Proceedings of the International Conference
of Doble Clients, Sec 4-4.
[4] R. Watson, October 2000, Diagnostic technique
facilitates timing and condition monitoring of circuit
breakers, Proceedings of Engineering Asset
Management Conference.
[5] M. Kezunovic, 2003, An Expert System for Automated
Analysis of Circuit Breaker Operations, Proceedings of
ISAP, Greece.
[6] IEEE Std C37.10-1995, Sept 1996, IEEE Guide for
Diagnostics and Failure Investigation of Power Circuit
Breakers, IEEE Power Engineering Society.
[7] S.Beattie, 1996, Circuit Breaker Condition Assessment
by Vibration and Trip Coil Analysis, IEEE Colloquium
on Monitors and Condition Assessment Equipment
Digest No. 1996/186, Leatherhead.
[8] H.Yin, 2002, ViSOM-A Novel Method for Multivariate
Data Projection and Structure Visualisation, IEEE
Trans. on Neural Networks. vol.13, Part.1, 237-243.
[9] L. Tarassenko, 1998, A Guide to Neural Computing
Applications, Arnold, ISBN 0 340 70580 2.
[10] R.L Harris, 2000, Information Graphics, Oxford
University Press Inc, USA.
[11] C. Chatfield, 1983, Statistics for Technology.
[12] W3C website, 2002, http://www.w3.org/XML/.