Proceedings of the International Conference on Structural Health Monitoring of Intelligent Infrastructure

Porto, Portugal, 30 June - 2 July 2021

ISSN 2564-3738

Using structural health monitoring to estimate the remaining lifetime due to fatigue
at a conductor from an overhead high-voltage transmission line
Raphael Faria de Mendonça1, Elsa Caetano2
1,2
CONSTRUCT-VIBEST, Department of Civil Engineering, Faculty of Engineering, University of Porto, R. Dr. Roberto Frias,
4200-465 Porto, Portugal
email: raphael.mendonca@fe.up.pt, ecaetano@fe.up.pt

ABSTRACT: This article presents the work developed in the context of the study of the turbulent wind effects on a high-voltage
transmission line. Turbulent wind loads are in the origin of vibrations, namely Aeolian vibrations, inducing damage at the cable
conductors and accessories. With the purpose of investigating such effects, a continuous dynamic monitoring system was
developed and installed on a span of a de-activated transmission line. In order to validate existing fatigue models applied to the
estimation of the remaining life of conductors, measured accelerations are processed, and CIGRÉ’s standard rules for the fatigue
evaluations are applied. The experimental results are then mathematically manipulated according to the Poffenberger-Swart
procedure, and the remaining lifetime estimation of the conductor is determined using the Rainflow method and Miner’s rule.

KEYWORDS: Aeolian vibrations; Conductors fatigue; Conductor remaining lifetime; Dynamic monitoring; Cable failure.

1 INTRODUCTION Santo Tirso, and Valongo and is currently de-activated. This

Despite the high level of standardisation currently adopted in span has an approximate length of 595 m, crossing an area of
the design of overhead high voltage transmission lines rural characteristics, with sparse vegetation and low
(LMAT), there remains a considerable lack of knowledge construction density. Images of this span and the instrumented
regarding the characterisation of the turbulent wind effects, the tower are shown in Fig.1, and an aerial view of the crossed area
structural behaviour of conductors, their dynamic interaction in Fig. 2. The choice of this span was based on the length
with the towers, and also the definition of damping levels greater than 500 meters and the location outside urban area with
achieved in service and extreme conditions. The occurrence of good exposure to the wind.
damage and collapse in LMATs is not uncommon, both in
terms of conductors and insulators, as well as in towers, which
can cause disturbances in the distribution of electricity with
significant economic repercussions. On the other hand, the
difficulties related to the instrumentation of live-lines have
limited the experimental characterisation analysis of these
structures’ behaviour.
However, technological developments in recent years have
opened new perspectives in the monitoring of these structures.
In this context, the Laboratory of Vibrations and Structural
Monitoring (ViBest) at FEUP has designed a continuous
dynamic monitoring system and implemented it on a LMAT’s
representative span of the electricity transmission network’s
infrastructure, taking into account the greenfield location and
the atmospheric conditions. The monitoring system aimed at
Figure 1. LRA.VM1 Line: a) span e b) instrumented tower.
studying, among others, the effects of aeolian vibrations, which
are the leading cause of the conductors’ fatigue failure.
This article aims to present the continuous monitoring
system, and some results of a study focused on the assessment
of the remaining lifetime of a conductor, comparing the
dynamic cable behaviour before and after installing
Stockbridge vibration dampers.

2 CHARACTERISATION OF THE LMAT

Location
The monitored span is part of the Riba de Ave - Vermoim 1
Line, which runs along the cities of Vila Nova de Famalicão, Figure 2. Aerial view of the zone crossed by P43-P44 span.

1233
Proceedings of the International Conference on Structural Health Monitoring of Intelligent Infrastructure

Cables accelerometers in two horizontal directions allows the

The conductor cables in the monitored span are of the ACSR characterisation of the dynamic interaction between these two
(Aluminum Conductors Steel Reinforced) type, consisting of substructures. The monitoring system was complemented with
aluminium wires with a reinforcement core of steel wires, and four accelerometers, located in the first and third measurement
the cross-section as shown in Fig.3a. The LMAT also has a sections closest to the tower. Images of the accelerometers
guard-cable of the same type but with a smaller number of installed at the tower and conductor are shown in Figs. 6 and 7.
wires and a smaller cross-sectional area, schematically shown Finally, to characterise the wind actions on the conductor and
in Fig.3b. tower, a sonic anemometer was installed at the top of the tower
(see Fig. 8).

Figure 3. Cables cross-section a) conductor and b) guard-

cable.
The instrumentation was applied to a conductor cable with a
24 mm diameter and reference BEAR - ACSR 325. With the
purpose of subsequent calibration of the numerical model, a
topographic survey was carried out, which allowed defining the
installed tension and the cable profile. This information is
systematised in Fig.4. It is also noted that no connection joints
or repairs were observed at the cable during the installation of
the monitoring system.
Figure 5. Instrumented section locations in the P43-P44 span.

Figure 6. Accelerometers: a) piezoelectric (tower) and b)

MEMs (conductor).
Figure 4. Cable profile and installed tension force.
Towers
The towers P43 and P44 supporting the LMAT’s monitored
span, are equal and designated as T1-29, belonging to the “T”
family towers with 29m height. These are common trussed steel
structures constituted by angle profiles and connected by bolts
and gusset plates.
Structural health monitoring system
With the main aim of characterising the conductor’s vibrations Figure 7. Accelerometers installed: a) piezoelectric (tower)
and the potential dynamic interaction with the towers, a and b) MEMs (conductor).
structural health monitoring system has been designed and
implemented. Having the intention of using this system in an
activated line, fibre optical technology was incorporated into
the system by means of specific developed Bragg sensor setups
installed in five sections of the cable according to the
representation of Fig.5.
The measurement of in-plane and out-of-plane strains in the
conductor at more than one section allows the identification of
modal components associated with the conductor’s vibration.
Simultaneous measurement of the tower vibrations using two Figure 8. Anemometer installed at the top of the tower.

1234
 Proceedings of the International Conference on Structural Health Monitoring of Intelligent Infrastructure

Data acquisition
The acquisition system was installed near the base of the P43
tower, accommodating all devices for data acquisition, storage
and management, as well as communication and power supply.
The monitoring system was active for a period of two years. In
order to assess the performance of Stockbridge dampers, a
conventional damping solution for the span was designed and
implemented, which permitted the monitoring of the damped
conductor during a period of 6 months.

3 FATIGUE, DAMPERS, AND VIBRATION RECORD

DEVICES
Fatigue at cable conductors
Significant attention should be given to cable conductors since
these components represent up to 40% of a transmission line’s
total investment. Hence, maintenance is essential for electricity
distributor companies. However, it is not uncommon in this
kind of structure to confront one of the most complicated
mechanical problems related to damage and failure: fretting
fatigue. Fretting happens due to relative movement between the
conductor’s internal wires and by their contact with clamps and
dampers. In overhead high-voltage transmission lines, this
phenomenon occurs mainly due to aeolian vibrations, also
designated as vortex-shedding resonance. These vibrations
occur for particular wind velocities and are due to the formation
of alternate vortices at the top and bottom of the conductor,
inducing a vertical cable movement that results in bending at Figure 9. CIGRÉ Safe Border Line and S-N curves [1]
the anchorage or suspension point near the tower. The
association of bending and tension stresses at the cable may
conduct to fatigue of the wires. Table 1. Constant values A and B for the CSBL curve.
Aeolian vibrations in cable conductors occur at low wind
velocities from 0.5m/s to 7m/s, in a frequency range of 3Hz to # of layers N < 2x107 N > 2x107
150Hz, and the vibration amplitudes are expressed as a function A B A B
of the conductor’s diameter, with values around 0.01 to 1 1 730 -0.2 430 -0.17
diameter [1]. >1 450 -0.2 263 -0.17

S-N curves and CIGRÉ’s Safe Border Line (CSBL)

Every Day Stress (EDS) and bending stress concept
The characterisation of the fatigue performance of a material is
commonly achieved using the stress-life method with the aid of The Every Day Stress is the tensile strength established by
S-N curves (Wöhler curves), which plot in a graph the cyclic CIGRÉ as a portion of the cable conductor’s Rated Tensile
stress (S) versus the number of cycles to failure (N) during a Strength (RTS). In general, in overhead high-voltage
laboratory test. The test stop criteria are either the failure of transmission lines, EDS values vary in a range of 15% to 20%
10% of the cable’s wires or of a specific number of wires. Due of the RTS. Although fatigue failure in conductors is intimately
to the difficult execution and expensive costs involved in this linked with high EDS values, the increasing energy demand has
kind of test, the International Council on Large Electric been leading companies, over the years, to use higher values of
Systems (CIGRÉ) provided a safe curve, called Safe Border EDS, which can easily achieve 35%.
Line, in order to make it easier for companies to determine the The accurate assessment of stress and strain at conductors is
lifetime of conductors without a substantial investment. Fig.9 a difficult task. Poffenberger and Swart [2] developed an
shows the CSBL curve in comparison with other test results for analytical formulation that establishes a relationship between
a variety of cable compositions [1]. the dynamic bending stress (σPS) at the wires’ outer layer and
The following equation can express the Safe Border Line the peak-to-peak vibration amplitude (YB). The Poffenberger-
curve Swart equation can be expressed as

Va A ˜ NiB (1) V PS K ˜ YB (2)

The constants A and B are related to the number of fatigue The constant K in this equation converts the vertical
cycles Ni for a specific level of stress and number of aluminium amplitude measured at 89 mm from the last point of contact
wire layers of the conductor. The stress amplitude is designated (LPC) into bending stress (0-to-peak), as shown in Fig.10, and
as σa and expressed in MPa. Table 1 shows the values of these is defined as
parameters according to [1].

1235
Proceedings of the International Conference on Structural Health Monitoring of Intelligent Infrastructure

Ea ˜ d ˜ p 2 should be placed at a distance of λ/4 from the fixed point, as

K  p˜ x
[ N / mm3 ] (3) shown in Fig.12, where λ is the wavelength of the mode to be
4 ˜ (e  1  p ˜ x) mitigated.
However, taking into account that the natural frequencies of
where Ea is the aluminium Young’s modulus, expressed in the assembly damper-cable are different from the natural
MPa, d is the wire diameter in the outer layer, expressed in mm, frequencies of the isolated cable, CIGRE recommends the
x is the distance between the measurement point and LPC, damper installation at a distance [4]:
usually adopted equal to 89 mm, and the parameter p is
described by O
x 0.85 ˜ (6)
2 T 2
p (4)
EI Where the wavelength is given by:
In equation (4), T is the tension force at the conductor 1 T
calculated using EDS, expressed in N, and EI is the conductor’s O ˜ (7)
minimal flexural stiffness in N.mm2 given by the following fn m
equation
In equation (7), fn is the cable natural frequency in Hz, T is the
S d a4 S d s4 tension force, expressed in N, and m is cable mass per unit of
EI min na ˜ Ea ˜  ns ˜ Es ˜ (5) length, expressed in kg/m.
64 64
where na is the number of aluminium wires, Ea is the aluminium
Young’s modulus, and da is the aluminium wire diameter. The
other parameters ns, Es, and ds are the same, but now regarding
to the steel wires present in a multi-layer cable conductor.

Figure 11. Stockbridge damper attached to the cable

conductor [3].

Figure 10. Schematic view of the conductor’s clamp and the

vibration amplitude measurement point.

Stockbridge dampers
Stockbridge dampers are the most widely used devices for
aeolian vibration control. In general, they consist of two rigid Figure 12. Theoretical and recommended distance of the
masses connected to both ends of a messenger cable. A clamp Stockbridge installation.
ensures the connection between the damper and the conductor.
This connection allows the transmission of displacements from
the conductor to the damper (Fig. 11). The inertial masses The resonance phenomenon occurs when the wind excitation
attached to the messenger cable induce a cable bending, frequency (Strouhal frequency) approximates the natural
causing energy dissipation by friction due to the relative frequency of the cable. Considering the Strouhal number equal
movements between the internal wires of the cable. Energy to 0.185, which is the recommended number for the specific
dissipation occurs in a frequency band close to their natural case of cables in overhead high-voltage transmission lines,
frequency, reaching maximum values when the Stockbridge is critical wind frequency can be extracted from the expression
excited in that frequency range [3]. If the sum of the energy
0.185 ˜U
dissipated by the conductor and the Stockbridge damper is f (8)
greater than the energy supplied by the wind, the conductor will D
vibrate with less amplitude and for less time.
Where U is the wind velocity in m/s, and D is the diameter of
However, to guarantee the system efficiency, it is essential
the conductor, expressed in m.
that dampers are placed in the correct position along the cable,
usually at the points of the cable with the largest amplitude of
vibration and closest to the anchored points. Installing the VIBREC500 vibration recorder
Stockbridge dampers at incorrect positions can result in an As part of the assessment of aeolian vibrations on transmission
increase in mechanical overload on the cable, which will lead line cables, devices used to measure vibration amplitudes and
to a reduced lifetime. Theoretically, a Stockbridge damper calculate the remaining lifetime are commonly used. In this

1236
 Proceedings of the International Conference on Structural Health Monitoring of Intelligent Infrastructure

work, the device VIBREC500-WT (Fig.13), manufactured in conductor is determined using the Miner’s rule using the
Switzerland, by Pfisterer Sefag was used to measure the aeolian expression
vibrations, recording peak-to-peak amplitudes, vibration m
ni
frequencies and temperature. The VIBREC weight is about 0.7
kg, including the battery, and has an autonomy of
D ¦N
i 1
(9)
fi
approximately one year, depending on the ambient temperature
and the acquisition time interval. The device was developed for and the procedure shown in Fig.14. The S-N curve obtained
cables supported by suspension clamps. Therefore, they are with the recorded data is compared with the CIGRÉ’s Safe
usually installed next to these sections, and the measurement Border Line [1], obtaining the final damage value, which is
sensor is located at a distance of 89 mm from the clamp. extrapolated to a one-year period.

Figure 14. VIBREC500-WT device at suspension clamp.

Finally, the remaining lifetime is given by the following
equation
1
V (10)
¦D
Figure 13. VIBREC500-WT device at suspension clamp.
Where D is the accumulated damage and V is the remaining
Due to the necessity to manage the battery life and memory, lifetime in years.
the VIBREC500-WT does not continuously record data. The
standard acquisition consists of storing 10 seconds of active- 4 REMAINING LIFETIME ESTIMATION METHOD
time data within 15 minutes of an idle interval, following the As the objective of evaluating and comparing whether the
recommendation of CIGRÉ, which determines a minimum signal acquisition time stipulated by VIBREC is sufficient to
monitoring period of three months. Assuming that the device characterise the phenomenon of fatigue in conductor cables, the
measures for three months, only approximately 24.8 hours of present work proposes a method of calculating the remaining
data would be recorded, which is equivalent to 0.3% of the lifetime due to fatigue using structural health monitoring data.
period of one year. This low percentage is one of the main
reasons for carrying out this work, which aims at characterising Periods of analysis
the quality of lifetime cable estimations based on such a short In order to use the same concepts and formulations employed
sampling period. Table 2 shows some VIBREC500-WT by VIBREC500-WT, a micro-electrical mechanical
specifications. accelerometer (MEM#3) has been installed close by the
Table 2. VIBREC500-WT data acquisition specifications anchorage point to simulate the device position. This sensor
measures the vertical accelerations of the cable (in-plane).
Active-time of acquisition 1 – 10s In possession of a relevant continuous monitoring database,
Active-time + Idle-time 15s – 60min two specific periods of analysis were considered. The first one
Wind velocity 0 – 30m/s was chosen before installing Stockbridge dampers and the
Frequency range 0.2 – 200Hz second one after the installation, both with a three-month
Amplitude of vibration 0 – 2mm duration as recommended by CIGRÉ. Different weather
conditions were also considered when choosing the period to
The VIBREC500-WT is also capable of calculating the take into account the effects of temperature variations. One
remaining lifetime of the conductor with a software aid. The interval was defined during three months in the summer
analysis performed by the software is following the most recent without dampers, and another three months of data were
IEEE [4] and CIGRÉ [5] standards. Accordingly, the measured collected in the winter with the dampers already installed.
peak-to peak amplitudes are converted into bending stresses, Fig.15 and Fig.16 show the 10-minutes wind mean velocity in
using the Poffenberger – Swart equation (2). both periods of analysis.
Regarding cycle counting, the VIBREC500-WT manual The calculations of the conductor lifetime were then made
states that the device automatically stores the number of half- considering different record durations, starting in 5 s in every
cycles in a data matrix, and the software converts the data into 15-minute period, and ending with the full 15-minute period.
a number of full-cycles.
The data are expressed through an S-N curve relating each
block of stress to the number of cycles, and the damage at the

1237
Proceedings of the International Conference on Structural Health Monitoring of Intelligent Infrastructure

Once the displacement signal is converted into a stress signal,

the cycle counting Rainflow method is applied, and Miner’s
rule is used to determine the conductor’s remaining lifetime.
Table 3. Poffenberger-Swart equation parameters
Aluminium Young’s modulus, Ea 70000 MPa
Steel Young’s modulus, Es 200000 MPa
Aluminium wire diameter, da 3.35 mm
Steel wire diameter, ds 3.35 mm
Number of aluminium wires, na 30
Figure 15. 10-minute mean wind velocity during summer. Number of steel wires, ns 7
Conductor Bending Stiffness, EImin 2.163x107 N/mm2
Cable Tension, T 23320 N
Poffenberger-Swart constant, K 31.981 N/mm3
Distance from LPC, x 89 mm

5 REMAINING LIFETIME ESTIMATION RESULTS

Several analyses have been performed using different
acquisition times before and after the Stockbridge installation
to estimate the remaining lifetime of the LMAT’s conductor
and assess the acquisition time used by VIBREC500-WT.
Table 4 and Fig.17 show the results of these analyses.
Figure 16. 10-minute mean wind velocity during winter.
Table 4. Lifetime estimations for LMAT’s conductor
Acquisition Life [years]
Processing of acceleration signals time [s/15min] Without dampers With dampers
Before using the acceleration signals in the cables to calculate 5 10.05 200
the fatigue lifetime, it is necessary to resort to data processing 10 7.74 167
procedures. The algorithm developed collects the 30 6.09 136
accelerometer signals and then analyses and remove the offsets, 60 5.86 132
outliers and missing values. 90 5.86 127
Another important procedure that should be done is to filter 120 5.77 124
the signal removing the frequency content out-of-range of wind 180 5.54 126
vibrations. The wind-induced vibration frequencies can be 300 5.55 123
determined using the equation (8) and the wind velocity range
600 5.27 119
0.5-7m/s for the cable BEAR-ACSR 325.

0.185 ˜ [0.5, 7]
f | [4 Hz , 60 Hz ] (11)
0.02345

A Butterworth signal processing filter was designed to have

a frequency response in the mentioned passband as flat as
possible.

Signal integration in the frequency domain

The Poffenberger-Swart equation converts the cable amplitude
of vibration into bending stress. This implies that the
acceleration signals should be doubly integrated in the
frequency domain to transform accelerations into velocities and
afterwards velocities into displacements. It is necessary to be
aware of the correct method to apply to integrate discrete Figure 17. Remaining lifetime x Acquisition time.
signals in the frequency domain preventing the drift effect,
which results in a corrupted signal. Table 3 shows the Based on the analyses, the conductor’s remaining lifetime tends
parameters used in the Poffenberger-Swart equation (2) to 5 years when increasing the time of acquisition in the cable
considering the characteristics of the conductor under analysis. without the Stockbridge damper. On the other hand, the cable
analyses with dampers show a considerable increase in the

1238
 Proceedings of the International Conference on Structural Health Monitoring of Intelligent Infrastructure

conductor lifetime, which tends to 119 years. This occurs due ACKNOWLEDGMENTS
to the energy dissipated by the dampers, reducing the amplitude This work was financially supported by: Base Funding -
of vibration and consequently reducing the stresses at the UIDB/04708/2020 and Programmatic Funding -
anchorage point. In Fig.17, it is possible to observe that the UIDP/04708/2020 of the CONSTRUCT - Instituto de I&D em
acquisition time of 10s for 15 minutes used by default in the Estruturas e Construções - funded by national funds through the
VIBREC500-WT is not sufficient for a reasonable estimate of FCT/MCTES (PIDDAC), and by the Project PTDC/ECM-
the lifetime, and even less for time intervals of less than 10s, as EST/2110/2014 – POCI-01-0145-FEDER-016877 - funded by
allowed in the device configuration. It can also be observed that FEDER funds through COMPETE2020 - Programa
the lifetime values obtained for the undamped and damped Operacional Competitividade e Internacionalização (POCI)
condition for the different periods of data analysis exhibit an and by national funds through FCT - Fundação para a Ciência
asymptotic variation after a period of 60s/15min of data e a Tecnologia, I.P. The financial support granted by Fundação
acquisition. Assuming the CICRÉ’s recommendation of data para a Ciência e Tecnologia (FCT) to the first author through
acquisition using 10s/15min time, at the 3-month end, the the doctoral scholarship 2020.07461.BD is also acknowledged.
amount of data analysed is 86400s compared to the 518400s
using 60s/15min. As usual, the lifetime is expressed in years,
and the data is extrapolated for this period. Considering the REFERENCES
VIBREC500-WT period of acquisition, it represents 0.27% of [1] EPRI. EPRI Transmission Line Reference Book: Wind-Induced
a year. As stated before, this time is not enough to characterise conductor Motion. Palo Alto, CA: 2006. 1012317.
the fatigue phenomenon due to aeolian vibrations. Considering [2] POFFENBERGER, J. C.; SWART, R. L. Differential Displacement and
the results obtained with this specific conductor, it is Dynamic Conductor Strain. IEEE Transactions Paper, Vol. PAS-84,
1965, pp. 281-289.
recommended to perform time domain analysis using at least [3] GUEDES, A. V; MATT, C. F.; CAVALCANTI, E. S. C. Experimental
120s/15min for a reliable estimation of the remaining lifetime Investigation of The Dynamic Behavior of Stockbridge Dampers. 18th
of a cable conductor, which represents 1036800s or 3.3% of a International Congress of Mechanical Engineering. Ouro Preto/MG,
year. Fig.18 shows the accumulated damage curve compared 2005.
[4] IEEE, “Guide for Aeolian Vibration Field Measurements of Overhead
with the CIGRÉ Safe Border Line for 120s of time acquisition Conductors. “std 1368, 2006.
without damper. [5] CIGRE SCB2-08 WG30 TF7 2007. “Fatigue endurance capability of
conductor Conductor/Clamp Systems-update of Present Knowledge.”
CIGRE-TB 332, 2007.
[6] ROCHA, D. Estudo do Comportamento Estrutural de Cabos em Linhas
Elétricas Aéreas de Muito Alta Tensão. Departamento de Engenharia
Civil da Faculdade de Engenharia da Universidade do Porto, 2012.
[7] Cigre SG22 WG01. Report on aeolian vibration. Electra, 1989.
[8] CAETANO, E; BASTOS, F; ROCHA, D. Projeto de Monitorização de
Linha Elétrica Aérea de Muito Alta Tensão- Fase 2. Relatório
Laboratório de Vibrações e Monitorização de Estruturas, Fevereiro 2016.
[9] ASTM E1049-85(2005). Standard Practices for Cycle Counting in
Fatigue Analysis. ASTM International, West Conshohocken, PA, 2005.
[10] FADEL, A. A.; ROSA, D.; MURÇA, L. B.; FERREIRA, J.L.A.;
ARAÚJO, J.A. Effect of high mean tensile stress on the fretting fatigue
life of an ibis steel reinforced aluminium conductor. ELSEVIER:
International Journal of Fatigue. v. 42, pp. 24-34, march. 2011.
[11] FIGUEIREDO, I. M. P. Tempo Efectivo de um Sistema: Ciclos Rainflow
e Ciclos Min-Max. 2004. 79 p. Dissertação de Mestrado – Faculdade de
Engenharia, Universidade do Porto, Porto, Portugal.
Figure 18. Accumulated damage x CIGRÈ Safe Border Line. [12] PFISTERER SEAFAG AG. Vibrec500 WT: User Manual V 1.0.
Switzerland: 2010. 12 p.

6 CONCLUSIONS
The installation of a monitoring system in a LMAT line
allowed continuous observation through data collection and
automatic analysis of the cable conductor’s response with the
main aim to assess the current recommendation of the data
acquisition system of the aeolian vibrations measurement
devices in overhead high-voltage transmission lines. The
proposed method in this work allowed to estimate reliably the
remaining lifetime of the cable due to fatigue before and after
dampers installation, evidencing unconservativeness in the
short-term monitoring using established sampling durations.
Although further studies should be made to validate this
conclusion, it is shown that the use of continuous monitoring
acceleration can provide a good alternative to evaluate the
remaining lifetime of a conductor if compared with the
traditional methods.

1239