Electric Power Systems Research 77 (2007) 15521559

Adaptive relays for overhead line protection

V. Calderaro, V. Galdi, A. Piccolo , P. Siano
Department of Information and Electrical Engineering (DIIIE), University of Salerno, via Ponte don Melillo, 1, 84084 Fisciano, Salerno, Italy
Received 14 March 2006; received in revised form 18 September 2006
Available online 8 December 2006

Abstract
In the liberalized energy market scenario protective relays play an important role in assuring continuous service in the power system where a
malfunctioning could lead to serious damages to a wide number of operators having access to the power system. Considering that power lines are
operated many times below a rated load current, in this paper an adaptive procedure is presented in order to manage power distribution systems
according to dependability or security requirements. In particular, a procedure to obtain an inverse time trip curve by means of a microprocessor,
connected to a relay, is presented. The procedure adapts the trip characteristic depending on the conductor temperature, wind speed, emissivity
and solar absorbity and is implemented on a microprocessor Rabbit 2200 considering a Drake conductor, 795 kcmil 26/7 ACSR.
2006 Published by Elsevier B.V.
Keywords: Protection system; Overhead line; Adaptive relay

1. Introduction
The reliability and efficiency of power systems depend on the
use of the automation in transmission, substation and distribution
systems, which is growing fast and is becoming increasingly
handy with the rapid development of the modern communication
technology [1].
The automation control system must be able to manage a
wide variety of emerging intelligent electronic devices (IEDs)
including new protection systems, power electronic equipment
for power quality management and smart capacitor banks. In particular, there have been significant advances in the field of protective relaying due to both the impact of computers and microprocessors and the introduction of the adaptive relaying concept.
Protective relays play an important role in the liberalized
energy market scenario since they assure continuous service in
the power system where a malfunctioning could lead to serious
damages to a wide number of operators having access to the
power system.
Traditionally, in distribution systems, transformers, cables
and overhead lines have been operated below a rated load

Corresponding author. Tel.: +39 089 964210/84; fax: +39 089 964210/84.
E-mail addresses: [email protected] (V. Calderaro),
[email protected] (V. Galdi), [email protected] (A. Piccolo),
[email protected] (P. Siano).
0378-7796/$ see front matter 2006 Published by Elsevier B.V.
doi:10.1016/j.epsr.2006.11.001

current. However, the rating values, provided by equipment

manufactures, are set by regulations based on particular critical conditions. Nevertheless ambient and operative conditions
are often not as severe as the worst case, therefore, equipments
can handle higher load currents, while remaining within their
design parameters [2].
In order to allow a longer continuous service, during overload
conditions, relays can be used to protect equipments by adapting
their characteristics to ambient and technical conditions.
Several studies in literature deal with adaptive relays, focusing the attention on the application aspects or on the adaptation
methodology. In particular in [3] the objective is the investigation of the effect of mid-point compensation of STATCOM on
the performance of impedance distance relay under normal load
and fault conditions and the implementation of an adaptive distance relaying scheme for transmission line protection. In [4,5]
a Multi Agent System based on adaptive protection relay system
is presented. In [4] the architecture and the function model of the
system are proposed and the validity of the proposed scheme is
verified by means of simulations. In [5] a cooperative protection
system, which processes local data in order to achieve adaptive protection, is presented. The effect of distributed generation
(DG) on the coordination protection is explored in [6] where a
high DG penetration on distribution system suggests an adaptive
protection scheme as a solution to some identified problems.
As stated by many authors, overhead lines are operated below
a rated load current and several papers evaluate the lines thermal

V. Calderaro et al. / Electric Power Systems Research 77 (2007) 15521559

rating. In [7] an adaptable methodology to establish overhead

static thermal ratings, based on a statistical analysis of historical
weather observations, is delineated. Nevertheless, the used worst
case approach appears inadequate in a liberalized energy market
characterized by a high competitiveness [8]. As consequence, in
[9], in order to predict the thermal behaviour on short and long
time horizons, also in the presence of data uncertainties, an interesting approach, based on the use Affine Arithmetic, is proposed.
An adaptive protection system for distribution lines is presented in this paper. The proposed approach takes into account
the variability of several technical and ambient parameters. In
particular, a novel procedure, able to obtain an optimal inverse
time trip curve, using microprocessor connected to a relay, is
presented. The procedure adapts the trip characteristic, depending on the conductor temperature, wind speed, emissivity and
solar absorbity thus being beneficial to power system in terms of
continuity of service without dealing with possible faults missed
by conventional relays.
In the first part of the paper the adaptive protection concept and the architecture of the considered protection system
are presented. The implemented algorithm is, then, illustrated in
Section 2, while some simulations results are shown in Section
3. Conclusions and discussion are presented in the last section.
2. Adaptive concept and the architecture of the
protection system
Actually, protection systems used in distribution systems
function in an off-line predetermined manner. Technicians and
engineers, after analyzing most of the critical power system
conditions, select the parameters for relay setting.
With the development of microprocessor-based relay, some
traditional obstacles, in determining real time settings of relay
for all normal and abnormal operating conditions, are overcome.
In fact in order to change the relay settings according to external
conditions, microprocessor-based relay can collect information,
handle complex logic and communicate with other equipments
by continuously monitoring and analyzing the power system.

1553

Protection systems are designed in order to initiate switching

actions and to clear faults rapidly. Standard protections are willing to accept false trips in order to ensure the reliability of fault
isolation, minimizing damages to system components and are
appropriate only when power system operates in normal state
(dependability approach). If the system is under stress, unnecessary trips will contribute, very likely, to wide spread system
failures, thus compromising system security. Therefore, a security approach will consist in a contrary management of the trip
operations [10].
For mature high voltage or extra high voltage systems the
bias is generally toward dependability because the power system
itself is redundant and it is vital to clear faults with certainty.
On the other hand, distribution systems are not redundant and
therefore security is the most important aspect [11].
An adaptive approach allows to shift between dependability
and security approach, impeding storms or heavy loading and
ensuring longer operative service.
Several definitions have been given in [12] about the adaptive protection and from their analysis two important needs are
clear:
1. computing and transmitting new operating parameters from
a central unit in order to alter automatically and on line the
protection system settings, functions and characteristics;
2. finding an optimal parameter configuration which results
in the most desirable operation, instead of a conventional
trade-off between conflicting requirements (dependability
and security).
However, with regard to the adaptive relay proposed in this
paper, as the application is related to a single line protection,
a central unit will not be necessary in order to change relay
settings.
The proposed architecture, shown in Fig. 1, consists of a
breaker, sensors, useful in obtaining current value and ambient conditions and an external microprocessor, which computes
the trip characteristic according to the actual conditions. The

Fig. 1. Architecture for an adaptive overhead lines protection.

1554

V. Calderaro et al. / Electric Power Systems Research 77 (2007) 15521559

variables causing the change of the operative conditions are

temperature, wind speed and solar absorbity.
The breaker protects an overhead line by faults and overloads. In particular, in this paper attention has been concentrated
on overload protection in order to achieve longer operative service. The sensors measure line current, wind speed and solar
absorbity. Starting from an average current value, obtained by
the acquisition of current values for a t time, the microprocessor evaluates the conductor temperature by means of previously
loaded algorithms and considering the history previous to the
measurement. According to an adaptive procedure, described
in the next paragraph, a new trip characteristic is evaluated and
stored into the relay. The adaptive procedure, in this way, aims at
overcoming the protection system management based on worst
case situation by means of static rating trip values.
3. Mathematical formulation and adaptive procedure
As known, the trip characteristics of the relay is closely correlated to the thermal behaviour of the power line.
Several mathematical models for dynamic thermal transient
of power lines have been proposed in literature [13,17]. In this
paper a classical model based on the following dynamic heat
equation is used [13]
dTc
1
[I 2 R(Tc ) + qs (t)
=
dt
mCp
qc (t, Tc , Ta , Vw , ) qr (t, Tc , Ta )]

(1)

where Tc and Ta are the conductor and ambient temperatures,

Vw and the wind speed and direction, qc and qr the heat fluxes
due to convection and radiation, respectively, qs the solar heat
gain, I and R the conductor current and resistance, and mCp is
the total heat capacity of the conductor.
In order to evaluate the time constant of the transient, an
approximation of Eq. (1) can be considered making linear certain
nonlinear terms in Eq. (1). Considering the equation term by
term, reported in details in Appendix A, it may be seen that the
ohmic heating term and the forced convection equation qc are
linear in Tc ; the solar heating term qr is independent of conductor
temperature while the radiation heat loss term and the natural
convection (zero wind speed) qr are both nonlinear in Tc . So,
considering several methods described in [1416] the radiation
cooling equation becomes linear in Tc and the following linear
non-steady-state heat balance equation can be solved

where Ti is the starting conductor temperature, and Tf is the final

conductor temperature of the thermal transient and the conductor
resistance, included in k2 in the Eq. (2), is that corresponding to
the average conductor temperature (Tf + Ti )/2 [13].
Starting by the above equations the adaptive procedure can
be introduced.
The first step for the adaptive procedure is a direct measurement of the wind speed, the solar absorbity and the current
flowing in the conductor in a time interval t. The average current value, valuated on several acquisitions in the time interval
t, is assumed constant until the next acquisition and is used to
evaluate the conductor temperature. In order to avoid overloads
in correspondence of current values, not considered during the
phase of acquisition, the time interval between two acquisitions
has to be much smaller than the thermal transient time constant.
The identification of the optimal trip curve begins after the
evaluation of the conductor temperature. In particular, the temperature of the conductor Ti at time t0 can be deduced by
considering its thermal history and solving Eq. (1)
 t0
dTc (t)
dt
(6)
Ti =
dt
The next temperature Tf for the determination of the adaptive
trip curves will be the theoretical final temperature obtained by
solving Eq. (1) in the [t0 , [ interval with fixed operational and
external variables and with starting condition Tc (t0 ) = Ti .
When the final temperature is determined, the evaluation of
the optimal time trip curve starts. As known, the overload current
can vary in a range [Imin , Imax ] and, consequently, the overload
final temperature TfOL reached by the conductor and its thermal transient change in correspondence of different overload
currents IOL [Imin , Imax ]. The proposed adaptive procedure
consists in dividing the protection overload current range, for
a discrete number of possible overload currents, in n regular
intervals and in solving Eq. (1).
For each value of IOL [Imin , Imax ], the transient thermal
conductor curves are shown in Fig. 2.
Considering the intersection between the maximum temperature value tolerable by the conductor Tc max and the thermal
transient curves, a set of couples (t, IOL ) = [(t1 , IOL1 ), (t2 , IOL2 ),
. . ., (tn , IOLn )], in correspondence of the tolerable maximum
temperature, is obtained as shown in Fig. 2.

d(Tc Ta )
(2)
= k1 (Tc Ta ) + k2 I 2
dt
Doing so, for a step change in electrical current, the solution
of linearized heat balance equation, that considers all terms taken
into account in (1), is
Tc (t) = Ti + (Tf Ti ) (1 et/ )
Tf = Ta
=

k2
I2
k1

(Tf Ti ) mCp
R(Tc ) (If2 Ii2 )

(3)
(4)
(5)
Fig. 2. Elaboration for the trip curve protection.

V. Calderaro et al. / Electric Power Systems Research 77 (2007) 15521559

So, a new trip characteristic time/current is deduced interpolating the couples of points (ti , IOLi ), as shown in Fig. 3.
Nevertheless, if the subsequent average current value I1 ,
obtained by an elaboration of a series of measurements and
acquisitions, is smaller than the previous average current value
I0 , the adaptive procedure does not begin and the trip curve for
the relay does not change.
The procedure is started up only if for two or more subsequent
acquisitions of current the average values I1 , I2 , . . ., In are such
as:

I(t0 ) < I(ti ),
i = 1, 2 . . . , n
(7)
|I(ti ) I(ti+1 )| < , i = 1, 2 . . . , n 1
where is a variation of 5% between I(ti ) and I(ti+1 ).
Such assumption has been introduced in order to avoid trip
curve adaptation in correspondence of fast current variations.

Fig. 4. Procedure for adaptive protection.

Fig. 3. Inverse time trip curve.

1555

1556

V. Calderaro et al. / Electric Power Systems Research 77 (2007) 15521559

Fig. 5. Comparison between the classical and new trip curves.

Naturally, the number of similar subsequent acquisitions

depends on the relationship between the time constant of the
thermal transient and the time interval between two acquisitions
of current.
The adaptive procedure can be explained by means of the
flow chart in Fig. 4, where the operations in the blocks Temperature evaluation, Evaluation Average current value and
Determining trip curve are run by the microprocessor, and
Acquisition data by the sensors. ta is the acquisition time
and t is the time interval between two subsequent acquisitions.
Counter and max are internal variables of the microprocessor
taking into account, respectively, the number and the maximum
number of times in which subsequent average current values are
similar according to Eq. (7).
In order to take into account problems related to noise introduced during the adaptive procedure, a reset operation must be
executed periodically.
The new trip characteristic is an inverse time curve that can
give a longer continuous service in case of tolerable overloads to
the distribution system against the classical time delayed curves
used to protect conductors by overload. The depicted trip curve,

Fig. 6. Overload protectiontrip curves for variable current flow.

shown in Fig. 3, is related to the overload zone until kIr , where

Ir is the rating current of conductor and k varies between 2 and
10. For current values greater than kIr the typical independent
time trip characteristic can be set.
In comparison with the classical step trip curve, a longer
continuity of service can be observed in Fig. 5 and quantified by
means of the following relationships
 I1
 I2
1 =
(t(I) t1 ) dI,
2 =
(t(I) t2 ) dI,
I2
I3

I3


3 =

I4

(t(I) t3 ) dI

(8)

where I1 , I2 and I3 with t1 , t2 and t3 are the current and time

thresholds classically applied to the protections of the distribution lines starting from the primary stations of the distribution
systems and I4 is the pick-up value, that is the minimum current
at which the relay certainly operates.

Fig. 7. Overload protectionadaptive curve varying speed wind.

V. Calderaro et al. / Electric Power Systems Research 77 (2007) 15521559

1557

Table 1
Simulation parameters
Parameter

Value

Analysis time
Latitude
Maximum allowable temperature

08:00 a.m. to 15 00 p.m.

40.85 N
80 C

Conductor resistance

R (25 C) = 7.28 e005 /m

R (75 C) = 8.68 e005 /m

Line orientation
Wind speed
Emissivity and absorptivity
Maximum carrying current

Eastwest direction
0.61 m/s 3600 s/h
0.5
850 A

Fig. 8. Overload protectionadaptive curve varying solar absorbity.

4. Case study
In this section the previously described adaptive approach is
implemented in order to evaluate real time trip characteristics,
when varying ambient or operative conditions.
The microprocessor is Rabbit series 2200 programmed in
Dynamic C and the interpolation of the numerical results is
carried out by Matlab in order to show the variation of the
time/current trip characteristic.
The conductor considered for the analysis is a Drake conductor, 795 kcmil 26/7 ACSR with a constant thermal time equal to
837 s [13]. The maximum temperature for a Drake conductor is
80 C and the analysis is performed with an acquisitions time ta
equal to 60 s. Operational and external parameters are valuated
by the Rabbit microprocessor in about 2030 s. The subsequent
series of current acquisitions is executed after t = 300 s from
the previous one. If the average current is smaller than the previous average value, the elaboration of the trip curve does not start.

The adaptation is carried out if for two subsequent acquisition

sets Eq. (7) is true (max = 2).
The trip curves, obtained starting from different average current values are shown in Fig. 6. The results have been presented
for the condition considered in Table 1.
Even if the current of the conductor remains the same, an
adaptation of the trip curve can be obtained by varying others
parameters like wind speed, emissivity or absorptivity. In particular, emissivity and absorptivity vary generally from about
0.2 to about 0.9 with age. The exact rate of increase depends
on the level of atmospheric pollution and on the lines operating voltage. In particular, the emissivity coefficient describes
the correlation between the surrounding atmosphere and surface
condition of the line. This coefficient is specified in the range
[0.230.98] where the small values are for new conductors. The
correlation between the coefficient and atmospheric pollution

Fig. 9. Adaptive sequence of trip curve for overload protection.

1558

V. Calderaro et al. / Electric Power Systems Research 77 (2007) 15521559

can be expressed considering the number of year the conductor

has been energised Y [18]:

Appendix A
Convective term:

Y
= 0.23 + 0.7
1.22 + Y

(9)

Specific details about the correlation between dust and overhead line capacity are in [19].
Absorptivity is generally higher than emissivity over the life
of the conductor.
So, placing a measurement device to evaluate wind and
recording the age of the conductor, a variable trip curve can
be obtained. The trip curves obtained by varying wind speed
between 0.61 and 0.70 m/s, in overload range, are shown in
Fig. 7 starting from an average current value equal to 500 A.
In the same figure a case with 3 m/s is shown too in order to
evidence the dramatic changing of the trip curve.
In the proposed system, in order to consider a precautionary
approach, the wind speed is acquired for several points along the
line and the smaller value is taken into account for the adaptive
procedure because, as known, the smaller value corresponds to
lower the capacity of the line and represents the worst case.
The last adaptive trip curves, in overload range, have been
obtained by varying the solar absorbity between 0.3 and 0.6 and
the trip curves presented starting from an average current value
equal to 500 A are shown in Fig. 8.
The adaptive sequence of real time trip curves, for a time
varying current profile, is presented in Fig. 9.
The proposed approach is robust because even if the sensor
measurements could be inaccurate, the worst case trip curve is
implemented in order to guarantee security to the power system.
Moreover, the small dimension of the Rabbit, its processing characteristics and the quickness in the trip curve elaboration justify
the use of microprocessor with respect to a classical computer.
In particular, data processing by means of a simple microprocessor requires few seconds (2030 s) so that, eventually, data
acquisition can be carried out in real time.

5. Conclusions
In this paper an adaptive protection system for overhead lines
has been presented. Even if the adaptive concept is not totally
new, the advent of microprocessor-based relay will permit not
previously possible applications. For this reason an application
of adaptive protection to overhead line has been the handled
issue.
The proposed approach considered the variability of several
operational and ambient parameters. After an overview on the
concept of adaptive protection, a procedure aiming to obtain an
inverse time trip curve by means of a microprocessor, connected
to a relay, has been presented.
A simple architecture of the protection system has been considered and simulations have been run by means of a Rabbit
2200 microprocessor on a Drake conductor. Obtained results
remarked a potential capacity of load not utilized highlighting
the reachable benefits in terms of continuity of service.

qc,0 = 0.283f0.5 D0.75 (Tc Ta )1.25

Df Vw 0.25
q
kf (Tc Ta )
c,low = kangle 1.01 + 0.371
f

Df Vw 0.6

kf (Tc Ta )
c,low = kangle 0.1695
f
Radiative heat loss:



Tc + 273 4
Ta + 273 4
qr = 0.138D

100
1000
Solar heat gain:


qs = Qs sin()A
= cos1 [cos(Hc ) cos(Zc ZL )]

Resistance:


R(Thigh ) R(Tlow )
R(Tc ) =
(Tc Tlow ) + R(Tlow )
Thigh Tlow

References
[1] V. Calderaro, V. Galdi, A. Piccolo, Distribution planning by genetic algorithm with renewable energy units, in: Conference: Bulk Power Systems
Dynamics Control and ControlIREP Symposium August 2227, Cortina
dAmpezzo, Italy, 2004.
[2] V. Calderaro, S. Corsi, V. Galdi, A. Piccolo, Optimal setting of the protection systems in distribution networks in presence of distributed generation,
in: Proc. of 40th International Universities Power Engineering Conference
UPEC 2005, September 79, Cork, Ireland, 2005.
[3] J.V.V.N. Bapiraju, J.U. Shenoy, K.G. Sheshadri, H.P. Khincha, D.
Thukaram, Implementation of DSP based relaying with particular reference to effect of STATCOM on transmission line protection, in: Power
System Technology. PowerCon 2004. International Conference on Volume
2, November 2124, 2004, pp. 13811385.
[4] J.H. Chen, S.H. Chen, Y.M. Yang, Study on adaptive protection relay
system based on multi-agent, in: Proc. of International Conference on
Machine Learning and Cybernetics 2005, Volume 1, August 1821, 2005,
pp. 114118.
[5] M.-Y. Yang, Y.-L. Zhu, A cooperative protection system with multi-agent
system, in: Transmission and Distribution Conference and Exhibition: Asia
and Pacific, 2005 IEEE/PES, August 1518, 2005, pp. 14.
[6] S.M. Brahma, A.A. Girgis, Development of adaptive protection scheme for
distribution systems with high penetration of distributed generation, Power
Deliv. IEEE Trans. 19 (January (1)) (2004) 5663.
[7] F.R. McElvain, S.S. Mulnix, Statistically determined static thermal ratings
of overhead high voltage transmission lines in the Rocky Mountain region,
Power Syst. IEEE Trans. 15 (May (2)) (2000) 899902.
[8] T.L. Le, M. Negnevitsky, M. Piekutowski, Expert system application for
the loading capability assessment of transmission lines, Power Syst. IEEE
Trans. 10 (November (4)) (1995) 18051812.
[9] A. Piccolo, A. Vaccaro, D. Villacci, Thermal rating assessment of overhead
lines by Affine arithmetic, Electr. Power Syst. Res. 71 (November (3))
(2004) 275283.

V. Calderaro et al. / Electric Power Systems Research 77 (2007) 15521559

[10] M.S. Sachdev, B. Chattopadhyay, T.S. Sidhu, A new approach to distribution system protection-adaptive relaying, in: Fifth International Conference
on Developments in Power System Protection, 1993, pp. 165168.
[11] A.G. Phadke, S.H. Horowitz, Adaptive relaying, Comput. Appl. Power
IEEE 3 (July (3)) (1990) 4751.
[12] M.J. Damborg, M. Kim, J. Huang, S.S. Venkata, A.G. Phadke, Adaptive
protection as preventive and emergency control, in: Summer Meeting Power
Engineering Society. IEEE vol. 2, July 1620, 2000, pp. 12081212.
[13] IEEE standard for calculating the current-temperature relationship of bare
overhead conductors, IEEE Std 738-1993, November 8, 1993.
[14] W.Z. Black, W.R. Byrd, Real-time ampacity model for overhead lines, IEEE
Trans. Power Apparatus Syst. PAS-102 (July (7)) (1983) 22892293.

1559

[15] W.H. McAdams, Heat Transmission, third ed., McGraw-Hill, New York,
1954.
[16] Determination of Bare Overhead Conductor Ratings Conductor Rating
Task Force, PA, NJ, and MD Interconnection, May 1973.
[17] G.J. Anders, Rating of Electric Power Cables in Unfavorable Thermal
Environment, Wiley-IEEE Press, May 2005.
[18] T.L. Le, M. Negnevitsky, M. Piekutowski, Expert system application for
the loading capability assessment of transmission lines, IEEE Trans. Power
Syst. 10 (November (4)) (1995).
[19] M.M. Salama, Thermal performance of an overhead transmission line under
the influence of dust accumulation, Energy Convers. Manage. 41 (August
(12)) (2000) 13231334.