APAP2011

The International Conference on Advanced Power System Automation and Protection

• ARTICLES •

CT saturation effects on performance of digital overcurrent relays

COELHO Aurélio Luiz Magalhães1 & SILVEIRA Paulo Márcio1
1
Federal University of Itajubá, Itajubá 37500903, Brazil;

Received ; accepted

In general, modeling the behavior of the core in current transformers (CTs) used for protection and the effects of its operation
outside the linear region are commonly reported in literature. The problem in using a mathematical CT model is to obtain the
parameters in a particular case in order to implement that model easily, efficiently and accurately. In this sense, a CT model was
implemented in MATLAB based on the theory published by the IEEE Power System Relaying Committee (PSRC) to analyze
various factors and conditions that can lead to the CT saturation. The current data obtained from the simulation were converted to
COMTRADE format to be used in a injection test equipment to verify the overcurrent relay response behavior. A similar analysis
was done in the RTDS (Real Time Digital Simulator) perfoming hardware-in-the-loop simulations (HIL) in order to compare
results and validate the tests, which are presented in this paper.

current transformer saturation, overcurrent relay, hardware-in-the-loop, real time digital simulator

Citation: Coelho A L M, Silveira P M. CT saturation effects on performance of digital overcurrent relays.

1 Introduction An important document is the guide IEEE Standard

C37.110-2007 [3] that contains some rules for selection of
To the extent that electrical power system expand, present- CTs to avoid AC and DC saturation. Many of these recom-
ing high levels of fault currents, greater attention is neces- mendations result in increasing the CT’s core size. On the
sary with the protection devices to ensure its proper opera- other hand, in designs where it is desirable to incorporate CTs
tion. For this purpose, these devices require an accurate as integral part of reclosers and circuit breakers, its dimen-
measuring of the current system. Therefore, the current sions are typically the limiting factor [4].
transformers (CTs) are key components in this system, since Therefore, the choice of CTs for these applications is as-
its provide access to high current of this system through sociated with its behavior when exposed to factors that can
reduced replica on the secondary side, allowing timely and lead its core to saturate.
correct identification of faults and disturbances in the net- For this analysis, CT modeling is important to understand
work by protective relays [1]. The correct operation of the the saturation phenomenon. Many models of iron core were
protection system is therefore dependent on the perfor- studied and considered by many researchers. Most published
mance of CTs, which to operate successfully, requires a re- papers have considered the use of ATP (Alternative Transi-
production as accurate as possible of the current signals ents Program) for CT modeling under transient conditions,
during system faults. In opposition to this desirable feature, since accurate and acceptable results were obtained with
the cores of the CTs have characteristics of nonlinear exci- success [4-8].
tation and when exposed to high fault currents may retain The use of high technology equipment to study and ana-
high levels of flow density leading the core to saturate [1]. lyze the electrical protection devices is becoming more and
Particularly in electrical distribution systems, the satura- more common. Nowadays, the market offers modern test
tion of low ratio CTs due to high fault current may cause platform such as: injection test equipment that generate
misoperation of the overcurrent relays. Because this prob- signals when programmed to simulate situations that happen
lem, may occur loss of production to various plants or in a system and Real Time Digital Simulators (RTDS). These
damages critical equipment connected to this system [2]. simulators have the ability to perform simulations expressing
data both on the computer screen and in analogical and dig-
Coelho, A L M with Power Quality Study Group, UNIFEI, Itajubá, Brazil. ital outputs. These two outputs modes allowing perform tests
(email: [Link]@[Link])
Silveira, P M with Power Quality Study Group, UNIFEI, Itajubá, Brazil. APAP2011 [Link]

(e-mail: pmsilveira@[Link])
 APAP2011
on real devices, further increasing the credibility of the tests The parameters of circuit represented on Figure 1 are:
[9].
This paper will investigate the transient behavior of CT’s ip Instantaneous primary current;
and the factors that can lead its core to saturate. Besides the ips Instantaneous ideal secondary current;
performance of a digital overcurrent relay under CT satura- is Instantaneous real secondary current;
tion is evaluated in two ways. Firstly, a mathematical model ie Instantaneous excitation current;
of a CT, based on the theory published by the IEEE PSRC ve Instantaneous excitation voltage;
[10], has been implemented in MATLAB. By using this Number of CT turns;
model, secondary current waveforms are generated giving Rs Winding secondary resistance;
rise to a database with different signals. Any of these signals Rb Burden resistance;
can be converted to COMTRADE format and used in an Lb Burden inductance.
injection test equipment. In this case, the current signals is  radian frequency = 2f; f=60Hz.
transformed in a analogical signal, amplified and injected in
a overcurrent relay. The use of a RTDS (Real Time Digital Through of Kirchhoff's Voltage Law is possible to solve
Simulator) is the second way. For this, the overcurrent relay the circuit represented in Figure 1 by
is evaluated perfoming a hardware-in-the-loop simulations
(HIL) in order to compare results of the two methods. Elab- d
ve − (ips − ie ). (R s + R b ) − Lb . (ips − ie ) = 0 (1)
orated CTs electrical models are used in the RTDS and the dt

results are compared and validated in both methodologies.

B CT excitation curve
2 Current Transformer Modeling
The excitation characteristic of the modeled CT is repre-
sented by rms values of the excitation current ( Ie ) versus
Based on [10], the CT mathematical model implemented in
excitation secondary voltage ( Ve ), as illustrated in Figure 2.
MATLAB is intended to provide a quick indication not only
In this model, only two parameters need to be extracted
of the existence of CT saturation in a specific application, but
from the excitation curve: the slope (S) of the saturated sec-
also an accurate indication of the actual waveform of the
tion and the saturation voltage ( Vs ) for which the excitation
secondary current, in such way that the degree of saturation
current is ten amps [10]. The reason for choosing the satura-
can be visualized as function of time [10].
tion voltage at the point where the exciting current is ten
In this model, the conditions at the low-end portion of the
amps, is due to the rules defined in IEEE C37.110-2007 [3].
saturation curve are not represented. The excitation current
in the region below the knee-point is a complex combina-
tion of magnetizing, hysteresis and eddy current compo-
nents. These parameters usually are not known in a particu-
lar case. If the excitation current reaches into the saturated
region, the part of the waveform current in the be-
low-knee-point region has negligible effect on the global
solution [10]. Consequently, the hysteresis and eddy currents
losses are not represented in this model.

Figure 2 Excitation curve of the model.

A CT circuit model

The equivalent circuit for this model is represented in Fig- The flux-linkages (λ) versus excitation current (ie ) curve
ure 1. The parameters of circuit represented on Figure 1 are: has been used to solve the differential equations implied by
equation (1).
The remanent flux is defined in per unit, as a ratio be-
tween a voltage below the knee point ( Vx ) and the saturation
voltage ( Vs ), as shown in Figure 2.

C CT transient response

For this model the CT transient secondary response is de-

Figure 1 Equivalent circuit of the model. pendent of the CT rated values entered by the user in the
algorithm implemented: burden values, the saturation volt-
 APAP2011
age, number of turns, secondary winding resistance and A System test based on injection test equipment
remanent flux values; and also by the disturbance charac-
teristics such as: level of fault current, the presence of dc The values of the transient secondary waveforms obtained
component and the system X/R ratio. from equations (2) and (4) were converted to COMTRADE
Filled these parameters, the ideal (𝑖𝑒 =0) transient re- format [11] using an algorithm implemented in MATLAB.
sponse of the secondary current in function of time in re- These values were loaded into a OMICRON test equipment.
sponse of a short-circuit is given by The OMICRON’s Advanced Transplay software is used
t
to import the COMTRADE data generated from CT mod-
Ip −
ips (t) = √2 [K . e τ − cos(ωt − cos −1 K) (2) eled in MATLAB. For the intended application, a protective
N
relay, SEL 351-S, manufactured by Schweitzer Engineering
where Ip is the rms symmetrical primary fault current and Laboratories, is used to evaluate the overcurrent function
K determines the features of the current waveform: under CT saturation. The trip signal of this relay is sent to
the OMICRON binary/analog inputs in order to register the
time of trip command. The test system is represented in
i) K = 1, when there is asymmetrical component in fault
1 Figure 3.
current; In this case τ = is the constant time of the
X/R
system;
ii) K = 0, when there is only symmetrical component in fault
current;

In practice, the excitation current is nonzero and its value

determines how much the primary current will be reflected
erroneously in the secondary. In this model this current is
given by Figure 3 Test system based on OMICRON test equipment.

ie (t) = A. |λ(t)|S (3)

B Hardware-in-the-loop simulation by RTDS
where A is a model constant defined in [10].
The same features of the CT modeled in algorithm imple-
Considering the excitation current is nonzero, the actual mented in MATLAB has been reproduced in a CT model of
secondary current is then the RSCAD software [9], wich is used to model and run a
case. The influence variables on the fault current defined by
is (t) = ips (t) − ie (t) (4) user in MATLAB were determined on the RSCAD. For this,
arrangement and logical of electrical and control compo-
To evaluate the equation (4) as a saturation phenomenon, nents in a distribution system were made to provide similar
the IEEE C37.110-2007 criteria [3] has been adopted. In waveforms in order to check and compare various situations
according to this standard, AC and DC saturation can be that can lead the CT to saturate. The response of the SEL
avoided by following equations (5) and (6), respectively. 351-S relay under these events was performed.
The RSCAD is a powerful tool for modeling and simu-
20 ≥ if . zb (5)
lation power systems in real-time for event analysis [9]. By
X using the RTDS is not necessary to use injection test
20 ≥ |1 + | if . zb (6)
R equipment, being the behavior relay sizing by hard-
ware-in-the-loop simulation (HIL). In HIL simulation, some
where if is the maximum fault current in per unit of CT
of the components of the virtual power system can be re-
rating and zb is the CT burden in per unit of standard bur-
placed by physical devices [12]. In this case, the device is the
den.
SEL relay.
Figure 4 shows the HIL simulation scheme used for these
3 Relay test platform tests. The current secondary signals are sent to D/A con-
verters on the RTDS. To extract these signals for the relay,
Here are described briefly the two test systems used in this the GTAO card (Analogue Output Card) was used. In this
study to evaluate the effects of CT saturation in the over- card the secondary current waveform is reproduced in pro-
current response of a digital commercial relay. portional voltage values in the range of -5 to +5 Vac.
Therefore, this card supports low-level signals. Then an
 APAP2011
amplifier was used because high-level signal was necessary tion, the difference between the response relay for an ideal
to saturate the CT core in this research. The analog signals and actual secondary current for a particular case has been
are then connected to the relay’s inputs. The relay senses considered. In general, three cases were selected to be used
the secondary currents from the RTDS system, and in case in this analysis as shown in Table 1.
of any fault, it sends out the trip signal to the simulated cir- By IEEE criteria [3] defined in equation (5), in all the
cuit breakers in the power system modeled in RSCAD. Trip cases there are CT saturation. In spite of fault magnitude be
signals from the relay are interfaced to the RTDS via digital higher in case 3, the second case has a factor burden that
input ports. In this research was used the GTFPI card (Front provides a severe CT saturation.
Panel Interface Card). Figure 5 compares ideal and actual CT secondary current
signals for the two methods for case 2.
As a result one can observe that the actual secondary
current magnitude is low and its waveform is severely dis-
torted causing the non-operation of the instantaneous func-
tion for these conditions in both methods.
For function 51, the relay operated, but with a delay of
almost 20 cycles, as shown in Table 1.

Table 1 Overcurrent relay response considering a symmetrical fault and

multiple of standard burden on CT saturation.
Relay operation time delay (in cycles)
Matlab Model RTDS Model
Figure 4 HIL simulation scheme by RTDS. Cases MI* MB*
Relay Function Relay Function
50 51 50 51
Case 1 7.5 5 0.948 1.5 0.517 3
4 Results and discussion Case 2 7.5 10 - 18.48 - 19.28
Case 3 20 2 0.43 0.96 0.276 0.98
To investigate the CT saturation phenomenon for the two
* MI and MB are the multiple of CT rated current and standard burden
methods, the following influence factors were analyzed: the respectively.
short circuit level, the presence of dc component and the
system X/R ratio, beyond the burden and remanent flux.
The instantaneous (50) and time-overcurrent (51) func-
tions of the SEL 351-S relay are evaluated under these
events.
For analyze in RTDS, a distribution system has been
modeled in RSCAD. The rated current of this system is ap-
proximately 250A. So it was used a CT 300/5A with the
following characteristics in according to the IEEE
C57.13-2008 [13]:
- Secondary terminal voltage: 100V.
- Standard burden: B1.0 (R b = 0.5 Ω; Lb = 2.3 mH).

A Effects of Symmetrical Fault Current Figure 5 CT saturation due to a high burden value – Case 2 of Table 1 for
the two CT models.
For this analysis will be shown some cases used to evaluate
the effects of the symmetrical fault current and CT burden
B Effects of Asymmetrical Fault Current
on the saturation phenomenon in order to check the re-
sponse of the instantaneous and time-overcurrent functions
of the SEL 351-S relay. In this evaluation, some cases have been used to check the
The following settings have been used for this relay: (i) effects of the asymmetrical fault current on the CT satura-
pickup of instantaneous function 18A, (ii) IEC Very Inverse tion. In all cases shown in Table 2 were used a standard
Curve, (iii) a pickup of inverse-time 5A and (iv) dial in a burden and has been adopted a fault current of 20 times the
value of 0.05. rated CT.
To verify the relay time operation delay under CT satura- The following settings have been used to evaluate the
overcurrent relay response: (i) pickup of instantaneous
 APAP2011
function 60A, (ii) IEC Very Inverse Curve, (iii) a pickup of Has been adopted a fault current of 7.5 times the CT
inverse-time 5A and (iv) dial in a value of 0.05. rated current and a X/R ratio of 25 in all cases on the Table
Different values of X/R ratio have been used to evaluate 3.
the relay response as shown in Table 2. The SEL 351-S relay has been adjusted with the follow-
ing settings: (i) pickup of instantaneous function 60A, (ii)
IEC Very Inverse Curve, (iii) a pickup of inverse-time 5A
Table 2 Overcurrent relay response considering an asymmetrical fault on
CT saturation.
and (iv) dial in a value of 0.05.

Relay operation time delay (in cycles)

Matlab Model RTDS Model Table 3 Overcurrent relay response considering an asymmetrical fault,
Cases MI* X/R
Relay Function Relay Function remanent flux and a high burden value.
50 51 50 51
Relay operation time delay (in cycles)
Case 1 20 10 4.25 2.94 3.01 1.65 λrem* Matlab Model RTDS Model
Cases MB*
Case 2 20 16 5.32 4.72 4.28 3.89 (p.u.) Relay Function Relay Function
Case 3 20 25 10.46 5.84 7.67 4.97 50 51 50 51
* MI is the multiple of CT rated current. Case 1 5 0.4 8.09 11.02 6.02 9.05
Case 2 7 0.8 - 27.10 - 21.72
By IEEE criteria [3] defined in equation (6), all the cases Case 3 6 0.9 9.17 17.59 8.04 13.43
* MI is the multiple of standard burden and λrem is the remanent flux in
of the Table 2 there are CT saturation. A sample of simulated
p.u.
secondary current is shown in Figure 6 for the worst case of
CT saturation presents in this table, the case 3. In this case, a
By IEEE criteria [3] defined in equation (6), in all the
high fault current is injected to the primary with X/R ratio of
cases of the Table 3 there are CT saturation. A sample of
25. The CT is driven into severe saturation within the first
simulated secondary current for the two methods is shown in
cycle and the secondary current remains low and distorted
Figure 7 for the worst case of CT saturation presents in this
for about six cycles.
table, the case 2. In this case, due high burden value, the
The CT secondary waveform response for the two
actual secondary current magnitude is low and its waveform
methods are almost identical as represented on Figure 6.
is very distorted causing the non-operation of the instanta-
Table 2 presents the relay operation time delay for a SEL
overcurrent digital relay. This table shows that, given the neous function for these conditions in both methods. Ob-
same level of fault current, a higher value of X/R ratio result served in Figure 7 that CT is driven into severe saturation
in a lower value seen by the relay in the first cycles and within the first cycles due the remanent flux and high X/R
consequently greater delay in its operation in comparison to ratio value.
the ideal case. For function 51, the relay operated in all cases of Table 3,
but with a greater delay in comparison to the ideal case, as
show n in a curve represented by Figure 8.

Figure 6 CT saturation due to an asymmetrical fault current - Case 3 of

Table 2 for the two CT models

C Effects of the remanent flux

For this analysis has been considered an asymmetrical fault Figure 7 CT saturation due to a high burden and remanent flux value –
current, a multiple of standard burden and remanent flux Case 2 of Table 3 for the two CT models
values on the CT saturation to evaluate the overcurrent relay
response as shown in Table 3.
 APAP2011
This work was supported by the CAPES and FAPEMIG, Brazil.

1 Segatto, E.C.; Coury, D.V. A power transformer protection with

recurrent ANN saturation correction. Power Engineering Society
General Meeting, 2005. IEEE , vol., no., pp. 1341- 1346 Vol. 2,
12-16 June 2005.

2 El-Amin, I.M.; Al-Abbas, N.H. Saturation of Current Transform-

ers and its Impact on Digital Overcurrent Relays. Transmission &
Distribution Conference and Exposition: Latin America, 2006.
TDC '06. IEEE/PES , vol., no., pp.1-6, 15-18 Aug. 2006.

3 IEEE C37.110-2007. IEEE Guide for the Application of Current

Transformers Used for Protective Relaying Purposes, 1996.

4 Kojovic, LJ.A. Impact of Current Transformers Saturation on

Overcurrent Protection Operation. Power Engineering Society
Summer Meeting, 2002 IEEE, Volume: 3, pp. 1078-1083, July
2002.

Figure 8 Time-overcurrent relay response under CT saturation – Cases of 5 Kezunovic, M.; Fromen, C. W. and Phillips, F., Experimental
Table 3. Evaluation of EMTP Based Current Transformer Models for Pro-
tective Relay Transient Study, IEEE Transactions on Industry Ap-
plications, Vol 9, No.1, pp. 405-413, 1994.
5 Conclusion
6 Guerra, F. C. F.; Mota, W. S. Current transformer model, IEEE
Transactions on Power Delivery, vol. 22, no. 1, pp. 187- 194,
An efficient and accurate CT model was implemented in 2007.
MATLAB based on theory published by IEEE PSRC to
evaluate the CT transient response and its effects on over- 7 Lucas, J. R.; McLaren, P. G.; Keethipala, W. W. and Jayasinghe, R.
current instantaneous and time-overcurrent functions re- P. Improved Simulation Models for Current and Voltage Trans-
formers in Relay Studies, IEEE Transactions on Power Delivery,
sponse of a commercial relay. This model showed a satis- vol. 7, no. 1, pp. 152-159, 1992.
factory response when compared with the transient behavior
of RTDS CT model configured with the same parameters, 8 Kojovic, LJ.A. Comparison of different current transformer mod-
presented similar CT signals response. The small difference eling techniques for protection system studies Power Engineering
Society Summer Meeting. IEEE. v. 3, p.1084-1089, 2002.
observed in the relay time operation delay in both methods
is acceptable because the mathematical CT models are not
9 RTDS Technologies, [Link]
equals. However in two methods was noted the influence of
saturation in the overcurrent relay response. 10 Swift, G. CT Saturation Theory and Calculator. IEEE PSRC, pp
Considering only low levels of symmetrical fault current 1-9, June 2001.
and standard burden values the relay operation time delay
was very short and did not cause problem in the relay oper- 11 IEEE C37.111. IEEE Standard Common Format for Transient
Data Exchange (COMTRADE) for Powers Systems, 1999.
ation under CT saturation.
However, the overcurrent relay response was severely 12 Chenfeng Zhang, Vamsi K Vijapurapu, Anurag K Srivastava, Noel
affected by CT saturation when its circuit as exposed to a N. Schulz, Jimena Bastos and Rudi Wierckx. Hard-
high levels of fault current, presence of dc component in a ware-in-the-Loop Simulation of Distance Relay Using RTDS.
Summer Simulation Multiconference, San Diego, CA United
fault current and a high value of ratio X/R, multiples of States, 2007.
standard burden and remanent flux in the core.
This paper shows that CT performance for a given appli- 13 IEEE C57.13. IEEE Standard Requirements for Instrument
cation depends on system conditions and be in accord with Transformers, 2008.
the criteria defined by IEEE standard C37.110-2007 to
evaluate saturation effects so check the protective relays
purposes.
The HIL simulation by RTDS proved to be a fast and ef-
fective method for testing in relays, and also allows greater
repeatability of the tests and a real-time analysis of wave-
forms distorted by CT saturation.