See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228897262

Identification of parameters for coupling capacitor voltage transformers

Article · January 2001

CITATIONS READS

5 1,317

3 authors, including:

W.L.A. Neves
Universidade Federal de Campina Grande (UFCG)
53 PUBLICATIONS   837 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Transmission line controlled switching View project

EMTP-Based Numerical Relays Models View project

All content following this page was uploaded by W.L.A. Neves on 09 September 2014.

The user has requested enhancement of the downloaded file.

 Identification of Parameters for
Coupling Capacitor Voltage Transformers

D. Fernandes Jr. W. L. A. Neves J. C. A. Vasconcelos

Departamento de Engenharia Elétrica – UFPB Companhia Hidro Elétrica do São Francisco
Av. Aprígio Veloso, 882 – Bodocongó Av. General Sam Martins, 1450 – Bongi
58.109-970, Campina Grande – PB – Brazil 50.630-060, Recife – PE – Brazil
[email protected], [email protected] [email protected]

Abstract - A method to obtain the coupling capacitor frequency response from the secondary side without the
voltage transformer (CCVT) model parameters from need to access its internal components.
frequency response curves is presented. Frequency In Brazil, some electrical energy companies have
response measurements of magnitude and phase, in the reported unexpected overvoltage protective device
range from 10 Hz to 10 kHz, were carried out for a operations in several coupling capacitor voltage
230 kV CCVT at our high voltage laboratory and used transformers leading to failures of some units. The
as input data to a full Newton-type fitting routine to reported overvoltages occurred during normal switching
estimate the CCVT parameters. Analytical CCVT conditions [8, 9].
functions were fitted to the measured data. The In the present study, a first step consisting on building
magnitude and phase errors, in the whole frequency a CCVT model for transient studies, is given towards the
range, are fairly small. The nonlinear behavior of the solution of some CCVT problems reported by CHESF
voltage transformer (VT) magnetic core is also taken (Companhia Hidro Elétrica do São Francisco). The model
into account. The obtained CCVT model may easily be includes the nonlinear behavior of the VT magnetic core.
used in connection with the EMTP (Electromagnetic The goal is to obtain a method to estimate the model
Transients Program). parameters (resistances, inductances and capacitances)
from the frequency response curves. In order to achieve
Keywords: Coupling Capacitor Voltage Transformer, this, frequency response measurements of magnitude and
Nonlinear Fitting, CCVT Model, EMTP. phase, in the range from 10 Hz to 10 kHz, were carried
out for a 230 kV CCVT in the high voltage laboratory and
used as input data to a full Newton-type fitting routine.
I. INTRODUCTION The analytical CCVT functions (magnitude and phase)
were fitted to the measured data.
Electric utilities, for many years, have used coupling
capacitor voltage transformers (CCVTs) as input sources
to protective relays and measuring instruments. However, II. FUNDAMENTAL PRINCIPLES
problems have yet been traced to incorrect inputs.
The fundamental principle of CCVT is the fidelity The basic electrical diagram for a typical CCVT is
with which the secondary voltage follows the primary shown in Fig. 1. The primary side consists of two
voltage under all operating conditions. Under steady-state capacitive elements C1 e C2 connected in series. The
conditions, this requirement may be achieved based on the voltage transformer provides a secondary voltage vo for
design and tuning of the CCVT. However, the fidelity of protective relays and measuring instruments. The
CCVT decreases under transient conditions due to inductance Lc is chosen to avoid phase shifts between vi
inductive, capacitive and nonlinear components [1]. and vo at power frequency. Small errors occur due to the
Therefore, the CCVT transient behavior must be well exciting current and the CCVT burden (Zb) [1, 2].
known. LINE
Many works including field measurements, laboratory
tests and digital simulations, have been conducted to study
the performance of the CCVT. Some studies have been
C1
concentrated on nonlinear behavior of the voltage
transformer (VT) magnetic core to accurately simulate the
transient response of CCVT [2, 3, 4]. vi Lc VT
Other works have considered the effect of stray
capacitances in some CCVT elements to explain the C2 FSC Zb vo
measured frequency responses in the linear region of
operation [5, 6, 7]. There are some problems in obtaining
the CCVT parameters. In [5] and [7], the used
measurement techniques need disassembling the CCVT
and in [6], a method was developed to measure the CCVT Fig. 1. Basic electrical diagram for a typical CCVT.
 Ferroresonance oscillations may take place if the magnetic core because in the frequency response
circuit capacitances resonate with the iron core nonlinear measurements, the core was not saturated. The
inductance. These oscillations cause undesired nonlinearity will be only included in time domain
information transferred to the relays and measuring simulations to improve the representation of the transient
instruments. Therefore, a ferroresonant suppression circuit effects in CCVT. The circuit shown in Fig. 2 is considered
(FSC) is normally included in one of the CCVT windings. with specific blocks of impedances Z1, Z2, Z3, Z4, Z5 and
Circuits tuned at power frequency (L in parallel with with all the elements referred to the voltage transformer
C) and a resistance to ground have been often used as secondary side, according to Fig. 4.
ferroresonant suppression circuits [4, 6] because they LINE
damp out transient oscillations and they require small
amount of energy during steady-state. r2C1
1 Z1 2 Z2 3

III. DEVELOPED ANALYTICAL METHOD vi/r Z4

2
r C2 r2Cp Z3 4 Zb vo
The diagram shown in Fig. 1 is valid only near power Z5
frequency. A model to be applicable for frequencies up to
a few kilohertz needs to take the VT primary winding and
compensating inductor stray capacitances into account [3, Fig. 4. CCVT model with specific blocks of impedances.
4, 5, 6].
In this work, the circuit shown in Fig. 2 was used to The expressions for the impedances in the s domain,
with s = jω, are:
[ ](
represent the CCVT. It comprises: a capacitor stack (C1,
C2); a compensating inductor (Rc, Lc, Cc); a step down Z 1 = (R c + sLc ) r 2 // 1 r 2 sC c ; )
transformer (Rp , Lp , Cp , Lm , Rm) and a ferroresonant
suppression circuit (Rf , Lf1 , Lf2 , −M , Cf) [5, 6, 10].
(
Z 2 = R p + sL p )r 2
;
The FSC design is shown in Fig. 3a. A nonsaturable (
Z 3 = Rm r 2
)// (sL m )
r2 ; (1)
iron core inductor Lf is connected in parallel with a
capacitor Cf so that the circuit is tuned to the fundamental
( ) (
Z 4 = sL f 1 + 1 sC f // sL f 2 ; )
frequency with a high Q factor [10]. The configuration of Z 5 = R f − sM .
FSC digital model is shown in Fig 3b. The damping Where, r is the VT ratio and the symbol // means that
resistor Rf is used to attenuate ferroresonant oscillations. elements are in parallel.
The CCVT model parameters R, L, C should
LINE
reproduce the transfer functions of magnitude and phase
C1 Cc
represented by vo/vi . They are calculated using the
technique described below for the minimization of
Lc Rc Lp Rp
nonlinear functions.
Cf Lf2
C2 Cp Lm B. Minimization of Nonlinear Functions
Rm Lf1 Zb
−M
Rf Methods for minimizing nonlinear functions are
usually iterative, that is, given an approximate solution xi ,
an estimate of the solution x* is obtained. The technique
used here is based on Newton’s method which uses a
Fig. 2. CCVT model for identification of parameters.
quadratic approximation to the function F(x) derived from
the second-order Taylor series expansion about the point
xi. In two dimensions, the second-order Taylor series
Cf Lf Cf
Lf1 Lf2 approximation can be written in the form:
 ∂F ( x1 , x 2 ) 
−M  ∂x1 
Rf F ( x1 + p1 , x 2 + p 2 ) ≈ F ( x1 , x 2 ) + [ p1 p 2 ] 
Rf  ∂F ( x1 , x 2 ) 
 ∂x 2 
 
(a) (b) (2)
 ∂ F ( x1 , x 2 ) ∂ F ( x1 , x 2 ) 
2 2
 
Fig. 3. (a) FSC design. (b) FSC digital model.  ∂x12 ∂x1 ∂x 2   p1 
+ [ p1 p 2 ]
1
 p 
 ∂ F ( x1 , x 2 ) ∂ F ( x1 , x 2 )   2 
2 2 2
A. Mathematical Model Development
 ∂x 2 ∂x1 ∂x 22 
 
In order to develop the analytical expressions of CCVT And for n dimension, the expression above in
model, we considered only the linear region of the VT matrix/vector form is:
 1 T 2 H1 X1
F ( x + p ) ≈ F ( x ) + p T ∇F ( x ) +
p ∇ F ( x ) p . (3)
2 X2
To obtain the step p, the function F is minimized by
forming its gradient with respect to p and setting it equal X3
to zero. Then, Y1
∇ F(x)p = −∇F(x).
2
(4) 2
The approximate solution xk+1 is given by: Y2
[ ]
−1
xk +1 = xk + p = x k − ∇ F ( x k ) ∇F ( x k ) .
2
(5)
H2
1
Y3
Newton’s method will converge if [∇ F(x)] is 2 -1
Fig. 5. 230 kV CCVT voltage transformer.
positive definite in each iterative step, that is,
zT[∇2F(x)]-1z > 0 for all z ≠ 0. The technique used here is In this configuration, H1 is the primary terminal, H2 is
the full Newton-type method [11,12]. It is a modification the ground terminal and X1, X2, X3, Y1, Y2, Y3 are the
of Newton’s method for that iteration in which [∇2F(x)]-1 low voltage secondary terminals. Terminals 1 and 2 can
is not positive definite: ∇2F(x) is replaced by a “nearby” be connected to ground.
positive definite matrix ∇ 2 F ( x ) and p is computed by
B. Frequency Response Measurements
solving ∇ 2 F ( x ) p = −∇F ( x ) .
Our objective is to minimize the merit function χ2(x) Frequency response measurements, magnitude and
2 phase, were carried out for the 230 kV CCVT. A low-pass
n
 yi − y (ωi ; x ) 
χ 2 ( x) = ∑ 
i =1 
σi
 ,

(6) filter was required to attenuate high frequency noises. A
3rd order RC active filter with a cut-off frequency of 15
where ωi is the i-th measured frequency value and yi is the kHz, was used [10]. A signal generator feeding an
i-th measured frequency response value of the n data amplifier whose maximum peak-to-peak voltage is
points. σi is the standard deviation for each yi . x is the 2000 V, was connected between the high voltage terminal
vector which contains the parameters R, L, C to be and the ground, according to Fig. 6.
determined and y(ωi ; x) is the analytical model function. Signal
A FORTRAN code was written to minimize the merit Generator
function χ2(x) using the method described above. Besides ∼ C1
the function y(ωi ; x) it is necessary to know its first and vi
Y1
second derivatives with respect to each parameter of the Power Inductive
Amplifier vo Filter
vector x. The algorithm is given below: C2 Tank
1. Supply the CCVT frequency response values yi for Y3
each frequency ωi and enter with a guess for the
parameters R, L, C (vector x);
2. Determine χ2(x) and evaluate χ2(x + p); Oscilloscope
3. Save the value of χ2(x + p) and for a user defined
number of iterations m, compare the actual value of the Fig. 6. Frequency response measurements for the
merit function to its old value m iterations before: 230 kV CCVT.
4. If the difference is greater than a user defined
tolerance, go back to step 2. Otherwise, stop the C. VT Nonlinear Characteristic Measurements
iterative process.
A sinusoidal voltage source was applied across
terminals X1 − X3 up to 2.2 p.u. and rms voltage and
IV. LABORATORY TESTS current data points were obtained. The rms v − i data were
converted into the peak λ − i data shown in Table 1 by
Frequency response measurements of magnitude and
using a routine developed in [13].
phase were carried out for CCVT. The rms v − i nonlinear
Table 1. Nonlinear characteristic of the VT magnetic core.
curve for the VT magnetic core was measured as well.
A. 230 kV CCVT Manufacturer Data [10]: Peak Current (A) Peak Flux (V.s)
0. 076368 0.025772
• Primary voltage: 230: 3 kV;
0.720881 0.189066
• Intermediate voltage: 30: 3 kV; 1.429369 0.396889
• Secondary voltage: 115 V and 115: 3 V; 2.147414 0.638099
• CCVT ratio: 2000 − 1154.7 : 1; 2.511675 0.748388
• Capacitances: C1 = 9660 pF and C2 = 64400 pF; 2.989304 0.806533
• Frequency: 60 Hz. 3.662012 0.863553
The configuration of the CCVT intermediate voltage 4.587227 0.903317
transformer is shown in Fig. 5. 5.712037 0.942706
 V. ANALYSIS OF THE RESULTS 20.0
Analytical model curve
Fitted curve
The nonlinear fitting routine was used for two CCVTs. 0.0
To check if the code was correct, 138 kV CCVT data
reported in literature [5, 6] were used as a benchmark. -20.0

Gain (dB)
Measurements were performed for a 230 kV CCVT at our
high voltage laboratory. -40.0

A. 138 kV CCVT Benchmark Case -60.0

From the 138 kV CCVT parameters of Table 2 [6], the -80.0

frequency response magnitude and phase were computed
using MICROTRAN [14]. The magnitude and phase -100.0
data were used as an input file to the full Newton-type 1.0E+1 1.0E+2 1.0E+3 1.0E+4
fitting routine and the parameters were recalculated back. Frequency (Hz)
One set of the initial guesses is shown in Table 3. The Fig. 7. 138 kV CCVT magnitude curves obtained from the
recalculated parameters are shown in Table 4. original parameters and the nonlinear method.
Figs. 7 and 8 show the analytical and fitted curves for
100.0 Analytical model curve
magnitude and phase, respectively. The curves are nearly Fitted curve
the same. The fitting errors are shown in Table 5. 80.0
The routine converges even for sets of initial guesses 60.0
far away from the final values. Although differences may 40.0
occur for some recalculated circuit parameters, for each
Phase (degree)

20.0
set of initial guesses the obtained analytical frequency
response curves are nearly the same [10]. 0.0
The gain is calculated as: -20.0
 41.7v o 
Gain = 20 log   , (7) -40.0
 vi  -60.0
where 41.7 is the VT ratio. The voltages vo and vi were -80.0
measured at several frequencies.
-100.0
The CCVT phase is the phase shift between applied
1.0E+1 1.0E+2 1.0E+3 1.0E+4
signal and secondary signal as follows: Frequency (Hz)
phase = ∠v o − ∠v i . (8)
Fig. 8. 138 kV CCVT curves of phase obtained from the
original parameters and the nonlinear method.
Table 2. 138 kV CCVT parameters [6]
Table 5. Magnitude and phase errors for 138 kV CCVT.
Rc = 228.0 Ω Lp = 2.85 H Lf2 = 247.0 mH
Lc = 56.5 H Rm = 1.0 MΩ Rf = 37.5 Ω Magnitude Error Phase Error
Cc = 127.0 pF Lm = 10.0 kH M = 163.0 mH Average Maximum Average Maximum
Cp = 154.0 pF Lf1 = 481.0 mH − 0.026 % 0.13 % 0.39o 0.86o
Rp = 400.0 Ω Cf = 9.6 µF − For the initial guesses shown in Table 3, the routine
converged in 6 iterations. This is one of the most
Table 3. 138 kV CCVT initial guess parameters
important characteristics of the full Newton-type method:
Rc = 174.0 Ω Lp = 7.0 H Lf2 = 270.0 mH fast decrease of the merit function in the initial iterations
[10, 12]. In other words, χ2(x) reached quickly a value
Lc = 17.0 H Rm = 34.0 MΩ Rf = 40.0 Ω
nearly 300,000 times lower than its initial value.
Cc = 58.0 pF Lm = 10.0 kH M = 193.0 mH From these results, the CCVT magnitude and phase
Cp = 5.0 pF Lf1 = 650.0 mH − can be reproduced by the fitted curves with very small
Rp = 3.0 kΩ Cf = 7.0 µF − errors.

Table 4. 138 kV CCVT recalculated parameters B. 230 kV CCVT Parameters from Measurements

Rc = 3.25 Ω Lp = 2.88 H Lf2 = 274.34 mH The 230 kV CCVT parameters were estimated from
Lc = 56.55 H Rm = 1.47 MΩ Rf = 36.80 Ω frequency response data points of magnitude and phase
measured in laboratory. The 230 kV CCVT initial guesses
Cc = 126.86 pF Lm = 10.98 kH M = 192.7 mH
and the fitted parameters are shown in Table 6 and 7,
Cp = 151.44 pF Lf1 = 649.78 mH − respectively. The measured and fitted magnitude and
Rp = 833.55 Ω Cf = 7.53 µF − phase curves are shown in Figs. 9 and 10, respectively.
 Table 6. 230 kV CCVT initial guess parameters The CCVT phase was obtained from the same
expression given in (8). The fitting errors for the
Rc = 16.0 kΩ Lp = 91.0 H Lf2 = 95.0 mH magnitude and phase are shown in Table 8.
Lc = 50.0 H Rm = 2.0 MΩ Rf = 4.0 Ω
Cc = 46.0 nF Lm = 7.0 MH M = 10.0 mH Table 8. Magnitude and phase errors for 230 kV CCVT.
Cp = 60.0 pF Lf1 = 10.0 mH − Magnitude Error Phase Error
Rp = 62.0 kΩ Cf = 140.0 µF − Average Maximum Average Maximum
5.1 % 16.09 % 5.53o 20.3o
Table 7. 230 kV CCVT estimated parameters.
Based on Figs. 9 and 10 and Table 8, magnitude errors
230 kV CCVT estimated parameters are fairly small for frequencies up to 2 kHz. Near 60 Hz
Rc = 6.71 kΩ Lp = 114.7 H Lf2 = 43.03 mH the magnitude and phase errors are very small. This is the
Lc = 86.16 H Rm = 27.94 MΩ Rf = 4.63 Ω region in which the CCVT operates most of the time.
Cc = 584.6 nF Lm = 16.3 kH M = 9.19 mH
C. 230 kV CCVT Transient Simulations
Cp = 0.23 nF Lf1 = 11.06 mH −
Rp = 10.5 kΩ Cf = 173.15 µF − Time domain simulations were performed with
-10.0
MICROTRAN to evaluate the importance of the FSC.
Laboratory measurement The CCVT is connected to the power system by a switch.
Fitted curve An open-close switch operation was simulated. Fig. 11
-15.0
shows the CCVT voltage waveform at the CCVT
secondary side when the FSC is removed. Fig. 12 shows
-20.0
the same case with the FSC included in the model. The
Gain (dB)

switch opened at 12.5 ms and closed at 50 ms. It can be

-25.0 seen that the transient was damped out if the FSC is in.
400.0
-30.0
300.0
-35.0 200.0
Secondary Voltage (V)

-40.0 100.0
1.0E+1 1.0E+2 1.0E+3 1.0E+4 0.0
Frequency (Hz)
Fig. 9. 230 kV CCVT curves of magnitude measured and -100.0
fitted by nonlinear method. -200.0
80.0
60.0 Laboratory measurement -300.0
40.0 Fitted curve
20.0 -400.0
0.0 0.00 0.05 0.10 0.15 0.20 0.25 0.30
Time (s)
Phase (degree)

-20.0
-40.0 Fig. 11. Secondary voltage waveform of the 230 kV
-60.0 CCVT without the FSC.
200.0
-80.0
-100.0
-120.0 100.0
Secondary Voltage (V)

-140.0
-160.0
-180.0 0.0

1.0E+1 1.0E+2 1.0E+3 1.0E+4

Frequency (Hz)
-100.0
Fig. 10. 230 kV CCVT curves of phase measured and
fitted by nonlinear method. -200.0

The 230 kV CCVT gain is given by:

  152.6v o -300.0
Gain = 20 log 
,
 (9) 0.00 0.05 0.10 0.15 0.20 0.25 0.30
  vi Time (s)
where 152.6 is the VT ratio measured in laboratory, Fig. 12. Secondary voltage waveform of the 230 kV
considering the secondary terminals Y1 − Y3. CCVT with the FSC.
 VI. CONCLUSIONS "Mathematical Models for Current, Voltage and
Coupling Capacitor Voltage Transformers", IEEE
In this work, is given a first step to solve some CCVT Transactions on Power Delivery, vol. 15, no. 1,
problems reported by CHESF. A routine has been January 2000, pp. 62-72.
implemented to estimate the CCVT model parameters [5] M. Kezunovic, Lj. Kojovic, V. Skendzic, C. W.
from the frequency response curves of magnitude and Fromen, D. R. Sevcik and S. L. Nilsson, "Digital
phase in the range from 10 Hz to 10 kHz. The analytical Models of Coupling Capacitor Voltage Transformers
CCVT functions of magnitude and phase were fitted for Protective Relay Transient Studies", IEEE
simultaneously to the measured data. Laboratory Transactions on Power Delivery, vol. 7, no. 4, October
measurements were carried out to validate the used 1992, pp. 1927-1935.
methodology. [6] Lj. Kojovic, M. Kezunovic, V. Skendzic, C. W.
The adopted CCVT model includes the nonlinear Fromen and D. R. Sevcik, "A New Method for the
characteristic of the voltage transformer and it may be CCVT Performance Analysis Using Field
used in connection with the EMTP. Measurements, Signal Processing and EMTP
Modeling", IEEE Transactions on Power Delivery,
vol. 9, no. 4, October 1994, pp. 1907-1915.
ACKNOWLEDGMENTS
[7] H. J. Vermeulen, L. R. Dann and J. Van Rooijen,
The authors are grateful to CHESF for providing the "Equivalent Circuit Modelling of a Capacitive Voltage
230 kV CCVT unit at our high voltage laboratory. The Transformer for Power System Harmonic
financial support of Mr. Damásio Fernandes Jr. from the Frequencies", IEEE Transactions on Power Delivery,
Brazilian National Research Council (CNPq) is gratefully vol. 10, no. 4, October 1995, pp. 1743-1749.
acknowledged. The authors also wish to thank the [8] H. M. Moraes and J. C. A. Vasconcelos, "Overvoltages
reviewers for their valuable suggestions. in CCVT During Switching Operations", (In
Portuguese), Proceedings of the XV Seminário
Nacional de Produção e Transmissão de Energia
VII. REFERENCES Elétrica, November 22-26, 1999, Foz do Iguaçu.
[9]J. C. A. Vasconcelos, Transfer Function Determination
[1] A. Sweetana, "Transient Response Characteristics of for Coupling Capacitor Voltage Transformer, Internal
Capacitive Potential Devices", IEEE Transactions on Report (In Portuguese), UFPB, February 2000.
Power Apparatus and Systems, vol. PAS-90, [10] D. Fernandes Jr., Estimation of Coupling Capacitor
September/October 1971, pp. 1989-2001. Voltage Transformer Parameters, M. Sc. Dissertation
[2] J. R. Lucas, P. G. McLaren, W. W. L. Keerthipala and (In Portuguese), UFPB, September 1999.
R. P. Jayasinghe, "Improved Simulation Models for [11] D. Kahaner, C. Moler and S. Nash, Numerical
Current and Voltage Transformers in Relay Studies", Methods and Software, Prentice Hall PTR, New Jersey,
IEEE Transactions on Power Delivery, vol. 7, no. 1, 1989.
January 1992, pp. 152-159. [12] W. H. Press, S. A. Teukolsky, W. T. Vetterling and
[3] M. R. Iravani, X. Wang, I. Polishchuk, J. Ribeiro and B. P. Flannery, Numerical Recipes in Fortran - The
A. Sarshar, "Digital Time-Domain Investigation of Art of Scientific Computing, Second Edition, New
Transient Behaviour of Coupling Capacitor Voltage York, Cambridge University Press, 1992.
Transformer", IEEE Transactions on Power Delivery, [13] W. L. A. Neves and H. W. Dommel, "On Modeling
vol. 13, no. 2, April 1998, pp. 622-629. Iron Core Nonlinearities", IEEE Transactions on
[4] D. A. Tziouvaras, P. McLaren, G. Alexander, D. Power Systems, vol. 8, no. 2, May 1993, pp. 417-423.
Dawson, J. Ezstergalyos, C. Fromen, M. Glinkowski, [14] Microtran Power System Analysis Corporation,
I. Hasenwinkle, M. Kezunovic, Lj. Kojovic, B. Transients Analysis Program Reference Manual,
Kotheimer, R. Kuffel, J. Nordstrom, and S. Zocholl, Vancouver, 1992.

View publication stats