You are on page 1of 5

Presentation 6.

CTs for Measurements of Current Harmonics in MV Networks

Dr. Christian Henze / Prof. Dr. Peter Schegner, Technische Universität Dresden, Germany
Matthias Reiß, Ritz Instrument Transformers, Germany

Abstract CT’s primary current is measured with three high


precision shunts in parallel. On the secondary side
The aim of this paper is to show how inductive several calibrated low inductive shunts from 0.3 
CT’s in MV networks behave regarding the to 10  are available. The voltages are recorded
measurement of current harmonics and to provide with DEWETRON DAQP-HV and DAQP-LV
an equivalent circuit model to design suitable CT’s. isolation amplifiers and the analog-digital converter
First of all, the frequency response of a physical of the NI PXI.
CT model is measured depending on different The normalized ratio nr and the phase
burdens. For this, the measurement setup and the displacement δ i are measured with frequency
obtained results are explained. Second, the theory sweeps at certain magnitudes of the primary
of the equivalent circuit diagram of a CT is
current Ip . These values are defined according to
extended for harmonic frequencies. Third, the
determination of the frequency depending equations (1) and (2).
parameters is shown. As a result, the measured w pIs
and calculated frequency responses are nr = (1)
w sIp
compared. Finally, suggestions are given to
improve the frequency response and the δ i = ϕis − ϕip (2)
measurements of current harmonics.
Therefore, the magnitude and the elements of the
frequency vector are predefined. For each
Motivation frequency a data window of about 100 periods per
Due to the increasing number of distributed channel is analyzed to obtain the amplitudes Iˆp
generation units network operators and clients and Iˆs as well as the phase angles ϕip and ϕis of
carry out more and more power quality
measurements. Several technical rules and the primary and secondary current. This is done by
standards (e.g. [1]) require the assessment of the applying the discrete Fourier transformation.
emission of harmonics of such installations. Mostly,
the frequency range up to 2.5 kHz is covered.
CT
Caused by the rising number of self-commutating
power electronics topologies harmonics of higher
orders (> 50th) are increasing continuously [2]. Rs
Therefore, a trend to measure harmonics up to ca.
9 kHz is noticeable.
The accuracy class of CT’s is valid for the rated
frequency only. Consequently, it is not possible to Rp
determine a measurement error for harmonic
measurements. This paper gives an overview
LV 2010
about the parameters influencing the frequency
Ch1 Ch2
response of CT’s. Furthermore, an algorithm for
the design of CT’s with improved frequency
Signal
depending behavior is presented. Signal Generator
Conditioning

Measurement of the Frequency GPIB interface ADC board


Response NI PXI (MATLAB)
(Control & Calculation)
The frequency response is measured with the
setup according to Fig. 1. The whole measurement Fig. 1 Measurement setup
procedure is controlled by a MATLAB program
running on a National Instruments PXI. The power Due to the construction of inductive MV-CT’s, no
amplifier LV 2010 generates the primary current of resonance points are expected up to 10 kHz. The
the CT depending on the output voltage signal of measurement results of a wide variety of different
the signal generator. The maximum output current CT’s confirm this assumption.
of the amplifier is 65 A at 10 kHz and 5.6 kVA. The

© OMICRON electronics GmbH 2011 – Instrument Transformer Measurement Forum


Presentation 6.2

The three major influence factors on the frequency 0.99


response are the burden, the core material and the
ampere turns. A physical CT model with 0.97
conventional core material and a resistive burden 0.95
yields a smaller error for higher frequencies 0.93

nr →
(Fig. 2). Even for a ten times lager burden than the
rated value the largest error accrues at rated or 0.91
smaller frequencies. 0.89
0.87
1.01 0.85
25 2000 4000 6000 8000 10000
1 f in Hz →
5
0.99
nr →

0.98 0

δ i in ° →
0.97
-5
0.96

0.95 -10
25 2000 4000 6000 8000 10000
f in Hz → -15
2 25 2000 4000 6000 8000 10000
f in Hz →
1.5
a) b)
δ i in ° →

1
Fig. 3 Frequency response of a CT Model (400:5, 1VA,
0.5 conventional core material) for a) S=25VA
cos()=1 and b) S=27VA with cos()=0,996
0

-0.5 For resistive-inductive burdens, the error and the


25 2000 4000 6000 8000 10000
f in Hz → phase displacement increase quite strongly for
higher frequencies (Fig. 3). To investigate these
0.3VA 1VA 10VA
effects, the equivalent circuit diagram will be taken
Fig. 2 Frequency response of a CT model (50:1, 1 VA, into account.
conventional core material) depending on the
burden with cos()=1
Equivalent Circuit Diagram
The equivalent circuit diagram is usually valid for
the rated frequency and linear elements only
(Fig. 4). The main problem of modeling a CT for
different frequencies in that way is to determine the
main inductance Lh and the losses in the core Rν
as a function of the frequency. The parameters of
the primary coil can be neglected. The parameters
of the secondary coil can be measured or
calculated by the manufacturer.

wpip wsis
i0 ω L '
R 'Cu
R 'b
Rν ω , Bˆ( ) uh (
ω Lh ω , Bˆ )
iν i ω L 'b

Fig. 4 Equivalent circuit diagram of a CT

© OMICRON electronics GmbH 2011 – Instrument Transformer Measurement Forum


Presentation 6.3

The induced voltage U h depends on the current of frequency only, i.e. multi frequent signals cannot
the secondary side I s and the outer and inner be investigated with this method.
burden impedance.
U ' = I ( R + R + jω ( L + L ) ) = Uˆ ' e jϕu (3)
Determination of the frequency
h s Cu b  b h
depending core parameters
§ ω ( L + Lb ) ·
ϕu = ϕis + ϕ = ϕis + tan ¨¨ ¸¸ (4) The manufactures of suitable iron cores for CT’s
© RCu + Rb ¹ provide the parameters μT and ϕFe for a certain
The amplitude of the magnetic flux density B̂ can frequency, usually 50 Hz / 60 Hz, within defined
be calculated by taking the geometry of the iron limits. Depending on the core material, these
core AFe , the frequency ω and the number of parameters vary in quite wide ranges for different
turns of the secondary coil w s into account. frequencies. Due to the rising iron losses at higher
frequencies ϕFe and μT are decreased, i.e. the
1 Uˆ 'h resistive current Iν is rising.
Bˆ = (5)
ω AFe w s
The permeability μT (6) and iron angel ϕFe (7)
ie (t )
describe the relationship between the magnetic
flux density B̂ and the magnetic field H inside the
iron. Because μT and ϕFe are necessary to design
a CT these values are known for the rated um ( t )
frequency.

(
μT = f Bˆ , ω = ) H
(6)

§ I ·
( )
ϕFe = f Bˆ , ω = tan ¨ ¸ (7) Fig. 5 Measurement of the B-H-curve for different
© Iν ¹ frequencies


H= (8) To measure the magnetization characteristics of a
μT test core, two windings are necessary (Fig. 5). A
To calculate the magnetizing current I 0 the mean defined frequency variable current is feed in the
iron length lFe is needed. excitation winding and the induced voltage um ( t )
is measured. As a result, the equations (13) and
Ûh lFe (14) lead to the B − H -diagram and the equations
I0 = H lFe = (9)
ω w s AFe μT (6) and (7) to μT ( f , B ) (Fig. 6) and ϕFe ( f , B )
I 0 = 2 I0 e (
j ϕu −ϕFe ) (Fig. 7), respectively. Now, the equations (1) - (11)
(10)
can be applied to calculate the frequency response
The magnetizing current and the turns of the of any CT with the same core material.
primary coil w p allow to calculate the primary
ie ( t ) w e
current I p . H (t ) = (13)
lFe
wp I p = I 0 + w s I s (11) dB ( t ) um ( t )
= (14)
Finally, the error F dt AFew m
w p Is
F= − 1 = nr − 1 (12)
w sIp
and the phase displacement δ i (2) of the CT are
computable. They are quantities for its
assessment.
To calculate the error of the CT for frequencies
besides the rate frequency, it is necessary to
determine the core parameters μT and ϕFe
depending on the magnetic flux density at each
frequency of interest. In consequence, the
equivalent circuit diagram is still valid for one

© OMICRON electronics GmbH 2011 – Instrument Transformer Measurement Forum


Presentation 6.4

μT in VS/Am 0.99
0,1
50Hz
0.97
80Hz

100Hz 0.95

nr →
200Hz
0.93
300Hz
0.91
400Hz
0,01
600Hz 0.89
25 1000 2000 3000 4000 5000 6000 7000
800Hz f in Hz →
5
1000Hz

2000Hz 0

δ i in ° →
3000Hz
-5
5000Hz
0,001 -10
7000Hz
0,010 0,100 1,000
Bpeak in T
-15
Fig. 6 Measured values of μT ( f , B ) 25 1000 2000 3000 4000 5000 6000 7000
f in Hz →
a) b) c) d)
Fe in deg
50 Fig. 8 Comparison of measurement (solid lines) and
50Hz calculation (dashed lines) of the frequency
80Hz response of a CT model (400:5, 1VA,
45 conventional core material) for a)/c) S=25VA
100Hz cos()=1; b)/d) S=27VA with cos()=0.996
40 200Hz

300Hz
As a result, the extended equivalent circuit diagram
allows the investigation of the influences of
35 400Hz constructive parameter changes on the frequency
600Hz response. Furthermore, the frequency responses
30 of two CT’s with different core materials can be
800Hz compared without a time-consuming and
1000Hz expensive measurement described in Fig. 1.
25
2000Hz

3000Hz
Measurement Suggestions
20
5000Hz The investigations have shown that resistive
burdens deliver much better results than resistive-
15 7000Hz
0,010 0,100 1,000 inductive burdens. Therefore, the secondary circuit
Bpeak in T should be low inductive, including the connection
Fig. 7 Measured values of ϕFe ( f , B ) cables. To gain repeatable measurement results, it
is import to pay attention to EMC problems that
may occur.
To demonstrate the accuracy of the calculation
model, the CT used in Fig. 3 is modeled according
to the steps above. The results of the
Conclusion
measurement and the calculation match very well This paper shows how inductive MV-CT’s behave
(Fig. 8). The maximum deviation of the ratio is ca. regarding the measurement of current harmonics.
1% and occurs at 7 kHz with an resistive-inductive First of all, CT’s usually don’t have resonance
burden. The phase displacement δ i fits very well. points for frequencies up to 10 kHz. The ampere
turns and the core material as well as the burden
are responsible for the accuracy at higher
frequencies. The equivalent circuit diagram of a CT
is extended by frequency dependent parameters
ν ( )
R ω, Bˆ and L ω, Bˆ h to( )
treat harmonic

© OMICRON electronics GmbH 2011 – Instrument Transformer Measurement Forum


Presentation 6.5

frequencies. These parameters can be measured


and are suitable as input parameters for the
equivalent circuit model. As a result, the measured
and calculated frequency responses fit quite well.
This allows the design of MV-CT’s with improved
frequency behavior up to 10 kHz.

Literature
[1] Bartak, G.; Holesnstein, H.; Meyer, J.:
“Technical Rules for the Assessment of
Network Disturbances”, D-A-CH-CZ, 2nd
Edition, 2007
[2] Klatt, M.; Dorado, A.; Meyer, J.; Backes, J.,
Li, R.: “Power Quality Aspects of RuRal Grids
with High Penetration of Microgeneration”,
CIRED, Frankfurt, June 2011

About the Authors


Christian Henze received his
Dipl.-Ing. degree in electrical
power engineering in 2006 and
his Dr.-Ing. degree in 2010 from
the Technische Universität
Dresden. Currently, he is
academic assistant at the
Institute of Electrical Power
Systems and High Voltage
Engineering, TU Dresden, Germany.

Matthias Reiß received his


Dipl.-Ing. degree in electrical
engineering in 1983 from the
Technical University Dresden.
Currently, he is plant manager
of Ritz Instrument Transformers
GmbH, Location Dresden,
Germany.

Peter Schegner received his


Dipl.-Ing. degree in electrical
power engineering from the
Darmstadt University of
Technology, Germany, in 1982
and his Dr.-Ing. degree from
Saarland University, Germany in
1989. Currently, he is Professor
at the Institute of Electrical
Power Systems and High Voltage Engineering, TU
Dresden, Germany.

© OMICRON electronics GmbH 2011 – Instrument Transformer Measurement Forum

You might also like