POSSIBLE MEASUREMENT ERRORS IN RELATION TO HARMONICS AND FLICKER
J.P. Braun, V.J. Gosbell, S. Perera
School of Electrical, Computer and Telecommunications Engineering, University of Wollongong
jbraun@ieee.org
Abstract
The measurement of harmonics and flicker in the MV and LV systems can be made either line-toline or line-to-neutral. Standards provide little guidance on this. As the choice is left to the
operator, it may result in some discrepancies in the reporting for compliance. This paper
investigates the possible errors between the two measurements in relation to harmonics and flicker
in two situations: (i) a large single phase MV load and (ii) the normal variation in LV systems due
to lack of perfect balance across the phases. The problem is first analysed theoretically and then
supported with field measurements.
1.
INTRODUCTION
Standards tend to stipulate the acceptable limits for
PQ disturbances, but leave the choice of measurement
to the utilities. For example the Electricity
Distribution code [1] of the state of Victoria only
specifies level limits and the measurement location for
unbalance and harmonics. Similarly, IEC 61000-2-2
[2] sets limits, but does not address the measurement
techniques. On the other hand EN 50160 [3] is very
specific as it sets the disturbance limits with respect to
the nominal voltage, which is defined as the line-toneutral voltage in a four-wire LV system and line-toline in the case of a three-wire MV system.
In both three-wire MV and four-wire LV systems, it is
possible to measure PQ disturbances between line-toline or line-to-neutral. In LV systems, no VTs are
required for the connection of the instrument to the
network, and therefore it is possible to use either
method as several instruments are designed with this
in mind. However, in MV systems, VTs are always
required and the type of measurement is determined
by their configuration.
In LV systems, the line-to-neutral measurements are
preferred as most loads are connected between line-toneutral. In addition, this gives voltage and current
readings from which it is easy to calculate volt-ampere
and impedance values [4]. However, delta-connected
loads such as motors are exposed to disturbance levels
that correspond to those measured line-to-line.
MV systems require voltage transformers which can
be connected to give signals proportional to line-line
or line-neutral values. Unfortunately, no standard
practice has evolved in distribution systems. The
advantage of line-line voltage signals is that these
reflect how most MV loads are connected. It also
reflects the voltage seen by the downstream LV loads
because of the use of delta/star distribution
transformers. The advantage of line-neutral voltage
signals is the ease of determination of volt-ampere and
impedance values.
The question that comes to mind is how the line-toneutral measurements compare with those made
between lines. The common belief is that these
measurements yield similar results but this is based on
the assumption of balanced PQ disturbances. Two
situations will be investigated in this paper where
there is not a high level of balance. The first is the
large single phase MV load (e.g. traction), connected
between two phases, and which can be examined
theoretically. The second is the normal unbalance
which occurs in LV systems due to the momentary
unbalance between the single phase loads. This needs
to be investigated by actual PQ monitor readings.
2.
THEORETICAL STUDY
2.1
Fundamental voltage
At fundamental frequency, vectors best represent the
various voltages as shown in Figure 1. The line-to-line
voltage can be computed with equation 1 where X and
Y representing two line-to-neutral phasors.
2
V XY = VYN2 + V XN
− 2VYN V XN cos(θ Y − θ X )
(1)
The traditional 3 can be used in the special case
where all line-to-neutral voltages are identical and the
angle between them 120 degrees.
VC
Vh_ca
Vh_cn
A
Vh_an
θB
V
θA
Figure 2 Unbalanced harmonics at order h
N
rectifier at its front end, it is representative of many
distorting loads. The study shows that the phase
angles of higher harmonics are sensitive to the load
power (up to 180 degrees) while at lower frequencies
the changes are much less significant. Hence
unbalance between harmonics is likely to be small at
low frequency, increasing with the harmonic order.
V
AB
VC
V BC
θC
The THD is dominated by the lower frequencies and
is more likely to be balanced.
Figure 1 Voltage phasors
2.2
Harmonic voltages
2.3
The line-to-line voltage at a given harmonic frequency
is the phasor difference of its respective phase-toneutral harmonic voltages. Therefore, the phasor
diagram in Figure 1 as well as equation (1) can be
used at each harmonic frequency. It can be noted that
the spectrum of the line-to-line voltage contains all
components present in the corresponding line-toneutral voltages.
In the worse case the relative line-to-line harmonic
voltage could exceed the line-to-neutral value by
about 15% as shown in equation (1). This can be
considered as a limiting value for large single phase
loads with no significant balanced distorting loads.
This shows the possibility that a site may be
acceptable if assessed on line-neutral values but be
unacceptable if assessed on line-neutral values.
Vh _ XY
V1 _ XY
=
2 ⋅ Vh _ XN
3 ⋅ V1 _ XN
= 1.15 ⋅
Vh_ba
Vh_bn
V AN
BN
Vh_bc
Vh _ XN
V1 _ XN
(2)
With normal unbalance effects in LV systems, the
situation is not so deterministic and the application of
theory is limited. While great care is taken to
minimise unbalance in a distribution system at system
frequency, the harmonic behavior can be rather
different. According to equation 1, the line-to-line
voltage harmonic magnitude could possibly range
from zero to the sum of the two line-to-neutral
harmonic voltages. A much more likely situation is
the intermediate case shown in Figure 2.
The magnitude and phase of the harmonic current
generated by a distorting load varies with the power.
The case of a 3kW single-phase variable speed drive
has been investigated [5]. As this type of load is
designed with a capacitor-filtered diode bridge
Flicker voltage
The analysis of the line-to-line flicker is much more
complicated than the one for harmonics because of the
underlining measurement process based on the “lamp
– eye – brain” model [6]. However, if the analysis is
based on a stationary flicker made up of sinusoidal
fluctuations, the method used so far becomes
applicable. A modulated fundamental frequency
waveform voltage can be expressed by the following
equation:

V XN = V cos(ωt + α )1 +


p
∑V
m
m =1

⋅ cos(Ω m t + α m ) 


(3)
The first term is the fundamental waveform while the
second (between parentheses) represents the voltage
fluctuation composed exclusively of sinusoids.
Using simple trigonometric relations, it is possible to
transform (3) into a sum containing two series:
V XN = cos(ωt + α ) +
p
VVm
cos((ω + Ω m )t + γ 1) +
2
m =1
∑
(4)
p
VVm
cos((ω − Ω m )t + γ 2)
2
m =1
∑
Equation (4) represents the spectrum of the line-to
neutral-voltage.
The line-to-line flicker voltage can now be examined
using equation 1 for each component of equation 4
present in the line-to-neutral voltage. And again, each
side band can assume a value ranging from zero
(cancellation) to the direct sum of the two line-toneutral voltages. Thus, if we make the assumption that
the fundamental is balanced, the magnitude ratio
between each side band to the fundamental or
harmonic can be at best increased by the same factor
2 ÷ 3 . As a result, the line-to-line flicker could be
lower or larger by up to 15% than that of the line-toneutral flicker.
The question that remains is how the Pst associated
with each type of measure are related. This requires
the understanding of how the IEC flickermeter
processes the waveform [6]. Figure 3 shows a
simplified block, (partial) diagram of the flickermeter.
U(t)
Voltage
Adaptor
Squarer
Filters
Squarer
LP Filter
(0.53 Hz)
IFL
measured the voltage between lines. The GPS
synchronises all samples and data to the Universal
Time Coordinated (UTC) within an accuracy of 1
microsecond [8]. The following Arbiter 1133A data
were used:
Phasors:
Harmonics: Magnitude and phase angle of
harmonics to the 50th. Available once
per second.
Flicker:
Figure 3 IEC flickermeter
3.
MEASUREMENT RESULTS
To investigate the validity of the above theoretical
considerations at typical sites, field measurements
were undertaken in a distribution network. A first set
of measurements was taken at a zone substation and a
second at the power laboratory of the University of
Wollongong.
Also, as shown previously, it is important that the
magnitude and phase of each voltage is measured at
the same time so as to permit their vectorial addition.
This also requires the simultaneous measurement of
the line-to-neutral and line-to-line voltages. This was
achieved by means of two GPS synchronised Arbiter
1133A PQ analysers as shown in Figure 4.
3.1
Pst and Plt computed as per IEC 610004-15 [6]. Synchronised intervals allow
direct comparison.
Measurements at a zone substation
Measurements on the 11kV bus are made through an
auxiliary transformer acting as a measurement VT.
The disturbances measured across the transformer are
those present in the 11kV system. Being a delta/star
transformer and with only a small load on the
secondary, no zero-sequence situation has been
assumed.
3.1.1 Harmonics
The dominant harmonic at this site was the 5th. Figure
5 and 6 show the voltage measurements. The line-toline phase relation for the 5th harmonic is shown in
Figure 7. This must be contrasted with the same
measurement made on the 17th harmonic (Figure 8),
showing clearly that higher order harmonics have
larger phase angle fluctuations.
4 .8
A
C
4 .6
4 .4
4 .2
Voltage (V)
While beyond the scope of the this paper, it can be
shown [7] that the instantaneous flicker level (IFL)
present at the output of the low pass filter in Figure 3
is the sum of the IFL due to by each side band on its
own. As a result, the Pst measured between lines can
theoretically range between zero and 1.15 times the Pst
in relation to phase-to-neutral voltage.
Magnitude and absolute phase angle
(relative to UTC) of the fundamental.
Available 20 times per second.
4
3 .8
3 .6
3 .4
A
B
C
N
B
3 .2
3
4 3 :0 0
4 6 :0 0
4 9 :0 0
5 2 :0 0
5 5 :0 0
5 8 :0 0
T im e [ m in : s e c ]
Figure 5 Line-to-neutral 5th harmonic voltage
8
CA
CB
7 .5
GPS
Arbiter 1133 A
Figure 4 Synchronised measurements of PQ disturbances
7
Voltage (V)
GPS
Arbiter 1133 A
6 .5
6
5 .5
BC
The first instrument was set up to measure the
voltages between lines and neutral while the second
5
4 3 :0 0
4 6 :0 0
4 9 :0 0
5 2 :0 0
5 5 :0 0
T im e [ m in : s e c ]
Figure 6 Line-to-line 5th harmonic voltages
5 8 :0 0
3 .5
150
3
2 .5
Pst (1 min)
100
CA
Angle (deg)
50
0
4 3 :0 0
4 6 :0 0
4 9 :0 0
5 2 :0 0
5 5 :0 0
5 8 :0 0
2
1 .5
1
-5 0
0 .5
AB
0
1 3 :0 0
-1 0 0
1 4 :0 0
1 5 :0 0
1 6 :0 0
1 7 :0 0
T im e [ H o u r s ]
-1 5 0
Figure 9
T im e [m in :se c ]
Line-to-neutral flicker for phase A
Figure 7 Line-to-neutral 5th harmonic phase angles
3 .5
150
3
100
Angle (deg)
50
AC
0
4 3 :0 0
4 6 :0 0
4 9 :0 0
5 2 :0 0
5 5 :0 0
5 8 :0 0
Pst (1 Min)
2 .5
-5 0
2
1 .5
1
AB
0 .5
-1 0 0
0
1 3 :0 0
-1 5 0
1 4 :0 0
1 5 :0 0
1 6 :0 0
1 7 :0 0
T im e [ H o u r s ]
-2 0 0
Figure 10 Line-to-neutral flicker for phase B
T im e [ m in : s e c ]
Figure 8 Line-to-neutral 17th harmonic phase angles
3 .5
The 95% harmonic distortion levels for the 5th
harmonic are:
AB:
BC:
CA:
1.87%
1.66%
1.91%
It can be seen that the there is a small difference in the
harmonic voltages assessed by the two sets of
readings, but not enough to affect the acceptability of
the site. It is usual to report a site using only the
largest of the three assessed voltages. This would give
values of 1.97% using line-neutral voltages and 1.91%
using line-line voltages. At higher harmonics, the
discrepancy would be expected to be larger while the
limits are smaller, and there is the possibility of
significant discrepancies.
3.1.2 Flicker
The flicker present at this location was relatively high
due mainly to the presence of a smelter up-stream in
the system. The measurements made line-to-neutral
are similar in all the phase as shown in Figures 9, 10.
The high peaks that can be noticed in these figures are
due to the operation of the ripple control system.
The line-to-line flicker, shown in Figure 11, is
surprisingly very similar to the line-to-neutral
measurements. This tends to suggest that the flicker
on all phases are correlated and balanced.
As this site was fairly well balanced for all
frequencies, no important difference was observed
between line-to-line and line-to-neutral measurements.
Pst (1 min)
A: 1.97%
B: 1.71%
C: 1.80%
3
2 .5
2
1 .5
1
0 .5
0
1 3 :0 0
1 4 :0 0
1 5 :0 0
1 6 :0 0
1 7 :0 0
T im e [ H o u r s ]
Figure 11
3.2
Line-to-line flicker between phase A and B
Measurements at user side
The measurements in the LV voltage side were made
in the power laboratory at the University of
Wollongong. The three-phase supply was monitored
in a similar fashion as at the zone substation. No
major loads were operating in the laboratory during
the measurements.
3.2.1 Harmonics
While the low order harmonics were again observed to
have relative phase angles of 1200, the case of the 11th
was rather different. It showed on average an angle of
900 between the phase A and B and 1400 for A and C
as shown in Figure 12. As a result, the line-to-line
harmonic voltages were affected as follows: AB was
smaller while BC and CA were respectively higher
compared to what they would have been if all the
angles were 1200 as for line-to-neutral. This is a clear
case where harmonic levels vary with the type of
measurement. Figure 13 and 14 show respectively the
line-to-neutral and line-to-line 11th harmonic voltages.
0 .5
200
150
0 .4
Pst (10 min)
Angle (deg)
100
AC
50
0
3 0 :0 0
3 2 :0 0
AB
3 4 :0 0
3 6 :0 0
3 8 :0 0
0 .3
0 .2
4 0 :0 0
-5 0
0 .1
-1 0 0
0
1 3 :0 0
-1 5 0
1 4 :0 0
1 5 :0 0
1 6 :0 0
1 7 :0 0
T im e [ H o u r s ]
T im e [ m in : s e c ]
Figure 16 Line-to-line flicker
Figure 12 Line-to-neutral 11th harmonic phase angles
4.
1 .2
A
CONCLUSIONS
1
Voltage (V)
0 .8
0 .6
0 .4
B
C
0 .2
0
3 0 :0 0
3 2 :0 0
3 4 :0 0
3 6 :0 0
3 8 :0 0
4 0 :0 0
Because of the lack of clarity in standards, the
measurement of PQ disturbances can be made either
line-to-line or line-to-neutral in MV and LV systems.
This paper investigated the possible discrepancies
between the two sets of measurement in the case of
harmonics and flicker.
T im e [m in :sec ]
Figure 13 Line-to-neutral 11th harmonic voltage
Based on a vectorial analysis done at the frequencies
contained in these disturbances, it was shown that the
line-to-line measurement could range between 0 and
1.15 times that measured between line-to-neutral
voltages at any given frequency. These two extreme
cases correspond to an angle between two vectors of
respectively 0 and 180 degrees. This is highly unlikely
in practice except at sites dominated by one single
phase load.
2
1 .8
CA
1 .6
Voltage (V)
1 .4
1 .2
1
0 .8
BC
AB
0 .6
0 .4
0 .2
0
3 0 :0 0
3 2 :0 0
3 4 :0 0
3 6 :0 0
3 8 :0 0
4 0 :0 0
T im e [ m in : s e c ]
Figure 14 Line-to-line 11th harmonic voltage
The 95% harmonic distortion levels for the 11th
harmonic are:
A: 0.42%
B: 0.30%
C: 0.37%
AB:
BC:
CA:
0.31%
0.35%
0.41%
Surprisingly, no major discrepancy is noticeable
despite a clear case of higher line-to-line voltage.
3.2.2 Flicker
Similar to the MV case, the flicker measured line-toneutral or line-to-line is almost identical for all phases
as shown in Figure 15 and 16.
Field measurements at more typical sites show that
these extreme cases are not observed and that
harmonics and flicker measured line-to-neutral are
very similar to those measured line-to-line. Some
visible discrepancies seem to occur only at high order
harmonics.
While more systematic measurements are required, it
cannot be excluded that discrepancies can possibly
occur notably at sites having single-phase distorting
loads or at more normal sites at high harmonic
frequencies. It is recommended that standards should
state clearly which set of voltages (line-line or lineneutral) should be used for assessment purposes.
5.
REFERENCES
[1]
Office of the Regulator-General, Victoria,
“Electricity Distribution Code”, January 2002
[2]
IEC 61000-2-2, “Compatibility levels for lowfrequency
conducted
disturbances
and
signalling in public low voltage power supply
systems – Basic EMC publication”, 2002
[3]
EN50160,
“Voltage
characteristics
of
electricity supplied by public distribution
systems”, 2000
0 .5
Pst (10 min)
0 .4
0 .3
0 .2
0 .1
0
1 3 :0 0
1 4 :0 0
1 5 :0 0
1 6 :0 0
T im e [ H o u r s ]
Figure 15 Line-to-neutral flicker
1 7 :0 0
[4]
J. Schlabbach, D. Blume, T. Stephanblome,
“Voltage quality in electrical power systems”,
IEE Power and Energy Series 36, 2001
[5]
A. Mansoor, W.M. Grady, A. Chowdhury, M.J.
Samotyj, “An investigation of harmonics
attenuation and diversity among distributed
single-phase power electronic loads”, IEEE
Transaction on Power Delivery, Vol 10, No 1,
January 1995, p467-473
[6]
IEC 61000-4-15, “Flickermeter – Functional
and design specifications”, 2003
[7]
K. Srinivasan, “Digital measurement of voltage
flicker”, IEEE Power Delivery, Vol. 6, No. 4,
Oct. 1999, p1593-1598
[8]
Arbiter Systems, Inc., “Model 1133A Power
Sentinel Operation Manual”, 2001