Professional Documents
Culture Documents
Classification of Power System Events: Voltage Dips: February 2000
Classification of Power System Events: Voltage Dips: February 2000
net/publication/3883500
CITATIONS READS
35 311
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Irene Y. H. Gu on 08 January 2016.
Abstract: The extensive monitoring programs that are run by power However, because of the large size of these databases new
utilities enable new insights of the power system operation and new tools are required that will enable the automatic processing of
characterisation methods must be used for the classification and the recordings. These tools should be able to extract all
analysis of the recordings. This paper focuses on events that cause a relevant information and offer all the necessary statistics.
temporary decrease in the fundamental frequency voltage magnitude
(voltage dip). The analysis of the recordings from surveys in This paper focuses on a large class of power system events:
medium and low voltage networks shows that new classes of voltage those that their main characteristic is a temporary decrease in
dips should be introduced in order to characterise the events that are the fundamental frequency voltage (voltage dip) in one or
captured by the power quality monitors. Aim of the paper is to show more phases. Voltage dips have attracted a lot of attention
the distinctive characteristics of each class and give the guidelines due to the problems that cause to equipment like adjustable
for the automatic processing of the recordings. Finally, a large speed drives, computers, industrial control systems etc. They
number of recordings from a medium voltage network are classified are mainly caused by faults and induction motor starting [ 1-
using these characteristics and the results are presented. 21.
It has been clarified already that voltage dips cannot be
Keywords: power system monitoring, power quality, voltage dips,
power system faults, power transformers, induction motors, Short characterised by a single magnitude and duration for each
Time Fourier Transform. phase. In [3], it is shown that these values are inadequate in
describing a voltage dip and especially its impact on sensitive
loads. In [4], the case in which the voltage dip contains
I. INTRODUCTION multiple components is highlighted.
Aim of this paper is to show that there are certain types of
The importance of power quality in the operation and voltage dips that should be considered in the analysis of
design of power systems is widely accepted. Utilities are monitoring data as showed the analysis of a large number of
forced to prove that the offered power meets the specified recordings. Furthermore, the distinctive characteristics of
standards and the requirements of the customers. each type are extracted and presented.
Furthermore, when the quality of power is not accordant with For the analysis of the different types of voltage dips,
these requirements, immediate action should be taken. For recordings are used that were performed at medium and low
these reasons power utilities run extensive monitoring voltage networks in Sweden (by Goteborg Energi Nat AB),
programs in order to provide statistics regarding the quality of Norway (by SINTEF Energy Research) and Scotland (by
the power and the necessary information for the solution of Scottish Power). Finally, a large set of recordings is
problems. Additional benefit of monitoring is that the classified, explained and presented in terms of the above
engineers obtain new knowledge about system operation. The mentioned types.
developments in communications and electronics make
monitoring even more attractive from a cost point of view and 11. BACKGROUND
is expected that more monitors will be installed in the coming
years. A. Short Time Fourier Transform.
The monitors according to their settings capture different The squared magnitude of the Fourier Transform is the
types of events and long sequences of measurements are classic method used to represent the frequency domain
transferred to the databases. Some of these events are due to information (spectrum) of a stationary signal. However, this
normal system operations, do not cause any problem and are approach does not provide any information on the time
not of particular interest, while others need to be categorised properties on the frequency components that the signal
according to the needs of the monitoring program. contains.
The Short Time Fourier Transform ( S m ) overcomes the
problem by partitioning the signal into short time segments
(during which the signal is considered stationary) then
applying a weighting function to the signal within each
segment, prior to evaluating the Fourier Transform. This
I I
0 100 2w 300 400 500 600
I I
0 100 200 300 400 500 600 0 50 100 150 200 250
I I
0 100 200 300 400 500 600 0 50 100 150 200 250
15
12
0 100 200 300 400 500 600 0 50 100 150 200 250
Time(msec) Tlme(msec)
Fig.3. (a)-(c) Phase-to-phase voltage waveforms during a fault (d) The Fig.5. (a)-(c) Phase-to-groundvoltage waveforms during a self-extinguishing
fundamental frequency magnitude of voltage. fault (d) The fundamental frequency magnitude of voltage.
- 748 -
C. Voltage dips due to transformer energising D. On the classification of voltage dips
Voltage dips due to transformer energising have been As mentioned in the introduction, automatic processing of
reported in the literature [7]but neither the frequency in the recordings is needed for the analysis of the databases
which this event appears in a power quality survey nor the created by power quality surveys. In this direction, the
characteristics of this event in terms of voltage have been distinctive features of each type must be extracted and utilised
presented. The main attention has been given to the effects of in a rule-based system.
the inrush current on the protection relays of the transformer The analysis of a large number of recordings from different
itself [8]. surveys showed that the following characteristics can be used
In a transformer under steady-state conditions there is a to distinguish between the different types of voltage dips:
particular value of flux in the core, for each point on the 1. The rectangular shape of the fundamental frequency
voltage waveform. When the transformer is energised, the voltage magnitude: voltage dips due to faults have this shape
initial value of flux in the core might not necessarily be the as shown in Fig. 1 due to the operation of protection that
steady-state value for this particular point on the voltage clears the fault and forces voltage to recover fast. On the
waveform. A transient will occur to change the flux in the other hand, voltage dips due to induction motor starting and
core to the steady state condition. In general, this will cause transformer energising recover gradually as show in Fig. 2
the flux to go above the saturation value once each cycle until and 6 providing that there is no protection maloperation.
the average value of the flux for a cycle has decayed to nearly 2. The symmetry between the different phases: voltages
zero. This temporary over-fluxing of the transformer core during induction motor starting are symmetrical because an
causes high values of the magnetising current. This induction motor is a balanced load and each phase will draw
phenomenon is known as magnetising inrush current. In tum, the same inrush current causing a voltage dip of the same
short duration voltage dips are caused that might result in an magnitude for all phases as in Fig. 2. In the case of
unwanted tripping of differential protective relays. As transformer energising this is not true. The inrush current
described in [7] and [9]this voltage dip can be long in depends on the degree of saturation of each phase therefore,
duration and drive more transformers into saturation. each phase will experience a voltage dip of different
Fig 6 shows the voltage waveforms during transformer magnitude (non-symmetrical) as shown in Fig. 6. Finally,
energising. The measurement was performed in an 1 1 kV voltage dips due to faults can be symmetrical or non-
network. The fundamental frequency voltage magnitude drops symmetrical depending on the type of the fault.
for a very short time and recovers gradually as the 3. The harmonic contents of voltage during the voltage dip: it
magnetising current decreases. The largest drop is is well documented that the voltage waveforms during
approximately 8% of the pre-event voltage. transformer energising present high harmonic distortion [e.g.
91. The analysis of voltage waveforms during transformer
energising showed that the harmonic distortion is temporary
and the even harmonics (second and fourth) are contributing
the most as shown in Fig. 7 which shows the harmonics up to
the fifth of the first voltage waveform of Fig. 6.
I I These features have been already applied in a ruled-based
0 100 200 300 400 500 system for power system event classification and the results
show strong potential [IO].
t
3.5 I "I,
I I
14
0 100 200 300 400 500
Time(msec)
Rg.6.(a)-(c) Phase-to-phase voltage waveforms during transformer Fig.7. Harmonicsof voltage (ofFig. 6) during transformer energising as
energising(d) The fundamental frequency magnitudeof voltage percentage of the fundamental frequency voltage using the STFT.
- 749 -
V. ANALYSIS OF A DATABASE USING THE PROPOSED Category 4 consists of recordings that did not present any
CLASSIFKATION significant drop in voltage. Most of these recordings were
captured by the monitors because of a variation in current.
Recordings that correspond to one month of monitoring in a
Category 5 contains recordings that although contain a
medium voltage network (33 and 11 kV), were analysed using
voltage dip could not be explainedklassified using the above
the STFI’. Using the features that were described in the
categories.
previous section, these recordings were classified in different
Voltage dips due to induction motor starting were not
categories as shown in Table 1. The monitors are triggered by
found. The explanation is that induction motors connected in
current or voltage disturbances and all the resulting
low voltage do not cause severe voltage dips in the medium
recordings are considered here.
voltage network.
Voltage dips due to faults turned out to be the category with
the largest number of recordings. They are subdivided into
three classes: VI. CONCLUSIONS
a) Voltage dips of duration longer than 3 cycles: this is the The analysis of a large number of recordings mainly from
largest class of all and their duration varies with the type of medium voltage networks showed that new characterisation
protection that is used to clear the fault that caused them. methods are needed because not all voltage dips can be
b) Voltage dips of duration shorter than 3 cycles: these are described by a single magnitude and duration for each phase
either due to self-extinguishing faults or fuse-cleared faults. assuming a rectangular shape of the fundamental frequency
However, it is not clear to the authors for how long a self- voltage.
extinguishing fault might persist before it disappears. It is The most interesting findings in the analysed data are that:
possible that self-extinguishing faults exist in the first 1. A large part of voltage dips due to faults are multistage,
subclass (voltage dips longer than 3 cycles). Records of the i.e. they present different levels of voltage dip magnitude
protection operation from all the voltage levels are needed to before the fault is cleared. As explained above this could be
accurately extract this type of voltage dip from the database. either due to a change in the nature of the fault or due a
c) Multistage voltage dips: in this group there are recordings change in the system configuration during the protection
that have two or even three different levels of voltage dip operation. Further work is needed in order to distinguish these
magnitude. In this category we recognised shallow voltage two cases.
dips of duration that points towards a fault in the transmission 2. A new category of voltage dips should be taken into
system that is cleared only after zone 2 operation of distance consideration: those that are due to transformer energising.
protection. The number of recordings due to this type of event is
Voltage dips due to transformer energising form a significant. This type of voltage dips might take place either
considerable part of the database. These are basically non- due to normal system operation (manual energising of a
rectangular, non-symmetrical voltage dips having a temporary transformer) or due to reclosing actions following the opening
harmonic distortion. This type of event might take place of circuit breaker to clear a fault.
either due to normal system operation (manual energising of a 3. Most of the recordings (70%) from a medium voltage
transformer) or due to reclosing actions following the opening network present a distinctive voltage signature, i.e. they can
of circuit breaker to clear a fault. be characterised in terms of voltage. These recordings present
Category 3 contains recordings in which voltage increases either voltage dips (due to faults or transformer energising) or
or decreases in a step way without a voltage dip. These events step changes in voltage due to load switching or voltage
could be due to on-load transformer tap changing, capacitor compensation related events.
switching or reactor switching. 4. A number of fault-type dips are of very short duration: less
than three cycles. These may be associated with fault clearing
by fuses, due to self-extinguishing faults, or other phenomena.
A further investigation of the recordings and of the various
fault-clearing mechanisms is needed.
I I I Durationz3cycles I 190records 1
I
There is a recognised need for automatic processing of the
recordings that are obtained from power quality monitors
“Itage dips due Duration < 3cycles 48 records because of the large size of the created databases. Towards
to faults
I Multistage I 65records 1) this direction the distinctive characteristics of the different
types of voltage dips are given in section 111.
I12 Voltage dips due to transformer energising I 85 records
1 The method can also be used as a basis for more
sophisticated event characterisation. An additional
I 3 1 Step changes in voltage I 134 records 1 classification and characterisation algorithm can be applied to
fault-type events, e.g. as proposed in [ I 11. This method
4 Insignificantdrop in voltage (less than 2%) 221 records characterises the event in such a way that the characteristics
do not change when the disturbance propagates to lower
5 Unclassified I3 records
voltage levels.
- 750 *