Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Abstract - This article presents the main results of a before current chopping on the first phase to cut was
large overvoltage protection simulation study, carried out 0.3ms, in order that potentially high overvoltages are
on vacuum breaker switched motors. Different protection generated.
solutions, as well as a great number of electrical network The first step was a detailed modelling of the vacuum
parameters, motor starting current and connecting cable breaker, based on comparisons with test results. Then, a
characteristics were varied. The study showed that specific software application, for automatic simulation
overvoltage protection installation is highly recommended tests execution and results recordings was developed.
for frequently started MV motors, switched by vacuum This application permitted to consider many network
breakers regardless of the cable length, or even motor parameters, in order to cover the majority of practical
starting current. The main conclusions and results of this situations that may occur.
study are presented and commented on. The well known The acceptable overvoltage level is defined in the IEC
ATP-EMTP and MATLAB software were used for these 71-2:1996 as 3.0pu in case of switching with reignitions.
analyses. This maximum level was accounted for in the study.
However, motors are able to withstand higher
Index Terms — Vacuum circuit breaker, overvoltage overvoltages, and this should be taken into account, when
protection, surge-arrester, R-C filter, motor switching, choosing an appropriate protection. Some withstand
ATP, EMTP, MATLAB, Alternative Transients Program, values of the overvoltage levels, as a function of the
Electromagnetic Transients Program. Line and Cable impulse rise-time are depicted on Fig. 1:
Constants
I. INTRODUCTION
The next section presents briefly the vacuum breaker Test: 26 Voltage on phase B
20
model development. Section III describes the complete
EMTP
test-circuit, including the VCB model and starting motor 15 Lab test
model. Section IV gives an idea of the used program
supply. Sections V and VI analyze the results of the 10
Voltage, [kV]
0
The first step in the study was the establishment of the -10
breaker model. A preliminary bibliographic research, [11],
-15
[12], [13], gave the main ideas in vacuum breaker
modeling for overvoltage analysis. A primary version of -20
the model was developed in ATP-EMTP software, by
-25
using the MODELs Section. The model fitting was 0 5 10 15
performed by superposing the simulation results with t, [ms]
those recorded in laboratory tests. For this purpose it was Fig. 3 Laboratory test circuit validation, comparison
necessary to model also the laboratory test circuit. The without reignitions
last is defined by the IEC 62271-110:2005, circuit A.
On Fig. 3, the frequency and peak value match are of
The dielectric strength recovery, chopping current level importance for validating the laboratory test circuit model,
and high-frequency arc quenching were analyzed from as well as to give the correct current chopping value.
the recordings and introduced in a convenient way in the
model. In a nutshell the vacuum breaker was modeled as The next figure presents an example of switching with
three ideal switches, controlled by a MODELs block as on reignitions. It served to analyze the dielectric strength
the next figure: evolution of the breaker as well as the conditions of arc
quenching, in order to determine the best way to model it.
EMTP
us(t) i(t) ul(t) 40 Lab test
0,1
20
0
The above model also takes into account a measured
delay between openings of the three phases of about -20
0.3ms.
The breaker model fitting has followed two main steps: -40
Gen.
3.3-11kV Cs, Prot. Cmot
Lmot
IV. PROGRAM FOR AUTOMATIC TESTS RUN
Prot. Rd
33nF Rmot
Zn 20 kOhm
Isolated,
40 Ohm,
0 Ohm Rinf
As it was said, the study has covered a great majority of
Source Motor realistic situations, concerning the voltage level, cable
Fig. 5 Electrical diagram of the tested circuit type and length, motor power (starting current). Thus, it
was necessary to develop an adapted program able to
The presented circuit is particularly convenient to generate, execute and analyze each different case, as
experience high overvoltages, due, partially, to the low well as to record the results of interest. This task was
capacitance at the source side (33nF). The latter depends performed by coupling three different software packages:
on the upstream circuit. In our case it represents a short 1. MATLAB was used for the overall control of the
upstream circuit i.e. the vacuum breaker is connected process, the case generation, execution and
close to the main busbar. In a future study we are going to post treatment.
investigate the impact of this capacitance on the 2. ATP-EMTP was used to perform the electrical
overvoltage levels and the protection recommendations. network simulations
3. MS Excel served for recording the final results.
For the generation of the different test cases the
following circuit parameters were systematically varied: Having a total number of more than 12 000 test cases,
• The source voltage: 3.3-11kV 5 statistical simulations per case were performed,
• The source earthing impedance: Isolated, 40 Ohm, matching an acceptable overall simulation time and
direct statistical parameters variation.
• The cable type: multicore, single core, grounded
on one side or both sides V. SIMULATION RESULTS
• The cable length: 10-1250m
• The motor starting current: 25-3000A A. Case without any protection
• The applied protection: without, surge arrester,
The first test case was without overvoltage protection
wave-sloping capacitor and R-C branch
devices. In such a way, preliminary information about the
• The protection emplacement: motor side or expected surge-voltage levels as well as the situations
breaker side where protection may not be needed is gathered. All of
• The protection connecting cable length: 3m for all, the studied network configurations showed high
and also 1m for surge arrester on each side (motor overvoltage levels that have reached values far above the
to protection and protection to ground). recommended dielectric strength for lightning strike. More
detailed analysis showed it was due to the phenomenon
B. Cable modeling of Virtual Current Chopping (VCC), [14], [15]. This
phenomenon is to cut the load current in an adjacent
The motor connecting cable was modeled by the LCC phase by capacitive and inductive surge transmission
(Line Cable Constants) routine, available in ATP-EMTP. between phases, originating from the first cutting phase.
The frequency dependent SEMLYEN model was chosen In such a case, due to the high chopped current value,
as most appropriate for modeling the single core cables, the produced overvoltage is very important. In return, the
and the distributed parameter Bergeron model was used created surge is transmitted back and forward to the
for the multicore cables. The corresponding frequency, at adjacent phases.
which the cable parameters are calculated is a function of
their length and isolation, the last in terms of its dielectric This phenomenon took place with all the motors,
constant, as recommended in the ATP-EMTP Rule Book. including those having a starting current of 3000 A.
Normally it is believed that the value of this current is
C. Starting motor model great enough to avoid that a fast transient forces it to zero
and it gets cut. However, simulations, confirmed by
As it could be seen on Fig. 5, the motor model is an laboratory recordings, showed that high frequency
electric circuit equivalent, characterized by an R-L three currents, passing through the breaker, can reach more
phase ungrounded load and parallel capacitors to ground, than 5000 A. An example of cutting with VCC is shown on
calculated according to the indications in the above the Fig. 6. The simulated cable length is 500m, for a
mentioned IEC standard (Lmot-Cmot circuit tuned at 15 1000A starting current machine at 11kV. The cable
kHz). The R-L values are calculated in order to obtain a capacitance is of around 221nF, and the total capacitance
starting current 6 times greater than the nominal one.
Schneider Electric 2008 - Conferences publications
PCIC Europe 2008. 5th . 10 – 12 June 2008 – Weimar - Germany
Petroleum and Chemical Industry Conference Europe – Electrical and Instrumentation Applications
to ground with machine's capacitance included is reduced to 1m on each side. The results obtained were
226.6nF. The corresponding natural oscillating frequency more optimistic. They showed that:
is 2.39 kHz. 1. Surge-arrestors were able to reduce substantially
4000 the overvoltages and can be recommended for
[A] overvoltage protection.
2000
2. Connecting them to the motor terminals is more
First phase opens
here efficient, but the connection cable length should be
very short <1m, the obtained overvotage level
0 could be sometimes even lower than 3pu, but is
more often greater. Thus the average value has
-2000
not been over 5.0pu for all the motors, voltage
levels and different cable lengths. Some limitations
were observed when cable screens were grounded
-4000 on breaker side only.
VCC starts here
3. It is possible to connect them on the breaker side
-6000
and still to get an overvoltage level of about 4pu;
0 1 2 3 4
(f ile DPV_etude_VCB_MVP_test_or_ss_PRF.pl4; x-v ar t) c:VCBB -VCB_B
5 6 [ms] 7 that is acceptable for some motors, in such case
Fig. 6 VCC phenomenon on adjacent phase B, 11kV, the connection cable length is respected easily.
1000A starting current machine, 500m SC cable 4. For protection on motor side, the overvoltage
levels were slightly dependent on cable type.
As it can be seen the VCC takes place around 1ms However, for three-core cables, with motors of
after the cutting in the first phase at around 0,5ms. starting currents in the range 100-600A @11kV,
A similar situation, with lower maximum overvoltage the observed average overvoltages were
levels is observed for larger motors, starting current up to somewhat higher (3.75pu, 1m arrester cable) than
3000A, even if the cable length is 1250m. The results are for 25-50- >600 A (3.3pu, 1m arrester cable).
also confirmed in [8] and [16]. 5. For protection mounted on motor side, with single
core cables, screens grounded on both sides, the
The cable type is also an important factor, influencing in average overvoltage levels were under 3.0pu for
a great manner the overvoltages. With multicore cable, cable length greater than 200m and system
the probability of high overvoltages is greater than that voltage under 5.5kV.
with single core cables, and it is less reduced with the 6. The voltage rise time is not modified by surge
cable length, compared with single core cable. This is due arrestors, i.e. it is possible to get an acceptable
to the higher capacitive coupling between phases, overvoltage maximum value but not being able to
because of the proximity of the cables as well as the lack stand the overvoltage rise time, also confirmed in
of screens on them. [18].
The neutral earthing of the power system wasn't found Surge arrestors did not avoid completely the risk for
to have a clear influence on the overvoltage level; similar VCC, although they have reduced it by limiting the
levels were observed with the three different earthing. overvoltages up to a given level.
[A]
A common means of overvoltage protection is the use
-500
of surge arrester. In fact this is the most widely used
protection and the least expensive. The same test cases
-1000
as previously were loaded for simulation. Now two
connection points are considered:
1. At motor terminals -1500
2. At the breaker
-2000
The obtained results showed that the surge arrestors Fig. 7 Current on adjacent phase B, 11kV, 1000A
efficiency is very dependent on the connection point as starting current machine, 500m SC cable, protection by
well as the mounting conditions. For example, mounting surge arrester (27.6kV@10kA)
with 3m connection cable on each side, motor to surge
arrester's hot spot and between arrester's cold spot and In this example the VCC phenomenon has been
the ground, blinded the surge arrester. The reached avoided, but the current in the adjacent phase is
overvoltage levels were unacceptable, regardless if the experiencing fast transients. Again the efficiency of the
surge arrester was connected on motor side or on arrester will be a function of its connecting cable length.
breaker side. This mounting is however a common
practice when the connection point to the earthing system C. Protection by surge capacitor
is far from the motor.
The surge capacitor is used when more efficient
In a second step the connection cable length was
Schneider Electric 2008 - Conferences publications
PCIC Europe 2008. 5th . 10 – 12 June 2008 – Weimar - Germany
Petroleum and Chemical Industry Conference Europe – Electrical and Instrumentation Applications
overvoltage protection is needed i.e. when the prices in this range are close to each other. The issue is
overvoltage level should strictly remain lower than 3pu. more related to the resistance which will see much more
Inserting surge capacitor helps increasing the voltage rise current in a normal steady state. We are going to discuss
time by reducing the natural oscillation frequency of the here the results obtained with the second R-C
circuit on the motor side. The same test cases were combination.
loaded for simulation. The tested capacitor value is
500nF, available for this voltage level range. The obtained In general, the results showed that the R-C filter is the
results showed that: best solution for overvoltage protection:
1. The wave-sloping capacitor is quite independent of 1. Some limits appear with multicore cables; this
motor cable type (single core or multicore) issue has been observed also with wave-sloping
2. The overvoltage levels are limited far below 3.0pu capacitors. But for single core cables, grounded on
(less than 2.0 pu was typical), but its application is both sides the R-C filter gave very satisfactory
limited only to motors with starting current of less results, limiting the over voltages far below the 3.0
than 300A@3.3kV, and less than 200A@11kV. pu level when connected on the motor side for all
3. With more powerful motors the natural circuit the motors, voltage levels and different cable
frequency becomes too high and the voltage lengths.
variation is faster. This leads to multiple reignitions 2. On the breaker side R-C filter have a limited
on the first phase to cut and a consequent application, in terms of starting current
transient transmission, followed by the VCC (<600A@3.3kV; <200A@11kV), for an overvoltage
phenomenon. The overvoltage levels are much limitation under 3.0pu.
higher than 5.0pu. The situation is close to the 3. Connecting them on the breaker side will generally
case without any protection. Increasing the limit the overvoltages under 5.0pu with some
capacitor value will certainly decrease the rise- exceptions on 11kV.
time, however fast transients would be also quickly
transferred between phases. A typical situation is shown in Fig. 9; the example is the
same as that in A. the arrester being replaced by the R-C
Fig. 8 shows a typical simulation result, the same case filter:
as previously, the surge arrester is replaced by wave- 1400
4000 800
2000 600
-4000 0
-6000 -200
0 1 2 3 4 5 6 7 [ms] 8
(f ile DPV_etude_VCB_MVP_test_or_ss_PRF.pl4; x-v ar t) c:VCBA -VCB_A
-8000
0 1 2 3 4 5 6 [ms] 7 Fig. 9 Current on phase A
(f ile DPV_etude_VCB_MVP_test_or_ss_PRF.pl4; x-v ar t) c:VCBB -VCB_B
Fig. 8 VCC phenomenon on adjacent phase B, 11kV, It can be observed that the current in phase A is
1000A starting current machine, 500m SC cable, chopped, and is later followed by reignition. At that time
protection by wave-sloping capacitor (500nF) the current value has already grown sufficiently to avoid
zero crossing. This delay of the reignitions is due to the
Compared to the situation without any protection, the presence of the resistor. The corresponding overvoltage
capacitor has decreased the frequency of the oscillations, was lower than 2.0pu.
but the VCC can be clearly observed.
-200
D. Protection by R-C filter [A]
-400
The R-C filter is the last overvoltage protection
equipment, considered in this study. The resistance and Impact on the adjacent phase
-600
capacitor values were chosen in the generally
recommended and available range of 30-50 Ohm, 200 –
500nF, [6]. Two sets of values were considered, in order -800
be more expensive solution, but in practice the capacitor Fig. 10 Current on phase B
Schneider Electric 2008 - Conferences publications
PCIC Europe 2008. 5th . 10 – 12 June 2008 – Weimar - Germany
Petroleum and Chemical Industry Conference Europe – Electrical and Instrumentation Applications
X. VITA