= +
2  3
Where,
U
a
: the instantaneous value of power frequency voltage at the instant of current
chopping;
i
a
: the level of current chopping;
m
: the magnetic efficiency;
L and C: the impedance of load;
Moreover, if the magnetic efficiency is assumed equal to 1, the calculated result of
U
LP
by equation 23 is around 22.6kV and the simulated result is around 21kV. The
small deviation comes from the equation 23, because the losses of load are not
considered in it.
10
Figure 2  4 Upper trace: phase current in the VCB; Middle trace: voltage on the load side; Lower
trace: voltage across the contacts of VCB
11
If the arcing time is decreased from 600s to 550s, the second situation takes over,
as shown in Figure 25 and Figure 26.
Figure 2  5 Voltage across the contacts of VCB under second situation
At t=7.57ms, the contacts of VCB are opened. At this moment, the level of current
chopping is calculated by the mean value with normal distribution. After the arc has
existed around 550s, the power frequency current reaches to the level of current
chopping and is chopped immediately, as shown in upper right trace of Figure 26.
The voltage across the VCB starts to have a very HF oscillation after the power
frequency current is chopped, as shown in lower left trace of Figure 25. The HF
oscillation is induced from the interaction of inductance of cable L
K
, capacitance of
gap C
S
, capacitance of cable and load C
L
. The frequency is given by, [4]
1
1
1.8
(2 )
S L
K
S L
f MHz
C C
L
C C
t
= ~
+
2  4
The first peak of oscillation is called suppression peak, as shown in lower left trace of
Figure 25, the value is given by, [13]
5.96
ch
I A =
2  5
12
( ) 2691.9
K
S P ch S L
S L
L
U I C C V
C C
= + =
2  6
The above HF voltage oscillation will be damped in short time interval. The
frequency of next voltage oscillation is much lower and can be defined by, [4]
2
1
4.6
2
L L
f kHz
L C t
= ~
2  7
It represents the natural frequency of the load, as shown in lower trace of Figure 24
and upper left trace of Figure 25.
Figure 2  6 Phase current in VCB under second situation
As time goes by, the TRV catches up the dielectric strength of vacuum gap at 8.25ms,
and then the first reignition will occur, as shown in lower right trace of Figure 25. As
a direct result of reignition, the HF current is injected into network and two HF
oscillations can be observed, as shown in Figure 26. The First HF oscillation is due
to the interaction of C
S
and L
S
, as shown in lower left trace of Figure 25. The value of
frequency can be defined by, [4]
3
1
50
2
S S
f MHz
L C t
= ~
2  8
13
This HF current is quickly damped and is not interrupted in its zero point by the VCB.
The second HF current is caused by the interaction of L
K
and C
L
, as shown in lower
right trace of Figure 26. The value of frequency can be defined by, [4]
4
1
0.25
2
K L
f MHz
L C t
= ~
2  9
The quenching capability of HF current of VCB plays an important role in
interrupting the second HF current. When the rate of rise of this HF current is lower
than the quenching capability of VCB in one of its zero points, the HF current is
interrupted and the TRV appears again, as shown in upper trace of Figure 27. After
multiple reignitions have occurred, the TRV is built to a high level, while the
dielectric strength of vacuum gap is also increased due to the movement of contacts.
Once the dielectric strength could withstand the peak of TRV, the load current will be
successfully interrupted. Besides above two situations, the third and the fourth
situations are not observed and recognized in test circuit with different combinations
of parameters of Siemens medium voltage VCB.
Figure 2  7 Upper trace: zoom in the first and the second reignitions of Figure 25; Lower trace:
zoom in the HF current of Figure 26
14
2.6.3 Sensitivity Analysis of VCB
As has been discussed in the previous sections, the parameters of VCB are chosen
randomly with normal distribution from information of vendor. Hence, it is very
important to examine and verify the reignition behavior of different parameters of
VCB. The test circuit is the same as shown in Figure 23. In this part, the first check is
performed by changing the recovery of dielectric strength while keeping the
quenching capability of HF current constant. Then, the quenching capability of HF
current is changed, while the other parameters are kept constant. Finally, fortyfive
different combinations of VCB parameters are analyzed. The performance is recorded
based on the random switching operations in three different arcing time intervals.
If the recovery of dielectric strength is increased from 20kV/ms to 30kV/ms, the
numbers of reignitions are decreased, as shown in Figure 28. Obviously, the reason
for this phenomenon is that the VCB with higher recovery of dielectric strength can
build up the higher dielectric strength in short time to withstand the TRV.
Figure 2  8 Voltage across the contacts of VCB; Left trace: dielectric strength = 20kV/ms,
quenching capability = 0.5kA/s; Right trace: dielectric strength = 30kV/ms, quenching capability
= 0.5kA/s
As shown in Figure 29, the higher quenching capability of HF current, the more
voltage spikes across the contacts of VCB. The effect of spike reveals a potential risk
in that if the VCB has lower quenching capability of HF current, it may experience
more failures of interruption. If the quenching capability is decreased from 0.7kA/s
to 0.5kA/s, the HF current will be successfully interrupted at the tenth HF current
zero point instead of the ninth. It is because the VCB is waiting for the HF current
acquires enough damping to make the rate of rise of HF current is lower than the
quenching capability of VCB,
15
Figure 2  9 Left column traces: voltage across the contacts of VCB and phase current in VCB,
dielectric strength = 30kV/ms, quenching capability = 0.5kA/s; Right column traces: voltage
across the contacts of VCB and phase current in VCB, dielectric strength = 30kV/ms, quenching
capability = 0.7kA/s
The accurate performance of VCB is difficult to simulate, the reasons are not only the
confidential information of vendor, but also the limitation of experiment. The model
of VCB in this chapter is based on the general parameters of Siemens medium voltage
VCB. Therefore, the sensitivity analysis of parameters is important to define the trend
which would lead to the serious multiple reignitions and voltage escalation in the
VCB. Before starting the analysis, the boundaries of simulations should be defined.
First of all, three arcing time intervals are chosen: 0150s, 150300s and 300600s.
Each opening time of VCB is generated randomly by uniform distribution. Then, as
shown in Appendix, the second boundary is about the combination of parameters,
totally fortyfive combinations of parameters are simulated within three arcing time
intervals. After each simulation is finished, the peak voltage across the contacts of
VCB is recorded and the final performance of VCB with different parameters is
drawn on Figure 210.
As mentioned in previous sections, the VCB has risks to get multiple reignitions and
voltage escalations during switching operation. The worst case occurs in the
combination of middle part of quenching capability of HF current and minimum part
of recovery of dielectric strength. Moreover, when do the comparison of parameters in
Figure 210, it can be concluded that the quenching capability of HF current has less
effect on the multiple reignitions and voltage escalations. Furthermore, the shorter
16
arcing time, the more serious multiple reignitions and the voltage escalations.
Consequently, the arcing time and the recovery of dielectric strength play a dominant
role in the multiple reignitions and the voltage escalations of VCB. The worst
combination of parameters will be used in the next several chapters.
Figure 2  10 Reignition behavior of Siemens medium voltage VCB
2.7 Conclusions
In this section, the detail characteristics and phenomena of VCB are discussed. With
the information of Siemens, a vacuum circuit breaker model that includes the current
chopping, the multiple reignitions and the voltage escalations is introduced. An
ohmicinductive load test circuit is given to perform the sensitivity analysis with
different combinations of VCB parameters. Moreover, the result of sensitivity analysis
has shown that the arcing time and the recovery of dielectric strength play a dominant
role in the estimation of multiple reignitions and overvoltage, whereas the quenching
capability of HF current has less effect. Consequently, if the Siemens medium voltage
VCB is used to analyze the overvoltage due to switching operation in power system,
the combination of middle part of quenching capability of HF current and minimum
part of recovery of dielectric strength imply that the worst multiple reignitions and
overvoltage could occur. In order to give the most suitable protective scheme of
overvoltage, the VCB parameters that could generate the worst case will be used in
the later chapters.
17
Chapter 3 Modeling of Different Equipments in
Power System
3.1 Modeling of Cable in ATPEMTP
3.1.1 Introduction
In order to acquire an accurate simulation result of overvoltage, adequate estimation
of cable is needed in the simulation. The purpose of this chapter is to accurately
model cable in ATPEMTP and make cable response in a large frequency spectrum.
Several models of cable are available in ATPEMTP, such as PI, JMarti and Semlyen
[3]. In this study, the JMarti model is chosen, because it could represent the
performance of cable best in a wide range of frequency [3]. The following basic
equations show the electrical characteristics of insulated cable, [16]
( ) ( ) ( ) Z R j L e e e e = +
3  1
( ) ( ) ( ) Y G j C e e e e = +
3  2
Where,
R: the series resistance per unit length of cable system;
L: the series inductance per unit length of cable system;
G: the shunt conductance per unit length of cable system;
C: the shunt capacitance per unit length of cable system;
The electrical frequency is the variable that makes the above quantities are
calculated as the function of frequency. Both Z and Y are calculated by using the
Cable Constants with information of geometry and material properties from CCI cable
datasheet and IEC 60092354 standard. In general, the information below must be
specified, [3, 16]
Geometry of Cable System:
 Position of each cable (XY coordinates);
 Diameters of conductor and thickness of different layers inside the cable;
Material properties:
 Resistivity and relative permeability of conductor;
 Resistivity and relative permeability of surrounding material;
 Relative permittivity of insulating material;
18
The model of cable is achieved by ATPEMTP which does not consider the effect of
stranded conductors, inner and outer semiconducting layers and metal screen of
insulation. Therefore, an approximate method is combined with Cable Constants to
take account of the effect of stranded conductors and semiconducting layers in the
model of cable. The CCI XAI Type O 8.7/15kV 1300mm
2
medium voltage single
core power cable of marine and offshore application is modeled as an example in this
chapter and will be used in the later studies.
3.1.2 Geometry of Cable
The typical CCI XAI single core cable consists of conductors, different layers, metal
screen, sheath and armor, as shown in Figure 31. The conductor is made of tinned
flexible compact copper, and it is a kind of stranded conductors. The insulation is one
of the most important parts of cable. It is used to prevent the connection between the
conductor and the sheath. Therefore, it should have the capacity to withstand the
electrical field of cable for both steady and transient states. Hard ethylene propylene
rubber (HEPR) is the insulation of CCI single core cable, and it has characteristics
which is similar to cross the linked polyethylene (XLPE). The semiconducting layers
are placed between the conductor and the insulation, and also between the insulation
and the metal screen to prevent the partial discharge. The major purpose of metal
screen is to avoid the electrical field outside the cable, provide the return path for
charging current and conduct the fault current to earth. The metal screen is the
individual copper tape. The material of inner and outer sheath of cable is SHF1
compound. Based on the IEC 60092359 standard, the SHF1 compound is the halogen
free thermoplastic. This kind of material has very good performance for situations
with low smoke emissions, low generation of corrosive and toxic gases and low fire
propagation characteristics. The copper wire braid armor is placed between the inner
sheath and the outer sheath to provide mechanical protection to the cable.
The geometry of cable is based on the information of CCI cable datasheets. The
geometric data of cable is shown in Table 31.
Table 3  1 Basic geometric data of cable
Construction
(mm
2
)
Conductor
Diameter
(mm)
Diameter Over
insulation
(mm)
Under Armour
Diameter
(mm)
Overall
Diameter
(mm)
1300
21.7
33.0
37.0
44.0
The information above is the basic characteristics of cable configuration. The real
structure of cable is very complicated and some compulsory parameters of simulation
should be calculated further.
19
Figure 3  1 Cross sectional structure of cable
1
r : Outer Radius of Core
(= Inner Radius of Conductor Semiconducting Layer)
2
r : Outer Radius of Conductor Semiconducting Layer
(= Inner Radius of Insulation)
3
r : Outer Radius of Insulation
(= Inner Radius of Insulation Semiconducting Layer)
4
r : Outer Radius of Insulation Semiconducting Layer
(= Inner Radius of Metal Screen)
5
r : Outer Radius of Metal Screen
(= Inner Radius of Inner Sheath)
6
r : Outer Radius of Inner Sheath
(= Inner Radius of Armor)
7
r : Outer Radius of Armor
20
(= Inner Radius of Outer Sheath)
8
r : Outer Radius of Outer Sheath
(= Overall Radius of Cable)
1
T : Thickness of Conductor Semiconducting Layer
2
T : Thickness of Insulation
3
T : Thickness of Insulation Semiconducting Layer
4
T : Thickness of Metal Screen
5
T : Thickness of Inner Sheath
6
T : Thickness of Amor
Based on the IEC 60502.21998 standard, the nominal thickness of hard ethylene
propylene rubber (HEPR) insulation of 1300mm
2
8.7/15kV cable is 4.5mm. Hence,
the T
1
and r
2
can be calculated by the equations,
1 3 2 1
16.5 4.5 10.85
1.15
T r T r
mm
=
=
=
3  3
2 1 1
10.85 1.15
12
r r T
mm
= +
= +
=
3  4
The IEC 60502.21998 and IEC 60092354 standards give the specific information of
thickness for the copper tape and the inner sheath. Besides these relevant standards,
datasheets of several manufacturers such as Prysmian, UC and TMC use the typical
values of thickness of copper tape and inner sheath for medium voltage power cable.
Therefore, the copper tape with thickness of 0.1mm and the inner sheath with
thickness of 1.4mm are chosen from the general values of manufacturers. The r
3
and
r
4
can be calculated,
3 2 2
16.5
r r T
mm
= +
=
3  5
21
4 6 4 5
18.5 0.1 1.4
17
r r T T
mm
=
=
=
3  6
The thickness of armor is chosen based on IEC 60092354 standard. The IEC
60092354 standard requires that if the fictitious diameter under armor is larger than
30mm, the nominal diameter of the braid wire should be 0.4mm as a minimum.
Besides the IEC standard, one of the major marine and offshore cable manufacturers
TMC uses 0.4mm as diameter of braid wire. Therefore, 0.4mm is chosen as the
approximate diameter of braid wire due to the information of standard and cable
manufacturer. At each cross of braid wire, the two individual wires are superposed
with each other. Consequently, the thickness of armor can be approached by doubling
the diameter of braid wire,
6
0.8 T mm =
3  7
The r
5
, r
6
and r
7
can be calculated by equations,
5 4 4
17 0.1
17.1
r r T
mm
= +
= +
=
3  8
6 5 5
17.1 1.4
18.5
r r T
mm
= +
= +
=
3  9
7 6 6
18.5 0.8
19.3
r r T
mm
= +
= +
=
3  10
The above geometric information is used in the cable model of ATPEMTP. The
material properties of cable will be explained in the next section.
22
3.1.3 Material properties
3.1.3.1 Core of Cable
In ATPEMTP, only solid and hollow conductor can be directly modeled [3]. However,
the design of CCI cable core is stranded. Therefore, the resistivity of core material
should be adjusted to perform the effect of small space between each strand. The
resistivity of common conductor material is shown in Table 32,
Table 3  2 Resistivity of different metal materials
Material Copper Aluminum Lead Steel
( m)
1.7210
8
2.8310
8
2.210
7
1.810
7
The equation of references [9, 10] provide one possible method to adjust the
resistivity of material of stranded core,
2
' 1
2
3
8
6
8
10.85
( )
10
1.72 10
300
10
2.12 10
c c
c
r
A
m
t
t
=
= O
3  11
Where,
A
c
: the efficient (nominal) cross sectional area of the core;
r
1
: the outer radius of core;
c
: the resistivity of material of conductors. In this study, the copper of conductor is
chosen based on the datasheet of CCI cable;
3.1.3.2 Semiconducting Layer and Surrounding Insulation of Cable
The different semiconducting layers of cable cannot be automatically included in the
cable model of ATPEMTP. However, the effect of different semiconducting layers
can be considered through changing the electrical permittivity of insulation and
keeping the capacitance between conductor and metal screen constant [16]. Hard
ethylene propylene rubber is the insulation of cable and the relative permittivity is
around 3.2. The conversion procedure is shown in the equation, [16, 17]
23
4
1
0
10
12
ln( )
2
17
ln( )
10.85
4.4 10
2 8.85 10
3.55
r
r
r
C c
tc
t
=
=
=
3  12
Where,
C: the capacitance of cable which is obtained from datasheet of CCI cable;
0
: 8.8510
12
F/m;
In the above conversion procedure, the semiconducting layers are represented as part
of insulating materials of main insulation, consequently the main insulation of cable
will extend to the conductor and the metal screen. As a result of this procedure, it
makes the electrical permittivity increase to leave the capacitance unchanged [17].
Besides the relative permittivity of insulation, the relative permittivity of surrounding
materials such as the inner sheath and the outer sheath is also required for the cable
model in ATPEMTP. As mentioned above, the material of inner sheath and outer
sheath is SHF1. The SHF1 is a kind of halogen free thermoplastic with the ethylene
vinyl acetate (EVA) as the essential polymer to make such thermoplastic [18].
Therefore, the relative permittivity of EVA is used to approach the material of SHF1
and its typical value is around 2.8 [19]. The relative permeability of conductor,
insulation and surrounding material is equal to 1 [8].
3.1.4 Sensitivity Analysis of Semiconducting Layer
The completed model of 1300mm
2
single core cable system is shown in Figure 32
and the length of each cable is 180m.
Figure 3  2 Cable model in ATPEMTP
24
This length refers to the distance between the switchgear and KMT7111 main gas
compressor A motor. Using the open circuit test [20], a step voltage excitation is
applied on the sending end of cable system and the response of voltage excitation is
simulated on the receiving end of cable system, as shown in Figure 33. The metal
screen and the armor of each cable are grounded during the open circuit test.
Figure 3  3 Open circuit test of cable system
Figure 3  4 Response of open circuit test of cable system
The response of Figure 34 shows that the semiconducting layers make the
propagation speed decrease, because the HF transients of cable propagate as
decoupled coaxial wave between core and metal screen. Consequently, the modeling
of core, insulation, semiconducting layers and metal screen determine the transient
behavior of cable. If the thickness of insulation is fixed, adding semiconducting layers
will increase the inductance of conductor and metal screen loop without changing the
capacitance. This will lead to decrease propagation speed and increase surge
impedance [17].
25
3.2 Modeling of Generator and Busbar
The selected FPSO vessel has four synchronous generators and each generator is
driven by a gas turbine. During normal production, three of them work together to
feed loads and one of them is used as cold redundancy. The generator is modeled as
an ideal voltage source in series with its subtransient impedance [7], as shown in
Figure 35. The effect of saturation, excitation, governor system and mechanical part
are not included.
Figure 3  5 Simplified model of generator
The subtransient impedance is modeled by the symmetric resistance and inductance
coupled line: Line_SY3. When the Line_SY3 model is used to represent the
subtransient impedance of generator, the zero sequence impedance and the positive
sequence impedance of generator are required. C
s
is the bushing capacitance which is
between the generator windings and the ground. The values of different parameters
are shown in Table 33, and all of these values are from the vendor test report.
Table 3  3 Electrical data of generator
Parameter Unit Value
Rated Voltage kV 11
Rated Apparent Power kVA 29070
Power Frequency Hz 60
Base Impedance 4.16
R0 0.1249
X0 0.3746
R1 0.0558
X1 0.8408
C
s
F 0.1
26
3.3 Modeling of Busbar
Rectangular busbars are placed in the high voltage switchgear of the selected FPSO
vessel. The length of each separate busbar is around 85cm to 100cm. The transient
response of busbar is similar to the short length of overhead transmission line because
they have similar environment of installation. Therefore, in the switching transient
analysis, the busbar can be modeled by the transmission line with specific
characteristic impedance [8].
In ATPEMTP, the busbar can be accurately modeled by the JMarti frequency
dependent model and equivalent PI model. Both of them require the radius and
positions of busbars conductors in XY coordinate, as shown in Figure 36.
Figure 3  6 Model of busbar in ATPEMTP
The cylindrical conductors are used to approach the performance of rectangular
conductors. Because the impedance of rectangular structure is complicated to estimate
and two models of ATPEMTP are unable to modify the conductors with rectangular
structure. The cross sectional area of each cylindrical conductor is determined by the
rated current. The rated current of each busbar conductor is 4000A and the cross
sectional area of each conductor is around 25.87cm
2
. The position of each conductor
is measured from the design report of switchgear. In this study, the PI model is used
because if the busbar is very short such as 100cm or below, the JMarti model has the
internal numerical problem. In fact, this numerical problem can be solved through
decreasing the time step of simulation, but it requires more than one gigabytes
memory space of computer hard disk and causes overflow of ATPEMTP.
3.4 Conclusions
This chapter has described the modeling of cable, generator and busbar. The cable is
modeled by JMarti frequency dependent model in ATPEMTP, which includes the
effect of stranded conductors and semiconducting layers. The resistivity of stranded
conductor is slightly increased due to the small space between each strand. The
response of open circuit test shows that the propagation speed is decreased by the
semiconducting layers.
27
The model of generator is represented by an ideal voltage source behind its
subtransient impedance without considering the mechanical part, the excitation and
the governor systems. The busbar is treated the same to the short length overhead line.
The cylindrical conductors that have the same rated current are used to approach the
effect of rectangular busbar by PI model.
28
Chapter 4 Surge Protection Device
4.1 Introduction
The overvoltage of power system can be limited by using surge protection devices. In
general, the protection devices can be divided into two groups. All kinds of surge
arresters belong to the first group. The second group of surge protection device
consists of surge capacitors and CR suppressors. The method of categorization is
based on the protective way [1]. The surge arrester can be used to limit the amplitude
of different kinds of overvoltages such as the switching overvoltage and the lightning
overvoltage, however it cannot modify the rate of rise of voltage surge. The function
of surge capacitor is to reduce the rate of rise of voltage surge and decrease the rate of
rise of TRV to decrease reignitions. Therefore, both groups of the protection devices
could provide suitable surge protection to electrical equipments such as motors and
transformers. In this study, the surge arrester is modeled and used in later chapters,
because the experience shows that when the overvoltage is generated by switching
operation, the limitation of amplitude of overvoltage could provide better protection
to equipments [8].
4.2 Impulse function for Current Source
The electrical response of surge arrester model should be equal or similar to the data
that is suggested by the manufacturer. In order to check the response of surge arrester
model, the different current impulse waves should be described first. In general, the
impulse wave is determined by its front time T
f
and its time to half peak value T
h
, as
shown in Figure 41.
29
Figure 4  1 Current impulse wave 1/5s
The front time T
f
is given by the following equation, [8]
0 . 9 0 . 3
0 . 6
f
T T
T
=
4  1
Where,
T
0.9
: the time when the wave reaches 90% of the peak value;
T
0.3
: the time when the wave reaches 30% of the peak value;
The time to half peak value T
h
is determined by, [8]
0.9 0.3
0.3
2
h c
T T
T T T
= +
4  2
Where,
T
c
: the time when the wave reaches to the half peak value in the wave tail;
The time T
r
can be calculated by following equation, [8]
0.9 0.3
0.3
2
r
T T
T T
=
4  3
Originally, the above method is applicable for determination of voltage impulse wave.
The time T
r
of current impulse wave is determined by the trace that passes through 90%
30
and 10% of peak value rather than 90% and 30% of peak value. However in order to
simplify the work, the current impulse wave is tackled the same to the voltage
impulse wave [8]. In ATPEMTP, the different current impulse waves can be
accurately generated by Heidler current source, which is shown in the following
equation, [3]
_
_
( )
( ) exp( )
1 ( )
n
f
Peak
n
f
t
T
t
f t I
t
tau
T
=
+
4  4
Where,
I
peak
: the peak value of current impulse;
n: the parameter determines the rate of rise of function. It is ranging from 1 to 10;
T
_f
: the constant determined by front time of wave;
tau: the constant determined by tail time of wave;
The reference [6] proposed the calculated parameters for Heidler function with
different current impulse waves. The optimized parameters are used to generate three
typical current impulse waves: the fast front wave 1/5s, the lightning impulse wave
8/20s and the switching impulse wave 30/60s, as shown in Table 41. The peak
value of the current impulse is chosen from datasheet of ABB MWD surge arrester.
Table 4  1 Parameters of Heidler function for different current impulse waves
Impulse
(s/s)
Peak Value of
Current
impulse (A)
n T
_f
(s) tau (s)
1/5 1000 5 2.07 4.62
1/5 5000 5 2.07 4.62
1/5 10000 5 2.07 4.62
8/20 1000 5 16.55 9.62
8/20 5000 5 16.55 9.62
8/20 10000 5 16.55 9.62
8/20 20000 5 16.55 9.62
30/60 125 5 56.7 17.5
30/60 250 5 56.7 17.5
30/60 500 5 56.7 17.5
30/60 1000 5 56.7 17.5
31
4.3 Modeling of Surge Arrester
The amplitude of overvoltage in power system can be limited by surge arrester. The
surge arrester provides a temporal path to earth which the superfluous charge is
removed from the system. When the power system is working under normal voltage, a
very small or no leakage current is conducting in the surge arrester. However, if the
overvoltage appears due to the different operations of power system, the surge arrester
can conduct the current without causing fault [21]. The following typical surge
arresters are used,
Open Spark Gap.
The open spark gap is an early overvoltage protection device. It is simple and cheap,
however when it works, a short circuit appears in the power system due to an arc is
conducting between the line and the earth.
SiC Surge Arrester
Silicon carbide (SiC) surge arrester is an improvement of surge arrester. The spark gap
is in series with nonlinear resistance which is made of the stacks of silicon carbide
discs. The spark gap gives high impedance during normal voltage. If the overvoltage
occurs and exceeds the flashover voltage of the spark gap, a current will conduct in
nonlinear resistance, where the surge energy is dissipated and overvoltage is limited
[21].
Metal Oxide Surge Arrester
Metal oxide surge arrester is a very important improvement of SiC surge arrester,
which does not have a spark gap. The major material of metal oxide surge arrester is
the mixture of zinc oxide and additional metal oxide such as MnO [8]. Due to the
large nonlinear voltampere characteristic, a very small leakage current flows through
the surge arrester during normal voltage, whereas the overvoltage impulses can be
diverted with current surges of many kA [21].
As mentioned above, the metal oxide surge arrester has a good performance for surge
protection. Therefore, the 11.3kV ABB metal oxide surge arrester is modeled and used
to protect the electrical equipments in power system of the FPSO vessel. The
measured voltampere characteristic of 11.3kV surge arrester is shown in Fig 42.
32
Figure 4  2 VI characteristic of 11.3kV surge arrester
With the information of voltampere characteristic of surge arrester, the reference [9,
22] have proposed two different models of surge arrester which can be applied in
ATPEMTP. But all of these models require additional information such as
temperature dependent resistance and resistanceinductance filter. The reference [14]
has proposed a simplified Schmidt surge arrester model which is also used in this
study, as shown in Fig 43.
Figure 4  3 Simplified Schmidt surge arrester model
Where,
R(i): the current dependent resistance which is used to represent the non linear
voltageampere characteristic of metal oxide surge arrester;
C
a
: the capacitance of surge arrester block.
R
a
and L
a
: the physical characteristics of ZnO grains;
The parameters of surge arrester are estimated to be: R
a
= 0.06, L
a
= 0.5H and C
a
=
0.14nF based on the reference [14]. The voltampere characteristic of R(i) is chosen
from Figure 42. The residual voltages of simulations are compared with the data
sheet of ABB 11.3kV surge arrester, as shown in Table 42.
33
Table 4  2 Comparison of residual voltage between simulation and data sheet
Impulse (s/s)
Peak Value of
Current impulse
(A)
Peak Value of
Residual Voltage
(kV) (Simulation)
Peak Value of
Residual Voltage
(kV) (Data Sheet)
1/5 1000 23.97 
1/5 5000 29.07 
1/5 10000 33.86 
8/20 1000 23.87 23.4
8/20 5000 27.54 26.1
8/20 10000 29.81 27.6
8/20 20000 33.49 30.6
30/60 125 21.11 20.3
30/60 250 21.98 21.4
30/60 500 22.89 22.1
30/60 1000 23.86 23.0
The residual voltages of simulations under different current impulse waves are shown
from Figure 44 to Figure 46.
34
Figure 4  4 Upper trace: 1/5s current waves with different amplitudes; Lower trace: response of
surge arrester after excitation by 1/5s current waves
35
Figure 4  5 Upper trace: 8/20s current waves with different amplitudes; Lower trace: response of
surge arrester after excitation by 8/20s current waves
36
Figure 4  6 Upper trace: 30/60s current waves with different amplitudes; Lower trace: response
of surge arrester after excitation by 30/60s current waves
As shown in above Figures, the residual voltage of surge arrester has relationship with
the shape of current impulse and the peak value of current impulse. If the front
steepness of current impulse or the peak value of current impulse is increased, surge
arrester experiences higher residual voltage. Due to that the model of surge arrester
includes the inductance of surge arrester, the peak value of current impulse is lagging
to the residual voltage. Table 42 has shown that the model of surge arrester has a
good tolerance for the 30/60s switching current impulse, however it has slightly
higher deviation for the current impulse with shorter front time such as the 8/20
lightning current impulse. This because that the model of surge arrester does not
consider the hysteresis of voltampere characteristic [8]. In this study, the major
37
purpose is the analysis of switching operation of vacuum circuit breaker. Therefore, it
is sufficient to use this model of surge arrester, because the front time of the reignition
current impulse is not as short as that of the lightning current impulse [8].
4.4 Conclusions
In this chapter, the concepts, advantages and disadvantages of different surge arresters
are compared first. Then the steep front, the lightning and the switching current
impulse is calculated by the Heidler function in ATPEMTP. Based on several
references, a simplified Schmidt surge arrester model is used to model the ABB
MWD 11.3kV surge arrester. The simulated response of surge arrester model is
compared with the datasheet of ABB and a good tolerance is achieved. This model
of surge arrester will be used in later chapters.
38
Chapter 5 Modeling of Induction Motor
5.1 Introduction
Induction motor is one of the most important electrical loads to allow continued
processing in a FPSO vessel. Based on the driven machines such as compressor and
pump, the rated power of high voltage induction motors are ranging from several
hundreds of kW to several MW. In this study, the high voltage induction motor is
modeled according to three different switching operations. The first one is the
switching operation of motor under starting condition. This situation is usually from
the inappropriate settings of protection relays. Next situation is the switching
operation of motor under full load condition. It may occur due to the process trip. Last
situation is the switching operation of motor under light load condition. It is the most
common situation which disconnects the induction motor from the power system in
the FPSO vessel.
Therefore, the accurate model of induction motor is necessary for taking transient
analysis. The distribution of overvoltage stress inside the motor winding is not
analyzed in this study, because the major purpose of this study is the determination of
switching transient overvoltage in high voltage side of the selected FPSO power
system and the influence of surge arrester.
5.2 Modeling of Induction Motor under Starting Condition
The disconnection of motor under starting condition can cause more severe switching
transient overvoltage than the motor under full and light load working conditions.
When the motor is starting up, the rotor does not acquire high speed to generate
enough back electromotive force (EMF) which is opposite to the source voltage. The
low back EMF cannot keep the TRV at a low level after opening the contacts of VCB.
Therefore, the modeling of induction motor under starting condition can be regarded
as doing switching operation under the situation where the rotor is locked. As shown
in Figure 51, the paper [5] proposed the Tequivalent circuit of induction motor to
represent the electrical characteristics of motor under starting condition.
39
Figure 5  1 Tequivalent circuit or induction motor
Where,
R
1
and X
1
: the impedance of stator;
R
2
and X
2
: the impedance of rotor;
X
m
: the magnetizing reactance;
s : the slip of motor;
V: the phase voltage;
In the equivalent network, the slip of motor is equal to 100%. Although the equivalent
network is simple and useful, it does not consider the bushing capacitance for each
phase and the natural oscillation of motor windings. In order to fix this problem, the
modified equivalent network is proposed and shown in the Figure 52,
Figure 5  2 Modified Tequivalent circuit of induction motor under starting condition
Where,
R
1
, X
1
, R
2
and X
2
: the impedance of motor for each phase;
C
g
: the bushing capacitance of each phase;
R
d
: the damping resistance of motor winding for each phase;
In the modified circuit, the stator magnetizing reactance X
m
is ignored, because X
m
is
larger than the combination of impedance of rotor during locked rotor condition or
starting condition. The values of R
1
, X
1
, R
2
and X
2
are calculated from the no load and
the locked rotor test reports of vendor [23]. C
g
is the bushing capacitance that comes
from the electrical study of project and the information of vendor. R
d
is the damping
40
resistance that is given by the equation, [24]
1
2
( )
d
m
g
m
R
R
C
L t
=
5  1
Where,
R
m
: the total resistance of each phase;
L
m
: the total inductance of each phase;
: the attenuation time constant of natural oscillation of load itself and is related to the
amplitude factor
of recovery voltage by the equation, [24]
2
( ) 1
1
ln( )
1
m m
L C
t
t
 

= +


\ + .
5  2
Where,
The typical value of amplitude factor
of motor is 1.5;
5.3 Evaluation Circuit and Simulation Results of Motor
under Starting Condition
The Figure 53 shows the evaluation circuit that is used to analyze the switching
transient overvoltage for a typical load in the selected FPSO vessel: main gas
compressor A motor KMT7111. Three main generators are connected to the busbar
through 4(31300mm
2
) cable system, which means that each phase has four
8.7/15kV single core cables. The KMT7111 motor is connected to the VCB via
2(31300mm
2
) cable system. The modeling of generator, the cable and the busbar
is explained in previous chapters.
41
Figure 5  3 Evaluation circuit of induction motor
In this case, the VCB is used to disconnect the motor from locked rotor condition with
rated voltage 11kV to approach the starting condition. It means that after the motor is
connected to the power system for an extremely short time period, the rotor can be
regarded as stalled due to the fact that the mechanical load, and then, the terminal
voltage of motor is still around the rated voltage. Under such circumstance, the VCB
performs the switching operations in different arcing time intervals to get comparable
switching overvoltages. Before making the explanation of three typical phenomena of
switching overvoltages, the electrical characteristics of starting motor is shown in
Figure 54, three key characteristics are compared with the test report of vendor in
Table 51. The phase voltage, the phase current and the power factor show an
acceptable tolerance based on the IEC 60092101 and the IEC 60034.012010
standards. The deviation of these characteristics comes from the parameters of
equivalent network.
Figure 5  4 Simulated voltage and current of Phase A of KMT7111 motor under starting
condition
42
Table 5  1 Comparison of characteristics of motor under starting condition between simulation
and data sheet
Item
Results of
Simulation
Results of Test Report
and Datasheet
Tolerance of
IEC 60092101
IEC 60034.012010
Peak Value of
Phase Voltage (V)
9016.1 8981.5 +6% to 10%
Peak Value of
Phase Current (A)
2918.5(3.29p.u) 2905.1(3.28p.u) +20%
Power Factor
under Starting
Condition
0.14 0.12
Maximum absolute
value: 0.07
The first switching operation is performed by setting the arcing time around 850s. In
this case, the phase A of KMT7111 motor is the first pole to clear. Figure 55 shows
that the maximum switching overvoltage of terminal of motor is almost 12kV (1.3pu)
in phase A. Another important factor is that the whole switching operation does not
have any reignitions, this can be explained with cooperation of Figure 56, which
shows the voltage across contacts of phase A of VCB. When the contacts of VCB are
opened, around 3ms, the increase of dielectric strength is always larger than the
increase of TRV. Therefore, the VCB could withstand the TRV and make successful
interruption without reignitions.
Figure 5  5 Simulated three phase voltage in terminal of KMT7111 motor; (Red trace: phase A;
Green trace: phase B; Blue trace: phase C)
43
Figure 5  6 Simulated voltage across the contacts of phase A of VCB
Besides the successful interruption without reignitions, if the arcing time is decreased
from 850s to approximately 350s and while the other parameters of circuit are kept
constant, the switching overvoltage in terminal of motor are increased drastically. It is
also the worst situation that could occur during a switching operation based on Figure
57. The peak value of overvoltage could reach to around 56kV (6.2p.u). Such a high
overvoltage could give a large stress to the insulation system of motor and decrease
the life time of motor. Besides the stress of overvoltage, the other special scenario is
shown on the Figure 58. Based on the explanation of previous chapters, the current of
phase A reaches to the chopping level before the other two phases, therefore, the
current of phase A should be interrupted first. However, the simulation shows the
totally different results where the current of phase C is interrupted before phase A and
phase B. Both of the above situations are the results of virtual current chopping.
44
Figure 5  7 Upper left trace: simulated three phase voltage in terminal of KMT7111 motor;
Upper right trace: simulated voltage of phase A in terminal of KMT7111 motor; Lower left trace:
simulated voltage of phase B in terminal of KMT7111 motor; Lower right trace: simulated
voltage of phase C in terminal of KMT7111 motor
Figure 5  8 Upper left trace: simulated three phase current in VCB; Upper right trace: simulated
current of phase A in VCB; Lower left trace: simulated current of phase B in VCB; Lower right
trace: simulated current of phase C in VCB
45
The virtual current chopping may occur if the first pole to clear has multiple
reignitions before the other two phases are interrupted [25]. The reignition of one
phase could cause the HF current flow into the other two phases through the electrical
coupling of load. The evaluation circuit of Figure 53 is simplified to represent the
path of HF current and is shown in Figure 59,
Figure 5  9 Simplified evaluation circuit for the explanation of virtual current chopping
After the VCB is opened, the current of phase B and C are far from the chopping level
and the power frequency current is still conducting through the arc. Phase A has
multiple reignitions during switching operation, as shown in Figure 510. The HF
current I
t
in phase A flows to ground through the bushing capacitance C
g
, and then I
t
could be divided into two parts while each part goes back to phase B and C via the
bushing capacitance C
g
, as shown in Figure 59.
Figure 5  10 Simulated voltage across the contacts of phase A of VCB
46
Figure 5  11 Simulated three phase current in VCB; (Red trace: phase A; Green trace: phase B;
Blue trace: phase C)
The HF currents of phase B and C are superimposed on the power frequency current
and both the HF currents of phase B and C have the same polarity and magnitude, but
the polarity is opposite to the HF current in phase A. If the peak value of HF current
in phase B or C is larger than the peak value of power frequency current, the HF
current could force the power frequency current to zero, as shown in Figure 511. This
unnatural current zero phenomenon is named virtual current chopping. Compared
with the normal level of current chopping 6A for a modern VCB, the virtual chopping
level could reach to several hundred amperes and lead to larger overvoltage than the
normal case.
After the definition of virtual current chopping, Figure 57 and Figure 58 can be
reviewed. After around 20 reignitions in phase A, the first virtual current chopping is
observed in phase C due to the HF current flows through the capacitive couplings in
the load side, as shown in lower right trace of Figure 511. Then, the HF current of
phase C is temporarily interrupted and TRV with the very steep du/dt appears between
the contacts of phase C in VCB, as shown in lower right trace of Figure 512.
47
Figure 5  12 Upper trace: simulated voltage across contacts of phase A of VCB; Lower left trace:
simulated voltage across contacts of phase B of VCB; Lower right trace: simulated voltage across
contacts of phase C of VCB
At this moment, the contacts of phase C in VCB do not move to enough distance to
generate a strong dielectric strength which could withstand such high TRV. Therefore,
the first reignition occurs in phase C. If the virtual current chopping occurs several
times, the multiple reignitions appear in phase C until dielectric strength becomes
strong enough, as shown in lower right trace of Figure 512 around 6.5ms. Besides
phase C, the virtual current chopping also skips to phase B and causes the similar
multiple reignitions. After several repetitions of virtual current chopping, a
considerable overvoltage is built up in the motor side. In phase A, the HF current is
unsuccessfully interrupted and the power frequency current is conducting through the
arc until the next zero current point. The reason of unsuccessful interruption is that the
multiple reignitions make the current difficult to interrupt and some points which the
power frequency current is larger than the HF current could be reached. Consequently,
the HF current is unsuccessfully interrupted by the VCB due to the numbers of zero
crossing of HF current becomes less and less, as shown in Figure 513.
48
Figure 5  13 Simulated current of phase A in VCB
If the arcing time is further decreased to around 100s, the peak value of switching
overvoltage could be very moderate such as 14kV (1.6pu) in Figure 514. In fact, the
reason of this special case can be explained from the unsuccessful interruption of
VCB. After the contacts of VCB are opened, phase A is chopped first, as shown in
Figure 515. Due to the short arcing time, the distance between two contacts is too
small to withstand the TRV. After the reignition occurs several times, the VCB fails to
interrupt the HF current in phase A due to the same reason as mentioned above and
the power frequency current takes over until the next current zero point. The
conducting period of HF current is shorter than the previous case and does not cause
the virtual current chopping. Consequently, the switching overvoltage is not as serious
as the previous case.
Figure 5  14 Simulated three phase voltage in terminal of KMT7111 motor; (Red trace: phase A;
Green trace: phase B; Blue trace: phase C)
49
Figure 5  15 Simulated three phase current in VCB; (Red trace: phase A; Green trace: phase B;
Blue trace: phase C)
5.4 Modeling of Induction Motor under Full Load Condition
In ATPEMTP, the Universal Machine Type 3(UM3) is used to model the induction
motor under full load condition, as shown in Figure 516 [3, 6]. The back EMF could
be represented in Type 3 model because this model is based on the dq0 rotor
reference coordination. Unlike the other models, the Type 3 model has not built in a
mechanical part. Therefore, the user should build the mechanical part of motor
separately by different electrical elements, for example the capacitance representing
the moment of inertia and the current source representing the load torque [3].
Figure 5  16 Universal machine type 3 model of ATPEMTP
5.4.1 Optimized Parameters of Motor under Full Load Condition
The UM3 model requires accurate parameters of Tequivalent circuit of motor which
should represent the steady state of running. As has been mentioned in the previous
section, the parameters of Tequivalent circuit are calculated from test reports of
vendor such as the no load test and the locked rotor test based on the reference [23]. It
is a convenient and easy method to calculate parameters that show a good approach
for the electrical characteristics of motor under starting condition, however, it has
limitations for calculating parameters of motor under full load condition.
50
When the induction motor is working under normal running condition, the current of
rotor bar is uniformly distributed and the leakage flux lines are formed as shown in
Figure 517 [26]. If the working condition is changed from running to starting or
locked rotor, the mechanical speed of motor is quite low and even equal to zero. Such
a low mechanical speed results in very high slip and slip frequency in the rotor. Some
squirrel cage induction motors have a deep and narrow rotor bar, then the slot leakage
flux is concentrated on the lower part of rotor bar. During the starting or locked rotor
situation, the reactance of lower part of rotor bar is larger than the upper part of rotor
bar because the flux is not homogeneous distributed, while the current of rotor bar is
mainly flowing through the upper part. Therefore, the current change makes the
reactance of rotor conductor decrease and the resistance of rotor conductor increase.
This impedance change is named as skin effect. The large squirrel cage induction
motors are designed to perform a high starting torque and a low starting current by the
application of skin effect during start up.
Figure 5  17 Flux and current distribution in rotor conductor
The reference [23] has proposed a method to perform the locked rotor test with 25%
rated frequency to avoid the serious skin effect. In this case, the leakage reactance of
rated frequency could be regarded as proportional to the frequency. However, the
locked rotor test of vendor is finished with 100% rated frequency. If the five
parameters of Tequivalent circuit are calculated by the proposed method [23] without
any changes, a large deviation is inevitable. In order to minimize the deviations
caused by the skin effect, the author proposed a modified method to calculate
parameters of Tequivalent circuit of motor under full load condition with minimum
influence of skin effect. Figure 51 shows a typical single phase Tequivalent circuit.
If the induction motor is under no load or locked rotor test, the Tequivalent circuit
could be simplified, as shown in Fig 518,
51
Figure 5  18 Left trace: locked rotor equivalent circuit based on the Figure 51; Right trace: no
load equivalent circuit based on the Figure 51
Under the locked rotor condition, the slip is 100%. The magnetizing reactance X
m
is
parallel by the rotor impedance. Due to that the X
m
is very large, the magnetizing
circuit section could be regarded as open circuit. Therefore, the impedance of motor
under locked rotor condition is,
1
1
BL
V
Z
I
=
5  3
2
1
3
BL
BL
P
R
I
=
5  4
'
1 2 BL
R R R = +
5  5
2 2
BL BL BL
X Z R =
5  6
'
1 2 BL
X X X = +
5  7
Where,
Z
BL
: the locked rotor impedance;
R
BL
: the locked rotor resistance;
X
BL
: the locked rotor reactance;
P
BL
: the absorbed power of motor;
Under no load condition, the circuit section of rotor could be regarded as open circuit
due to that the slip is very small and the rotor resistance is extremely large. Therefore,
the impedance of motor under no load condition is,
52
1
1
NL
V
Z
I
=
5  8
0
2
1
3
NL
P
R
I
=
5  9
2 2
NL NL NL
X Z R =
5  10
1 NL m
X X X = +
5  11
Where,
Z
NL
: the no load impedance;
R
NL
: the no load resistance;
X
NL
: the no load reactance;
P
0
: the no load power of motor;
Based on the equation 57 and 511, the Tequivalent circuit can be modified to
eliminate the magnetizing reactance X
m
and the rotor reactance X
2
, as shown in
Figure 518.
Figure 5  19 Modified Tequivalent circuit of induction motor
The equivalent impedance of modified Tequivalent circuit is,
'
2
1 1
1 1 '
2
1 1
( ) ( )
( ) ( )
NL BL
th
NL BL
R
j X X j X X
s
Z R jX
R
j X X j X X
s
+
`
)
= + +
+ +
5  12
53
The real and imaginary parts of equivalent impedance are,
'
2
2
1
1 '
2 2
2
1
( )
( ) ( 2 )
NL
th
NL BL
R
X X
s
R R
R
X X X
s
= +
+ +
5  13
2
'
2
1 1 1 1
1 '
2 2
2
1
( )( )( 2 ) ( )
( ) ( 2 )
NL BL NL BL NL
th
NL BL
R
X X X X X X X X X
s
X X
R
X X X
s
 
+ +

\ .
= +
+ +
5  14
The no load and the locked rotor reactance could be calculated by the equation 57
and 511. S is the slip of the motor and the stator resistance R
1
is measured by test
report. Besides these known parameters, either real or imaginary part of equivalent
impedance has two variables: the stator leakage reactance X
1
and the rotor resistance
R
2
. The value range of X
1
is from 0 to X
BL
, and from 0 to X
NL
. Due to the reason of
minimum cover (X
BL
<X
NL
), the value of X
1
is chosen between 0 and X
BL
. The
determination of rotor resistance R
2
is cooperated with the theory of skin effect.
Based on the equation 55, R
2
is calculated by the locked rotor test. As a result, the
skin effect plays a dominant role and makes the rotor resistance R
2
larger than the one
in full load condition. In the other word, the value of R
2
should be between 0 and
(R
BL
R
1
) in full load condition. Therefore, the basic idea of modified calculation
method is to find out new parameters of X
1
, X
2
, X
m
and R
2
which could perform the
accurate characteristics of full load condition through adjusting the values of X
1
and
R
2
within their intervals. The flowchart of algorithm is shown in Figure 520.
54
Figure 5  20 Flowchart of modified method
Two variables X
1
and R
2
are set to zero in the initialization. The slip of motor is
chosen from the rated slip, for example the rated slip of KMT7111 motor is 0.76%.
X
NL
and X
BL
are set to values which are calculated based on the equation 57 and 510.
R
1
is equal to the measured value of test report. An identified impedance Z
th_i
is also
arranged in the initial stage before doing calculations. This identified impedance is the
No
No
No
Yes
X
1
X
BL
R
2
R
BL
 R
1
Calculation of R
th
, X
th
,
R
th
, X
th
If R
th
and X
th
(error)
X
1
(j) = X
1
R
2
(j) = R
2
R
th
(j) = R
th
X
th
(j) = X
th
j = j + 1
R
2
= R
2
+ Step size
End
X
1
= X
1
+ Step size
Initial Values (s, R
1
, X
1
, R
2
, X
BL
, X
NL
, R
th_i
, X
th_i
, j)
Yes
Yes
55
equivalent impedance of motor under full load condition and can be estimated by the
equations,
_
rated
th i
rated
V
Z
I
=
5  15
_ 2
3
rated
th i
rated
P
R
I
=
5  16
2 2
_ _ _ th i th i th i
X Z R =
5  17
Where,
V
rated
: the rated phase voltage;
I
rated
: the rated phase current;
P
rated
: the t rated power of motor;
R
th_i
: the real part of identified impedance;
X
th_i
: the imaginary part of identified impedance;
After the initialization stage, the algorithm starts to execute the main function. If the
stator leakage reactance X
1
is smaller than the locked rotor reactance X
BL
, the
algorithm steps into next judgment which is about the rotor resistance R
2
. If the R
2
is
smaller than the difference between the locked rotor resistance R
BL
and the measured
stator resistance R
1
, the first task will be calculated. Besides the R
th
and the X
th
, the
output of the first task also includes the differences between the equivalent impedance
and the identified impedance,
_ th th th i
R R R A = 5  18
_ th th th i
X X X A = 5  19
If both differences are smaller than the defined error , the stator leakage reactance X
1
,
the rotor resistance R
2
and two differences R
th
and X
th
are recorded into the array.
Then, the calculation returns back to the judgment of rotor resistance R
2
with one step
size increased. The calculation is repeated until the rotor resistance R
2
is larger than
the difference between the locked rotor resistance R
BL
and the measured stator
resistance R
1
. Next, the rotor resistance is given to zero and the stator leakage
reactance is increased by one step size to repeat calculations. After the calculation of
two loops is finished, the relationship of R
th
and X
th
are recorded into two arrays.
Finally, If the equivalent impedance R
th
and X
th
calculated based on the parameters of
56
X
1
and R
2
have minimum R
th
and X
th
, the parameters of X
1
and R
2
could be used to
represent the motor under full load running condition. In the other word, the deviation
of parameters due to skin effect could be minimized.
In order to verify the above modified method, the parameters of an Alstom 14MW
induction motor which is used in Capixaba Golfinho FPSO B project are calculated
and compared with the parameters of datasheet. In total, the algorithm finds out
389956 points of R
th
and X
th
which fulfill the requirement of defined error , for
example the is set to 3 for this motor. As shown in Figure 521, the star trace is the
R
th
and the dot trace is the X
th
. If the R
th
reaches to the lowest , the X
th
has the
largest . When the X
th
has the lowest , the of R
th
is located in the middle part.
Consequently, the X
th
with the lowest deviation and the R
th
with middle level of
deviation are chosen to calculate the parameters and the comparison is shown in
Table 52.
Figure 5  21 Values of R
th
and X
th
Table 5  2 Comparison of parameters from different methods
Item R
1
X
1
X
m
R
2
X
2
Trace on Figure
522
Method[18] 0.045 1.39 84.51 0.24 1.39 Blue
Modified Method 0.045 2.07 83.83 0.069 0.708 Red
Information of Data
Sheet
0.045 2.32 92.5 0.074 0.745 Black
57
Figure 5  22 Comparison of torque, current and speed characteristics with different parameters
shown in Table 52
The result of modified method has very closed approach to the information of vendor,
as shown in red trace of Figure 522. However, the blue trace which represents the
parameters from the method [23] gives large deviation for the full load condition,
because the rated torque point is moved to the left. The starting part of torque and
current characteristics are not fully represented, nevertheless the specific starting
model is used to fix this problem.
5.5 Evaluation Circuit and Simulation Results of Motor
under Full Load Condition
When the motor is working under full load condition, the voltage, current, power
factor and mechanical speed should be equal to or around the rated value. The
response of modified parameters and the dq0 model is evaluated by the same circuit
in Figure 53 to check whether the simulation results such as the phase voltage and
the phase current are fulfilling or not. The load is KMT7111 Main Gas Compressor A
Motor and the parameters are calculated by the modified method. The electrical and
the speed characteristics under full load condition are shown in Figure 523 and Table
53.
58
Figure 5  23 Voltage and current of phase A of KMT7111 motor under full load condition
Results of simulation have shown that the modified parameters have an acceptable
tolerance to the response of motor test report. The next procedure is similar to the
motor under starting condition. Typical switching operations are performed to check
the response of overvoltage. Figure 524 shows two cases with different arcing time
and results. The first case is the successful interruption without reignition in the VCB
and the arcing time is around 450s. Second case is the successful interruption with
multiple reignitions and the arcing time is around 100s.
Table 5  3 Comparison of electrical characteristics of KMT7111 motor under full load condition
between simulation and test report
Item
Results of
Simulation
Results of
Test Report
Tolerance of
IEC 60092101
IEC 60034.012010
Peak Value of Phase
Voltage (V)
9015.1 9010 +6% to 10%
Peak Value of Phase
Current (A)
878.1 876.4 
Power Factor under Full
Load Condition
0.9 0.89
Maximum absolute
value: 0.07
Mechanical Speed (rad/s) 187.1 187.07 20% to +20%
Slip (%) 0.74 0.76 20% to +20%
59
Figure 5  24 Upper left trace: simulated three phase voltage at terminal of load, 450s arcing time;
Upper right trace: simulated three phase voltage at terminal of load, 100s arcing time; Lower left
trace: simulated three phase current in VCB, 450s arcing time; Lower right trace: simulated three
phase current in VCB, 100s arcing time; (Red trace: phase A; Green trace: phase B; Blue trace:
phase C)
Unlike the switching operation of motor under starting condition, the amplitude and
the rate of rise of TRV is much lower for the motor under full load condition, as
shown in Figure 525. The overvoltage due to multiple reignition does not reach to a
serious level and the virtual current chopping is not observed at this moment. This is
because when the motor is running, the back EMF appears on the winding of motor.
Once the motor is disconnected, the rotor cannot stop rotating instantaneously. It
continues generating the back EMF so that the TRV builds up slowly and reduces the
switching overvoltage into insignificant level.
Figure 5  25 Left trace: simulated voltage across the contacts of phase B of VCB, 450s; Right
trace: simulated voltage across the contacts of phase B of VCB, 100s
60
When the arcing time is decreased to a very short time around 50s, the virtual
current chopping is observed in VCB and the considerable overvoltage around 33kV
(3.7p.u) occurs at the terminal of motor, as shown in Figure 526. This is because that
the VCB does not have enough time to build up high dielectric strength due to short
arcing time. As a result, the multiple reignitions occur in VCB and HF current is
injected into network to cause virtual current chopping.
Figure 5  26 Left trace: simulated three phase voltage at terminal of load, 50s arcing time; Right
trace: simulated three phase current in VCB, 50s arcing time; (Red trace: phase A; Green trace:
phase B; Blue trace: phase C)
5.6 Modeling of Induction Motor under Light Load
Condition
The modeling of induction motor under light load condition is performed by adjusting
the slip and the load torque in the UM3 model which is proposed in Section 5.4.
Based on the requirements of different driven machines, the phase current of light
load motor is set to 20% rated phase current for a compressor and 15% rated phase
current for a pump.
5.7 Evaluation Circuit and Simulation Results of Motor
under Light Load Condition
The switching operation of motor under light load condition is the most common
situation in a FPSO vessel. Before doing switching operations due to task such as
maintenance, the load of motor should be decreased first. The evaluation circuit which
is proposed in Figure 53 is also used in this part to verify the electrical characteristics
and the switching overvoltage of KMT7111 motor under light load condition.
In this case, the output power of KMT7111 motor is decreased to 17.65% rated
power to approach the motor that is working under light load condition. The
61
parameters of dq0 model are the same with the motor under full load condition
besides the load torque and the slip.
Figure 5  27 Left trace: simulated voltage of phase A of KMT7111 motor under light load
condition; Right trace: simulated current of phase A of KMT7111 motor under light load
condition
Table 5  4 Comparison of electrical characteristics of KMT7111 motor under light load condition
between simulation and test report
Item
Results of
Simulation
Results of
Test Report
Tolerance of
IEC 60092101
IEC 60034.012010
Peak Value of Phase
Voltage (V)
9060.7 8981.5 +6% to 10%
Peak Value of Phase
Current (A)
175.68 177.39 
Power Factor under Light
Load Condition
0.71 0.65
Maximum absolute
value: 0.07
Mechanical Speed (rad/s) 188.3 188.3 20% to +20%
Slip (%) 0.11 0.11 20% to +20%
Output (%) 17.6 15 
As shown in Figure 527 and Table 54, the electrical characteristics of model of
KMT7111 motor has shown sufficient acceptance to the test report. Hence, this
model is used to represent the KMT7111 motor under light load condition. The
typical switching operation with around 100s arcing time is shown in Figure 528.
After the current is chopped, the TRV appears between two contacts of phase B which
is the first pole to clear. The level of TRV is much lower than the starting and the full
load conditions. The switching overvoltage appears on the phase B with the peak
value around 9.1kV (1.01pu). The multiple reignitions and the virtual current
chopping are not observed in the motor under light load condition.
62
Figure 5  28 Upper trace: simulated three phase voltage at terminal of load, 100s arcing time;
Lower left trace: simulated three phase current in VCB, 100s arcing time; (Red trace: phase A;
Green trace: phase B; Blue trace: phase C) Lower right trace: simulated voltage across the
contacts of phase B of VCB;
5.8 Location and Protective Effect of Surge Arrester
As has been discussed in the previous sections, the arcing time of VCB plays an
important role in the switching transient overvoltage, especially for the motor under
starting condition. When the motor is under full and the light load condition, the
switching transient overvoltage shows clear low level, except the situation where the
arcing time of VCB in switching of a motor under full load condition is very short. As
mentioned in Chapter 4, a surge arrester is a suitable device to protect the electrical
equipments from the switching transient overvoltage. If the surge arrester is used in
the network, the major issue is the location of surge arrester. Generally speaking, each
surge arrester has a limited protective zone of only a few to up to several tens meters.
According to a general rule of thumb, the distance between the surge arrester and the
equipment to be protected should be as short as possible [27].
In order to verify the electrical response and the protective effect of surge arrester, the
basic insulation level of 11kV motor should be introduced first. The basic insulation
level (BIL) is the level of full wave lightning impulse that a network should withstand
without having any damage or flashover [21]. As the requirement of insulation
coordination, the IEC 6003415 standard suggests the BIL of 11kV motor to be 49kV
peak value. This voltage is applied between the coil terminals of motor and the earth.
In the simulation, it is equivalent to the phase voltage in the terminals of motor. As
shown in Figure 57, the switching overvoltage of KMT7111 motor under starting
condition with 350s arcing time is the worst case at this moment, and the peak value
63
of phase voltage reaches to around 56kV (6.2p.u). Such high overvoltage could
seriously damage the main insulation of motor and decrease the lifetime. Therefore,
the surge arrester is connected to the terminal of KMT7111 motor or the load side of
VCB, and then the limitation effect of overvoltage in above the worst case is checked.
Figure 529 shows that the switching overvoltage is clearly mitigated by using surge
arrester in the terminals of load. The peak value of phase voltage is limited to around
26kV (2.9p.u) and it is below the BIL of motor. Moreover, another important factor is
that the surge arrester cannot decrease multiple reignitions in the VCB.
Figure 5  29 Simulated three phase voltage in terminal of KMT7111 motor, 350s arcing time,
surge arrester is connected at terminal of KMT7111 motor; (Red trace: phase A; Green trace:
phase B; Blue trace: phase C)
Figure 530 shows the situation that the surge arrester is added to the downstream of
VCB. The distance between the surge arrester and the KMT7111 motor is around
180m. In this case, the switching transient overvoltage is mitigated to around 36kV
(4p.u). Although the peak value of overvoltage is below the BIL of motor, the
performance is not as good as the case where the surge arrester is connected at the
terminals of motor. It is because that the traveling waves reflect in the cable and make
the phase voltage in terminals of motor is higher than the protective level of surge
arrester [27].
Figure 5  30 Simulated three phase voltage in terminal of KMT7111 motor, 350s arcing time,
surge arrester is connected at load side of VCB; (Red trace: phase A; Green trace: phase B; Blue
trace: phase C)
64
5.9 Conclusions
This chapter has introduced three different models of motor under starting, full load
and light load conditions. The author proposed modified method to estimate the
accurate parameters of motor under full load condition and the calculated parameters
also show the low deviation for motor under light load condition. The main gas
compressor A motor KMT7111 is chosen as an example to determine the switching
transient overvoltage under starting, full load and light load condition. Consequently,
when the KMT7111 motor is disconnected during start up, the most serious
overvoltage could reach to a very high level which is above the BIL of motor.
Moreover, the overvoltage above the BIL of motor is not observed for the
disconnection of KMT7111 motor under full load and light load conditions. Once the
surge arrester is used, the amplitude of overvoltage is mitigated effectively and the
most appropriate location of surge arrester is the terminal of motor.
65
Chapter 6 Analysis of Switching Transient
Overvoltage in the Power System of A FPSO Vessel
6.1 Introduction
The purpose of this study is to check the switching transient overvoltage of different
motors and confirm the protective effect of surge arrester in the 11kV power system
of the selected FPSO vessel. Due to the large numbers of equipments, a simplified
layout of 11kV power system is used, as shown in Figure 61. Based on the client
requirements, the configuration of main generators is 3+1. It means that three main
generators work under normal production and one main generator is used as a cold
redundancy. The simplified layout ignores the cold redundancy of main generator.
Some induction motors and transformers which have the rated power around 1MW to
4MW are also ignored from the simplified layout. Originally, the 350mm
2
three core
cable is used to connect between the VCB and motors such as PMV2201B,
KMT7941, PMT6711 and PMV2201A. Due to lack of dimensions inside the three
core cable, the three 150mm
2
single core cables are applied to replace the three core
cable.
Four typical induction motors under starting, full load and light load conditions are
analyzed in this chapter: the 10.2MW main gas compressor A motor KMT7111
connected by 180m cable to a VCB, the 10.2MW main gas compressor B motor
KMT7131 connected by 300m cable to a VCB, the 5.5MW water injection pump
motor PMT2611 connected by 160m cable to a VCB, and the 1.25MW refrigerant
compressor motor KMT7941 connected by 240m cable to a VCB. Therefore, the
relationship between motors with different rated power and switching transient
overvoltage is recorded. Besides the effect of rated power of motor, the influence of
length of cable is also considered by doing comparable analysis between the main gas
compressor A motor and the main gas compressor B motor. Finally, the surge arrester
is applied in the terminals of four motors under starting condition. The switching
transient overvoltage of them is compared with the situation where motors do not
have surge arresters. The modeling of different equipments is the same to the previous
chapters. The above information of switching transient overvoltage is calculated by a
statistical way.
66
Figure 6  1 Simplified layout of 11kV power system of FPSO vessel
67
6.2 Results of Switching Operation of Motor under
Starting Condition
The interruption of motor under starting condition is usually from the inappropriate
settings of protection relays and a long duration of starting current during
commissioning of a FPSO vessel. When it occurs, as explained in previous chapters,
the multiple reignitions could lead to a serious overvoltage in the windings of motor.
As shown from Figure 62 to 65, four typical loads are disconnected by the VCB and
the relationship between cumulative probability and maximum phase voltage is
recorded.
Figure 6  2 Cumulative probability of maximum phase voltage at the terminal of KMT7111
motor under starting condition without a surge arrester connected
Figure 6  3 Cumulative probability of maximum phase voltage at the terminal of KMT7131
motor under starting condition without a surge arrester connected
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
0
0.2
0.4
0.6
0.8
1
1.2
0 1 2 3 4 5 6 7 8 9
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
68
Figure 6  4 Cumulative probability of maximum phase voltage at the terminal of PMT2611
motor under starting condition without a surge arrester connected
Figure 6  5 Cumulative probability of maximum phase voltage at the terminal of KMT7941
motor under starting condition without a surge arrester connected
Figure 62 and Figure 63 show the influence of length of cable, the maximum phase
voltage of KMT7131 motor is slightly lower than the KMT7111 motor, because the
KMT7131 motor is connected to the VCB by a longer cable than KMT 7111 motor.
If the length of cable is increased, the lumped capacitance of cable is also increased. It
means that the surge impedance and the rate of rise of transient voltage are reduced.
As a result, the KMT7131 motor experiences lower overvoltage. Besides the effect of
length of cable, the above figures have shown another interesting phenomenon that
the small machine produces a higher switching overvoltage than the large one.
Generally speaking, it is a common behavior for switching operation of induction
motor [28]. Two basic reasons can be used to explain this phenomenon. First of all, if
the VCB does not experience reignitions, the switching overvoltage is caused by the
current chopping, for example the overvoltage of KMT7941 is around 1.4p.u and the
overvoltage of KMT7111 is around 1.1p.u respectively. The equation 23 shows that
this overvoltage is proportional to the level of current chopping and surge impedance
of load. Since the trend of surge impedance is inverse proportional to the rated
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Switching Overvoltage (p.u)
69
apparent power of motor, as shown in Figure 66 [29], the small motor produces the
higher switching overvoltage which comes from the current chopping.
Figure 6  6 Surge impedance per phase of synchronous machines; (The impedance of induction
motors can be assumed to be less than 80% of the value in this Figure)
Second, if the VCB has multiple reignitions, the overvoltage may reach to 10.4p.u for
the KMT7941 motor, whereas the other three motors keep the overvoltage below
9p.u. As has been mentioned in Chapter 2, the level of current chopping of modern
VCB is around 6A. When the motor has lower rated power, the phase current of motor
is also reduced, such as the RMS value of phase current of KMT7111 is 627A and
the RMS value of phase current of KMT2611 is 346A. As a result, if the VCB
performs switching operation for motors, the phase current of small machine reaches
to the level of current chopping before the large machine, as shown in Figure 67. A
time interval T appears between the chopping time of small machine and the chopping
time of large machine. Therefore, when the small machine has multiple reignitions,
the time interval T could make reignition more serious and get more possibilities to
have the virtual current chopping. This explanation is approved by the simulation as
shown in Table 61.
Figure 6  7 Comparison of current chopping for small and large motors; (Green trace: phase
current of small motor; Red trace: phase current of large motor)
70
Table 6  1 Comparison of reignitions characteristics for KMT7111, PMT2611 and KMT7941
motors
Machine
Type
Times of Switching
Operations with
Reignition
Times of Virtual
Cuurent Chopping
Total Times of
Simulation in One
Electrical Period
KMT7111 12 10 50
PMT2611 18 18 50
KMT7941 17 16 50
With a surge arrester connected at the terminal of motor, the switching overvoltage is
obviously decreased, as shown in Figure 68 to 611. As has been mentioned in
Chapter 5, the surge arrester successfully mitigates the maximum overvoltage to
relatively low level. However the surge arrester does not limit the cumulative
probability for overvoltage due to multiple reignitions
Figure 6  8 Cumulative probability of maximum phase voltage at the terminal of KMT7111
motor under starting condition with a surge arrester connected
Figure 6  9 Cumulative probability of maximum phase voltage at the terminal of KMT7131
motor under starting condition with a surge arrester connected
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Switching Overvoltage (p.u)
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
71
Figure 6  10 Cumulative probability of maximum phase voltage at the terminal of PMT2611
motor under starting condition with a surge arrester connected
Figure 6  11 Cumulative probability of maximum phase voltage at the terminal of KMT7941
motor under starting condition with a surge arrester connected
6.3 Results of Switching Operation of Motor under Full
Load Condition
The switching operation of motor under full load condition can be resulted from the
process trip such as the differences of temperature and pressure in the processing
equipment. As shown from Figure 612 to 615, four different motors have remote
cumulative probabilities to acquire the phase overvoltage above 3.5p.u, for example,
the phase overvoltages of both the KMT7111 and the PMT2611 motors have 4%
probability to get the values between 4 to 4.5p.u, and the KMT7131 motor has 2%
probability to have the phase overvoltage around 3.9p.u. On the other hand, the small
power motor KMT7941 has slightly higher cumulative probability, where it is 10%
to have the phase overvoltage between 3.5 and 4.5p.u. This slightly increase can be
explained by the theory that is shown in Figure 67 as well. Although the phase
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5
C
u
m
u
l
a
t
i
v
e
Maximum Phase Voltage (p.u)
72
overvoltage due to the multiple reignitions and the virtual current chopping may reach
to the largest value around 4.2p.u (38kV) for the KMT7111 motor and the
KMT7941 motor, such overvoltage is still below the BIL of 11kV motor. Therefore,
the surge arrester is not really necessary in this case.
Figure 6  12 Cumulative probability of maximum phase voltage at the terminal of KMT7111
motor under full load condition without a surge arrester connected
Figure 6  13 Cumulative probability of maximum phase voltage at terminal of KMT7131 motor
under full load condition without surge arrester connected
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
73
Figure 6  14 Cumulative probability of maximum phase voltage at the terminal of PMT2611
motor under full load condition without a surge arrester connected
Figure 6  15 Cumulative probability of maximum phase voltage at the terminal of KMT7941
motor under full load condition without a surge arrester connected
6.4 Results of Switching Operation of Motor under Light
Load Condition
As has been mentioned in the beginning of this chapter, the switching operation of
motor under light load condition is the common situation which disconnects the load
from the network in a FPSO vessel. The relationship of cumulative probability and the
maximum phase voltage is shown from Figure 616 to 619. These simulations of four
motors illustrate that the serious multiple reignitions, the voltage escalations and the
virtual current chopping are not observed in switching operation of motor under light
load condition. The reason is that the rate of rise and the amplitude of TRV in the
VCB are really low, as shown in Figure 528. The main switching overvoltage is
caused by the current chopping. The below figures show that the motor with lower
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
74
rated power, the higher phase overvoltage. That is because the motor with low rated
power reaches to the high surge impedance, as shown in Figure 66. Consequently, the
small motor such as KMT7941 experiences slightly higher overvoltage than the other
motors. When compared with the previous two cases of motor working conditions, the
motor under light load condition has relatively low overvoltage and such overvoltage
is far below the BIL of 11kV motor. As a result, the surge arrester has little effect on
this overvoltage and it is unnecessary to add the surge arrester into the network to
protect the motor under light load condition from overvoltage.
Figure 6  16 Cumulative probability of maximum phase voltage at the terminal of KMT7111
motor under light load condition without a surge arrester connected
Figure 6  17 Cumulative probability of maximum phase voltage at the terminal of KMT7131
motor under light load condition without a surge arrester connected
0
0.2
0.4
0.6
0.8
1
1.2
1.005 1.01 1.015 1.02 1.025 1.03 1.035 1.04 1.045
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
0
0.2
0.4
0.6
0.8
1
1.2
1 1.05 1.1 1.15 1.2
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
75
Figure 6  18 Cumulative probability of maximum phase voltage at the terminal of PMT2611
motor under light load condition without a surge arrester connected
Figure 6  19 Cumulative probability of maximum phase voltage at the terminal of KMT7941
motor under light load condition without a surge arrester connected
0
0.2
0.4
0.6
0.8
1
1.2
1.08 1.1 1.12 1.14 1.16 1.18 1.2
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
1.15 1.2 1.25 1.3 1.35 1.4 1.45
C
u
m
u
l
a
t
i
v
e
P
r
o
b
a
b
i
l
i
t
y
Maximum Phase Voltage (p.u)
76
Chapter 7 Conclusions and Future Work
7.1 Conclusions
In this study, the switching transient overvoltage in the 11kV power system of
Floating Production, Storage and Offloading (FPSO) Vessel is analyzed. The whole
work is starting from the modeling of different equipments in the power system of the
selected FPSO vessel. After the modeling of different equipments is finished, a
simplified layout of 11kV power system in the selected FPSO vessel is described and
used to check the switching transient overvoltage for four typical induction motors.
The first important phenomenon is that the motor under starting condition could have
the most serious overvoltage, which comes from the multiple reignitions and the
virtual current chopping of VCB. In this case, the peak value of overvoltage could be
above the basic insulation level (BIL) of 11kV motor. To be compared with the motor
under starting condition, the motor under full load condition has very remote
probability to have large overvoltage due to the multiple reignitions and the virtual
current chopping. However, the level of overvoltage does not exceed the BIL of 11kV
motor. If the motor is under light load condition, the maximum overvoltage is very
moderate and far below the BIL of 11kV motor.
The second important phenomenon is that the motor with lower rated power produces
the higher overvoltage for the starting, the full load and the light load working
conditions. As has been discussed in the Chapter 6, this phenomenon is clearly
observed in the PMT2611 and the KMT7941 motors.
The third important phenomenon is that the VCB has enough capacity to interrupt the
current of four different motors in the starting, the full load and the light load working
conditions. If the current of first pole to clear of VCB is not successfully interrupted
and the power frequency current is conducting through the arc, the successful
interruption will be achieved in the next zero point of power frequency current.
Finally, the surge arrester is connected to the terminals of four motors to protect the
motors from the switching transient overvoltage, especially the motor under starting
condition. The amplitude of overvoltage is limited to the protective level of surge
arrester. The reduced level of overvoltage is below the BIL of 11kV motor.
The recommendations of whole work could be concluded: First of all, the surge
arrester is recommended for all induction motors in the 11kV power system of the
selected FPSO vessel. The most suitable location of surge arrest is the terminal of
motor. Second, due to the most serious overvoltage occurs in the starting condition of
motor, the commissioning of the selected FPSO vessel should be followed the strict
77
principles, especially the impropriate settings and the wire mistakes of protection
relays should be avoided.
7.2 Future Work
In the future, the work should focus on the accumulation of more accurate data of
equipments such as the parameters of VCB, the geometric information of cable and
the parameters of motor. Moreover, the parameters of motor are based on the power
frequency in this study. The frequency of switching transient analysis is ranging from
power frequency to several MHz. Therefore, the high frequency characteristics of
motor should be measured by an impedance analyzer and combined with the power
frequency model to represent the motor with a wide range of frequency. In this study,
the different short circuit faults cannot be truly represented due to the limitation of
generator model. In the future work, if the generator is modeled as accurate as
possible, the switching transient overvoltage during different short circuit faults can
be analyzed.
78
References
[1] J. F. Perkins, "Evaluation of Switching Surge Overvoltages on Medium
Voltage Power System," IEEE Transactions on Power Apparatus and Systems,
vol. PAS101, p. 8, 1982.
[2] T. Abdulahovic, "Analysis of High Frequency Electrical Transients in
Offshore Wind Parks," Master of Science, Electric Power Engineering,
Chalmers University of Technology, Goteborg, Sweden, 2009.
[3] W. S. M. a. T.h. Liu, ElectroMagnetic Transient Program (EMTP) Theory
Book. United States of America, 1994.
[4] J. H. a. M. Lindmayer, "Mathematical Modeling of the High Frequency
Behavior of Vacuum Interrupters and Comparision with Measured Transients
in Power Systems," presented at the XVIIth, Berkeley, 1996.
[5] J. K. a. P. Zunko, "A Statistical Vacuum Circuit Breaker Model for Simulation
of Transient Overvoltages," IEEE Transactions on Power Delivery, vol. 10, p.
7, 1995.
[6] T.h. L. W. Scott Meyer, Alternative Transients Program (ATP)Rule Book,
1987.
[7] J. A. Martinez, "Parameters Determination for Power Systems Transients,"
presented at the Power Engineering Society General Meeting, Barcelona,
2007.
[8] M. Popov, "Switching ThreePhase Distribution Transformers with a Vacuum
Circuit Breaker," PhD, Electrical Engineering, Mathematics and Computer
Science, Delft University of Technology, 2002.
[9] J. A. M. a. D. W. Durbak, "Parameter Determination for Modeling Systems
TransientsPart V: Surge Arrester," IEEE Transactions on Power Delivery, vol.
20, p. 6, 2005.
[10] V. P. a. A. G. Phadke, "Modeling of Circuit Breakers in The Electromagnetic
Transient Program," IEEE Transactions on Power Delivery, vol. 3, p. 7, 1988.
[11] M. T. G. a. M. R. Gutierrez, "Voltage Escalation and Reignition Behavior of
Vacuum Generator Circuit Breaker During Load Shedding," IEEE
Transactions on Power Delivery, vol. 12, p. 8, 1997.
[12] S. Theoleyre, MV Breaking Techniques. France: Schneider Electric, 1999.
[13] W. S. Mei, "Stochastic Studies of Overvoltages Resulting from Circuit
Breaker Operation in Industrial Installations," Master of Philosophy, Electrical
Engineering, The Hong Kong Polytechnic University, Hong Kong, 2004.
[14] L. v. d. S. M. Popov, G.C. Paap, "Investigation of the Circuit Breaker
Reignition Overvoltages Caused by NoLoad Transformer Switching Surges,"
European Transactions on Electrical Power, vol. 11, p. 10, 2001.
[15] M. Lerche, "Circuit Breaker Characteristics in Medium Voltage Equipment
under Various Network Configurations," Master of Science, Electrical
79
Engineering, Technical University of Denmark, Denmark, 2009.
[16] J. A. M. B. Gustavsen, D. Durbak, "Parameter Determination for Modeling
System TransientsPart II: Insulated Cables," IEEE Transactions on Power
Delivery, vol. 20, p. 6, 2005.
[17] J. A. MartinezVelasco, Power System Transients Parameter Determination:
CRC Press, 2010.
[18] E. Chemical. Formulations for HalogenFree Flame Retardant (HFFR)
Cables.
[19] D. F. W. M. A. Laughton, Electrical Engineer's Reference Book: Newnes,
2003.
[20] B. G. Atef Morched, Manoocher Tartibi, "A Universal Model for Accurate
Calculation of Electromagnetic Transients on Overhead Lines and
Underground Cables," IEEE Transactions on Power Delivery, vol. 14, p. 7,
1999.
[21] F. H. Kreuger, Industrial High Voltage. Delft: Delft University Press, 1992.
[22] J. M. W. Schmidt, B. Richter, K. Feser, L. Kehl, D. Qiu, "Behavior of
MOSurgeArrester Blocks to Fast Transients," IEEE Transactions on Power
Delivery, vol. 4, p. 9, 1989.
[23] P. C. Sen, Principles of Electric Machines and Power Electronics. Canada:
John Wiley & Sons, 1997.
[24] H. T. N. Ueno, Y. Murai, M. Okada, "MonteCarlo Simulation of Overvoltage
Generation in the Inductive Current Interruption by Vacuum Interrupters,"
IEEE Transactions on Power Delivery, vol. Pas103, p. 8, 1984.
[25] G. G. B. Kondala Rao, "Development and Application of Vacuum Circuit
Breaker Model in Electromagnetic Transient Simulation," presented at the
Power India Conference, India, 2006.
[26] R. D. Marcel Benecke, Gerd Griepentrog, Andreas Lindemann, "Skin Effect in
Squirrel Cage Rotor Bars and Its Consideration in Simulation of
NonSteadyState Operation of Induction Machines," PIERS online, vol. 7, p.
5, 2011.
[27] M. B. W. Schmidt, B. Richter, H. Gramespacher, "Towards the Overvoltage
Protection by Modern ZnO Surge Arrester," presented at the IEEE Power Tech
Conference, Bologna, 2003.
[28] A. T. C. Alexey M. Chaly, "The Influence of a Vacuum Circuit Breaker and
Circuit parameters on Switching Overvoltages Generated during Interruption
of Starting Motors," presented at the 17th Int. Symposium on Discharge and
Electric Insulation in Vacuum, Berkeley, US, 1996.
[29] F. X. Forstmaier, "Motor Surge Protection: at the Bus or at the Motor," IEEE
Transactions on Industry and General Applications, vol. IGA1, No.5, p. 12,
1965.
80
Appendix
Arcing Time
(s)
A (kV/ms) D (kA/s)
Peak Voltage across the Contacts of
VCB (kV)
0150 20 0.5 91.59
0150 20 0.55 91.46
0150 20 0.6 92.98
0150 20 0.65 91.55
0150 20 0.7 91.5
0150 25 0.5 91.27
0150 25 0.55 91.34
0150 25 0.6 92.28
0150 25 0.65 91.15
0150 25 0.7 91
0150 30 0.5 91.13
0150 30 0.55 91.19
0150 30 0.6 91.76
0150 30 0.65 90.98
0150 30 0.7 90.98
150300 20 0.5 91.38
150300 20 0.55 91.43
150300 20 0.6 92.75
150300 20 0.65 91.39
150300 20 0.7 91.29
150300 25 0.5 91.27
150300 25 0.55 91.18
150300 25 0.6 91.33
150300 25 0.65 90.85
150300 25 0.7 91.13
150300 30 0.5 87.67
150300 30 0.55 90.65
150300 30 0.6 90.65
150300 30 0.65 89.16
150300 30 0.7 89.23
300600 20 0.5 88.66
300600 20 0.55 90.89
300600 20 0.6 91.43
300600 20 0.65 90.63
300600 20 0.7 84.75
300600 25 0.5 83.51
81
300600 25 0.55 83.11
300600 25 0.6 87.51
300600 25 0.65 81.13
300600 25 0.7 87.26
300600 30 0.5 70.28
300600 30 0.55 54.07
300600 30 0.6 70.1
300600 30 0.65 61.17
300600 30 0.7 53.46