CIRED 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
PRACTICAL ASPECTS OF TRANSFORMER INRUSH CURRENT CALCULATION USING
TWO SOFTWARE PACKAGES
J.G. SLOOTWEG*, C.P.J. JANSEN**
*Essent Netwerk B.V., P.O. Box 856, 5201 AW, ’s-Hertogenbosch
**KEMA T&D Consulting, P.O. Box 9035, 6800 ET, Arnhem
The Netherlands
[email protected],
[email protected]SUMMARY overvoltages, it is important to be able to simulate the switch-
on of a transformer accurately. This is, however, not very
In the paper, the simulation of the energizing of transformers straightforward, because it requires more advanced simulation
and more specifically the calculation of inrush currents is approaches than used in routine network analysis and because
dealt with. First, the physical origin of inrush currents is more elaborate data than the usual name plate data of the
discussed. Then, two different transformers, namely a 20 transformer must be available.
MVA 10.6/66 kV and a 160 MVA 66/220 kV transformer, are
each modelled in two different software packages, namely The investigation of transformer energizing and the
ATP and DIgSILENT Power Factory. The models are calculation of inrush currents form the subject of this paper. It
described elaborately and the values of all relevant quantities is organized as follows. First, the physical origin of inrush
are given. currents is shortly commented upon. Then, the topology and
The models are used to simulate the energizing of the studied characteristics of the investigated system and its modelling in
transformers from a generator and to investigate the impact ATP and DIgSILENT Power Factory are described. Finally,
of various quantities, such as remanent flux and switching simulation results are presented and compared with
instant. Finally, the simulation results are compared with measurements.
measurements and it is observed that a satisfactory level of
correspondence has been achieved.
ORIGIN OF INRUSH CURRENTS
INTRODUCTION
The working principle of a transformer is governed by
Faraday’s and Ampère’s laws [3-5]. Faraday’s law states that
The working principle of a transformer is based on the
induced voltage is equal to the derivative of flux linkage, or,
presence of a magnetic flux, that couples the transformer
stated differently, that the coupled flux equals the integral of
windings. The transformer’s iron core, that carries the flux
the applied voltage. Ampère’s law relates the flux linked by a
and is hence referred to as the magnetic circuit, is
coil to the current flowing through it.
dimensioned for flux densities occurring during normal
During normal operation, the transformer current and voltage,
operation. However, when the transformer is connected to the
as well as the flux in the iron core, vary sinusoidally. When
grid after standstill, it must be magnetized, or energized, and
the transformer is not loaded, i.e. one of the terminals is open,
flux densities much higher than the values during normal
the phase shift between voltage on the one hand and flux and
operation may occur. These currents are referred to as inrush
current on the other, is approximately 90 degrees.
currents.
Assume now that the transformer is idle and a voltage is
In practice, inrush currents may pose problems for several
applied. When the presence of remanent magnetism is
reasons [1, 2]:
neglected and the connection happens at the instant at which
• Protection schemes may act upon the amplitude and/or the flux would equal zero during normal operation, i.e. at the
the shape of the transformer inrush current, mistakenly voltage’s maximum, this does not give rise to any exceptional
assuming that the observed currents are caused by some current or flux. The flux is the integral of the applied voltage
fault. and is exactly equal to what it would be if this had not been
• When a transformer is energized from a weak grid or a the first period of the applied voltage.
single generator, the voltage drop caused by the large However, when the transformer is connected at the moment
inrush currents may be such that undervoltage protection that the voltage equals zero, the flux will reach twice its
schemes, if present, are triggered and again disconnect normal height when integrating the voltage. Had the
the transformer that is just connected. transformer already been connected, the integration would
have started from the negative maximum of the flux and end
• Due to their non sinusoidal shape and the presence of at the positive maximum of the flux. However, because this is
higher harmonics, inrush currents can lead to resonance the first period of the voltage, the integration starts at zero
overvoltages caused by the transformer non-linear flux and the flux reaches twice the maximum value occurring
impedance and the capacitance of the line or cable during normal operation.
connected to it. Because these high flux densities, the iron core is driven into
In order to be able to prevent the occurrence of unintended saturation. Therefore, the transformer starts to exhibit non-
protection system actions as well as damage due to resonance linear behaviour. As a result, the current is no longer
CIRED2005
Session No 1
CIRED 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
approximately proportional to the flux. Instead, much more 34.0 MVA 20.0 MVA 20.0 MVA
current is needed to increase the flux and the current can 10.5 kV 10.6/66.0 kV 66.0/10.6 kV
easily reach much more than two times its maximum. This
G
~
large current is referred to as the inrush current. As stated
earlier, inrush currents can lead to undesirable effects, such as Transformer 1 Transformer 2
the triggering of protection devices and resonant
10.5 kV 66.0 kV 10.5 kV
overvoltages. The presence of remanent magnetism can
50 MVA 50 MVA
increase this effect, leading to even larger inrush currents. 10.5 kV 10.5/70.5 kV
G
~
MODELLING
160 MVA
Transformer 3 66.0/220.0 kV
System description
10.5 kV
The system studied in this paper consists of transformers and Transformer 4
generators in island operation, as depicted in figure 1. The 34.0 MVA 20.0 MVA
10.5 kV 10.6/66.0 kV 220.0 kV
generators have a rating of 34 MVA and 50 MVA at a voltage
of 10.6 kV. The first generator is connected to a 66 kV busbar
G
~
through a 26.5 MVA 10.6/66 kV transformers. It is driven by
Transformer 1
a gas turbine. The second generator is connected to the same
66 kV busbar through a 50 MVA transformer. 10.5 kV 66.0 kV
To investigate the inrush phenomenon and to validate the Figure 1. Investigated systems
developed models, firstly a second 26.5 MVA 10.6/66 kV
transformer and then a 160 MVA 66/220 kV transformer is ATP Model
connected to this 66 kV busbar. In the first case, only one
generator is in operation, whereas in the second case, both are The ATP model was developed using ATPDraw. The
in operation. The relevant parameters of the generator are generators were modelled as synchronous generators using the
given in table 1 and of the transformers in table 2. The model SM59 for which all relevant parameters are given in
parameters not given in the generator documentation have table 1. The transformers were modelled using the BCTRAN
been estimated and are marked with *. The one line diagrams dialog box. For those transformers connected directly to the
of the systems are depicted in figure 1. generators, which were assumed to be already energized at the
start of the simulation, all nonlinearities were neglected. In
Table 1. Generator parameters case of the transformer to be energized, the non-linear
Quantity Value Quantity Value
behaviour was modelled using the pseudo-nonlinear hysteretic
Generator Generator Generator Generator
1 2 1 2 inductor component Type 96 [6].
Snom 34.083 50 MVA Xq’ 0.5 p.u.* 0.5 p.u.* The Φ-I characteristic of the nonlinear inductor proved
MVA critical for matching the simulation results with the
Unom 10.5 kV 10.5 kV Xq’’ 0.131 0.151 p.u. measurements. However, it was unfortunately not available
p.u.*
Xl 0.10 p.u.* 0.10 p.u.* X0 0.075 p.u. 0.085 p.u.
for the investigated transformers. Therefore, it was calculated
Rl 1.481 1.317 Td0’ 6,54 s 5.49 s using the supporting routine HYSDAT that comes with ATP.
mp.u. mp.u. The values for Φ and I to be inserted in HYSDAT were
Xd 2.007 p.u. 1.910 p.u. Td0’’ 0,04 s* 0.018 s derived using the “View+” function of the BCTRAN dialog
Xd’ 0.194 p.u. 0.226 p.u. Tq0’ 0,50 s* 0.04 s* box. It was assumed that the positive saturation point, to be
Xd’’ 0.131 p.u. 0.151 p.u. Tq0’’ 0,15 s* 0,15 s*
Xq 1.8 p.u.* 1.72 p.u.*
inserted in HYSDAT, lied at a voltage of 1.18 p.u., and the
corresponding no-load current was assumed to equal 10 times
Table 2. Transformer parameters that at a voltage of 1.1 p.u. The assumption of a factor 10
Quantity Value increase in the current was based on measurements taken from
Transformer 1, 2 Transformer 3 Transformer 4
one of the transformers, of which no-load measurements up to
Snom 20 MVA 50 MVA 160 MVA
UHV 66 kV 70.5 kV 220 kV a voltage of 1.18 p.u. were available. These were used to
ULV 10.6 kV 10.5 kV 66 kV extrapolate the value for the other transformers of which only
Uk 14.7 % 11.0% 17.8 % measurements up to a voltage of 1.1 p.u. were available.
PCu 87.6 kW 116 kW 459 kW
P0, 100% 10.3 kW 31.3 kW 71.4 kW Power Factory Model
I0, 100% 0.77 mp.u. 2.3 mp.u. 1.04 mp.u.
In Power Factory the generators were modelled using the
standard synchronous machine model. The parameters for the
generators are given in table 1. The transformer model in
Power Factory used for the inrush simulations, is the two
winding transformer model shown in figure 2 [7]. The model
contains the leakage reactances at the low and high voltage
CIRED2005
Session No 1
CIRED 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
side together with the winding resistance at low and high As can be seen from the simulations, the shape and amplitude
voltage side. The model also shows the magnetizing reactance of the inrush currents depend strongly on the instant of
Xm and the ion losses resistance RFe. connection for both transformers. The cause of this is the
physical mechanism that causes the occurrence of the inrush
Rprim Xprim Rsec Xsec current, which makes them dependent on voltage at
connection instant as well as on remanent magnetism, as
described earlier.
Uprim Usec Further investigations with ATP on the impact of the presence
XM RFe of remanent magnetism (which was created by first starting up
n1 : n2
the simulation with the transformer connected and then
disconnecting and again reconnecting it) have shown that
Figure 2. Transformer model in Power Factory
when this was included, the simulation results varied even
more widely [8]. However, in the rest of this paper, remanent
To model the effect of inrush currents, the magnetizing
magnetism is not included in the ATP simulations in order to
reactance must be modelled as a function of the magnetizing
reduce the number of quantities to be varied.
current. For that, the magnetic flux is given as a function of
The simulations with Power Factory also showed that the
the magnetizing current in figure 3.
inrush current is dependent on the instant of connection and
on the assumed amount of remanent magnetism. In the Power
Φ
saturated Factory model, the instant of connection and the amount of
remanent magnetism were tuned together with the value of the
Φknee slope of the saturated flux (figure 3) in order to get the best
possible agreement between the simulation results and the
measurements described in the next section. The tuning
showed that the amount of remanent magnetism and the value
unsaturated of the saturated flux were up to a certain extent
interchangeable.
150
iM [A]
Figure 3. Flux saturation model in Power Factory 100
50
The unsaturated part of the curve is determined by the
simulation software from the no-load characteristics of the
0
transformer. The saturated part and also the knee flux, the
normal operating point of the transformer, is given by the Φ-I -50
characteristic of the transformer. As already mentioned, this
characteristic was not known for the transformers. For the -100
knee flux a typical value 1.04 p.u. is used. For the ratio
between the gradient of the saturated and unsaturated curve, a -150
0.01 0.03 0.05 0.07 0.09 [s] 0.11
typical value of 500 is used. This value is tuned to get a better (f ile SW_20_2.pl4; x-v ar t) c:X0055C-X0051C c:X0055A-X0051A c:X0055B-X0051B
match between simulated and measured curve of the inrush 200
[A]
current during the simulations. To model the remanent 150
magnetism in Power Factory, the value of the transformer
100
magnetizing flux is set at a predefined value (between 0 and
0.7 p.u.) at the start of a simulation. 50
0
SIMULATION RESULTS AND MEASUREMENTS
-50
Simulation results -100
-150
Now, simulation results obtained with the models described
-200
earlier are presented. Figures 4 and 5 show simulation results 0.015 0.035 0.055 0.075 0.095 [s] 0.115
obtained with ATP. Figures 6 and 7 show simulation results (f ile SW_20_2.pl4; x-v ar t) c:X0055C-X0051C c:X0055A-X0051A c:X0055B-X0051B
obtained with Power Factory. In figures 4 and 6, the phase Figure 4. Simulation of the connection of transformer 2 with ATP
currents for the switching of the 26.5 MVA transformer
(transformer 2) for two different instants are depicted. In
figures 5 and 7, the phase currents for the switching of the
160 MVA transformer (transformer 4) for two different
instants are depicted.
CIRED2005
Session No 1
CIRED 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
2000.00
2000
[A]
1500
1000.00
1000
500
0.00
0
-500
-1000.00
-1000
-1500
-2000.00
-2000 0.00 25.00 50.00 75.00 [ms] 100.
0.008 0.028 0.048 0.068 0.088 [s] 0.108 38-1T261: Phase Current A/LV-Side in A
38-1T261: Phase Current B/LV-Side in A
(f ile SW_160_2.pl4; x-v ar t) c:X0024B-X0001B c:X0024C-X0001C c:X0024A-X0001A
38-1T261: Phase Current C/LV-Side in A
1500 1500.00
[A]
1000 1000.00
500 500.00
0 0.00
-500 -500.00
-1000 -1000.00
-1500 -1500.00
0.00 25.00 50.00 75.00 [ms] 100.
0.008 0.028 0.048 0.068 0.088 [s] 0.108 38-1T261: Phase Current A/LV-Side in A
(f ile SW_160_2.pl4; x-v ar t) c:X0024B-X0001B c:X0024C-X0001C c:X0024A-X0001A 38-1T261: Phase Current B/LV-Side in A
38-1T261: Phase Current C/LV-Side in A
Figure 5. Simulation of the connection of transformer 4 with ATP
Figure 7. Simulation of the connection of transformer 4 with Power Factory
150.00
Comparison of simulations and measurements
100.00
In figure 8 and 9, two measurements of the connection to the
50.00 66 kV busbar are shown for transformer 2 and transformer 4.
When these figures are compared to figures 4 and 5 and to
0.00
figures 6 and 7 respectively, it can be concluded that for
-50.00
transformer 2 both the amplitude and shape of the measured
inrush currents are quite similar to the simulation results
-100.00 obtained with ATP and Power Factory.
For transformer 4, the observed discrepancies both with
-150.00
0.00 25.00 50.00 75.00 [ms] 100.
respect to shape as well as to amplitude are larger, particularly
38-2T02: Phase Current A/HV-Side in A in ATP, but to a lesser extent also in Power Factory. It proved
38-2T02: Phase Current B/HV-Side in A
38-2T02: Phase Current C/HV-Side in A not feasible to get a better match. Explanations for this
200.00
observation could be the that the magnetizing characteristic of
the iron differs more from the assumed characteristic than in
100.00 case of transformer 2 or that the effect of the simplifications
in the transformers model (e.g. neglecting the third winding of
transformer 4, which was not present in transformer 2)
0.00 negatively affect the similarity between simulation results and
measurements.
Further, in some cases the phases had to be exchanged to get a
-100.00
better correspondence. This is, however, not seen as a major
weakness of the model, because in practice it does not matter
-200.00 very much which phase exhibits which behaviour, as long as
0.00 25.00 50.00 75.00 [ms] 100.
38-2T02: Phase Current A/HV-Side in A the behaviour of each of the phases in the measurements is
38-2T02: Phase Current B/HV-Side in A
38-2T02: Phase Current C/HV-Side in A reflected in the simulation results as well.
Figure 6. Simulation of the connection of transformer 2 with Power Factory
CIRED2005
Session No 1
CIRED 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
how inrush currents can be simulated using two state of the art
power systems simulation software packages. Particular
attention was paid to the modelling of the transformer
including magnetic saturation: it was indicated that in most
cases not all necessary data is available, so that either
measurements must be available to tune the model or
assumptions must be made. In the latter case, the modelling
approach presented in the paper could be used advantageously
for formulating these assumptions.
Simulation results obtained with the developed models were
presented and it was shown that the shape and the amplitude
of the inrush currents strongly depend on the connection
instant, as well as on the assumptions with respect to the
amount of remanent magnetism. Finally, the simulation results
were compared with measurements. It was concluded that the
degree of correspondence was satisfactory, although
differences remain, which vary from case to case. Factors
explaining these could be discrepancies between the shape of
the modelled and the real magnetizing characteristic and the
impact of remanent magnetism.
Overall, the conclusion is that the modelling of inrush
currents is a complex task, because of a lack of appropriate
data and due to the strong nonlinearity and the stochastic
nature of the phenomenon. In order to cope with the latter,
always a number of simulations should be run to investigate
the extremes in the obtained results. Further, it can be
Figure 8. Measurements of the connection of transformer 2 concluded that investigations of inrush currents purely based
on simulations should be handled with care, because
significant errors can easily occur due to the great sensitivity
of the results for changes in the input parameters.
ACKNOWLEDGEMENT
Delesto B.V. is acknowledged for carrying out the
measurements, providing the equipment data and financially
supporting the work reported in the paper.
REFERENCES
[1] M.M. Adibi, 1992, R.W. Alexander, B. Avramovic,
“Overvoltage control during restoration”, IEEE Transactions
on Power Systems, v. 7, n. 4, pp. 1464-1470.
[2] C.P. Cheng, S. Chen, “Simulation of resonance over-
voltage during energization of high voltage power network”,
International Conference on Power System Transients, New
Orleans.
[3] M.J. Heathcote, 1998, The J&P Transformer Book, Reed
Educational and Professional Publishing, Oxford.
[4] J.J. Winders, 2004, Power Transformers, Marcel Dekker,
New York.
[5] http://www.allaboutcircuits.com/vol_2/chpt_9/12.html
[6] Can/Am EMTP User Group, 1995, Rule Book-Alternative
Figure 9. Measurements of the connection of transformer 4 Transient Program, Oregon.
[7] DIgSILENT, 2004, Power Factory Version 13; User
CONCLUSIONS Manual, Gomaringen.
[8] M. Rioual, C. Sicre, 2001, “Energization of a no-load
In this paper, a practical approach towards the simulation of transformer for power restoration purposes: Impact of the
transformer inrush currents was discussed. First, the physical sensitivity to parameters”, International Conference on
origin of inrush currents was described. Then it was discussed Power System Transients, Rio de Janeiro.
CIRED2005
Session No 1
