A SIMPLE TRANSFORMER MODEL FOR ANALYSIS OF TRANSFERRED
LIGHTNING SURGES FROM MV TO LV LINES
Alexandre Piantini Welson Bassi
Institute of Electrotechnics and Energy
University of São Paulo
Av. Prof. Luciano Gualberto, 1289, 05508-900, São Paulo - SP (Brazil)
Tel.: + 55 11 818 4720 - Fax: + 55 11 212 9251 - E-mail: piantini@iee.usp.br
ABSTRACT
This paper presents a distribution transformer model which
has been developed aiming at the analysis of transferred
lightning surges from MV to LV lines, taking into account
the load conditions, as well as the experiments that were
conducted for the verification of its validity. The results of
the comparisons between measured and calculated
voltages indicate that the proposed model can be an useful
tool for the evaluation of transferred surges, taking into
account the compromise between accuracy and simplicity.
Some examples of the applicability of the model, in terms of
the calculation of surges in LV networks are also
presented.
INTRODUCTION
Voltage and current surges transferred from the medium
voltage lines are important sources of disturbances on
secondary circuits. Such transients, usually caused by
direct or nearby lightning, have to be dealt with in any
analysis of the performance of overhead distribution lines.
A common way of calculating the voltages on the low-
voltage side is to represent the transformer by a capacitive
network. This method, however, may be highly inaccurate,
as has been demonstrated by laboratory tests. Furthermore,
although more sophisticated models have been proposed
[1], their validity is usually limited to no-load conditions.
In this paper a simple distribution transformer model is
developed, aiming at the analysis of voltages and currents
transferred to the low-voltage side, taking into account the
load conditions, when lightning strikes the primary
conductors. As the main motivation for the study was
related to the surges on the low-voltage network, the
transformer is considered to be protected by a set of
arresters at the MV side. Under this condition it can be
considered linear, and therefore it is treated as a quadripole.
Jorge M. Janiszewski Nelson M. Matsuo
Polytechnic School
TRANSFORMER MODEL
Transformer characteristics
Initially, tests were performed on a distribution transformer
commonly used in Brazil (three-phase, with rated power of
30 kVA, 13.8 kV - 220/127 V, delta-star connection), so as
to determine its characteristics in terms of input, output and
transference impedance as a function of frequency. As the
transformer was supposed to be protected by arresters, it
was assumed that the voltages on the three phases have
approximately the same amplitude. The input impedance
Z11 was then obtained with the HV terminals
interconnected and with the LV terminals in open circuit.
Voltage signals of approximately 16 V peak-to-peak, in the
frequency range of 10 kHz to 1 MHz, were applied to the
HV terminals. A digital storage oscilloscope with sampling
rate of 500 MS/s and bandwidth of 300 MHz was used for
measuring the applied voltage and the current at the HV
terminals 1 (external) and 2 (middle), as well as the phase
shift between voltage and current signals. A probe
associated with an amplifier was used for current
measurements.
Fig. 1 shows the circuit used for the determination of Z11.
For each frequency its modulus was calculated by the ratio
of the amplitudes of the voltage at point P1 and the current
at point P3. The phase shift was given by the difference in
time between the zero crossings of the voltage and current
signals. The output impedance Z22 was obtained
analogously. For the determination of the transfer
impedance Z21 the calculations considered, with reference
to Fig. 1, the voltage transferred to secondary (at the point
P2) and the current at the point P3.
Fig. 1 - Circuit for the determination of the transformer input impedance
Z11. G: signal generator; amp.: amplifier; osc.: oscilloscope.
Figs. 2 to 4 present the curves, as functions of frequency, of Z21 [kΩ]
Z [ohms]
21

the transformer impedances. 250

0.25

0.2
200

Z11
Z [kΩ]
[ohms]
11
0.15
150
35
35000

30
30000
0.1
100

25
25000

0.05
50
20
20000

15
15000 00
10 10 100 1000
10000
Frequency [kHz]
5
5000

00 Z21Fase
phase [degrees]
Z [graus]
21

10 100 1000
100
100
Frequency [kHz]

Z11
Fase phase
Z [graus][degrees]
11
50
50
00
10 100 1000
-20
-20
00
10 100 1000
-40
-40
-50
-50

-60
-60

-100
-100
-80
-80

-100
-100
-150
-150

Frequency [kHz]
-120
-120

Frequency [kHz]
Fig. 4 - Transformer transfer impedance Z21 as function of frequency.

Fig. 2 - Transformer input impedance Z11 as function of frequency.

As the HV terminals were short-circuited, the inductive
effect of the windings is practically eliminated, since the
primary windings are connected in delta. This explains the
essentially capacitive behaviour of the input impedance
Z11. On the other hand, as the LV windings are connected
Z22 [kΩ]
Z [ohms]
22
in star, the inductive effect of the windings leads to a
1400
1.4
predominantly inductive behaviour of the output impedance
1.2
1200
Z22 up to the frequency of approximately 750 kHz. The
1.0
1000 transfer impedance Z21 has a more complex frequency-
0.8
800
dependent behaviour, alternating between capacitive and
0.6
inductive characteristics, with various ressonance points.
600

0.4
400
Considering these results, simulations were then made in
0.2
200 order to obtain a circuit, as simple as possible, that could
0
0
reasonably reproduce the transformer behaviour in the
10 100 1000 frequency range adopted in this study (from 10 kHz to
Frequency [kHz]
1MHz). After some attempts the model shown in Fig. 5,
Z22
Fasephase
Z [graus][degrees]
22
composed by resistive, inductive and capacitive elements,
100
100
was considered to be representative of the transformer.
80
80

60
60

40
40

20
20
00
10 100 1000
-20
-20

-40
-40
-60
-60
-80
-80
Frequency [kHz]

Fig. 3 - Transformer output impedance Z22 as function of frequency.

Fig. 5 - Transformer model for calculation of transferred surges.
 Validation voltage [V]
50
50

Two kinds of tests were performed in order to verify the 40

40
validity of the transformer model. In the simplest
configuration, impulse voltages (1.2/50 µs waveform) with
30
30

20
amplitudes of approximately 8 kV were applied to the 20

high-voltage terminals interconnected, and the voltages 10

10

transferred to the secondary side were measured under 00

different load conditions. Figs. 6 to 8 present some -10

0 4 8 12 16 20
-10
comparisons between measured (at terminals X1 and X2)
-20
and calculated transferred voltages, obtained for different -20
time [us]
load conditions.
(b)

Fig. 7 - Voltages transferred to the secondary (resistive load of 100 Ω).

voltage [V] a) measured b) calculated
200
200

150
150
terminal X1
100
100 terminal X2
voltage [V]
50
50
35
35
00 30
30
0 10 20 30 40
-50
-50 25
25 terminal X1
terminal X2
-100
-100 20
20

-150
15
15
-150
10
10
-200
-200
time [us] 55
00
(a) 0 20 40 60 80 100
-5
-5

-10
-10
voltage [V] time [us]
250
250

200
200
(a)
voltage [V]
150
150
30
30

100
100

25
25
50
50

00 20
20

0 10 20 30 40
-50
-50 15
15

-100
-100 10
10

-150
-150 55
-200
-200 0
0
time [us] 0 60 80 100
20 40
-5
-5

(b) time [us]

Fig. 6 - Voltages transferred to the secondary under no-load conditions.

a) measured b) calculated (b)

Fig. 8 - Voltages transferred to the secondary (resistive load of 50 Ω).

a) measured b) calculated

voltage [V]
The second configuration included also the presence of a
60
60
simulated low-voltage line and of a medium voltage ZnO
50
50 terminal X1 arrester (rated current of 10 kA). Either resistors or
terminal X2
40
40 capacitors, placed at the end of the line and connected
30
30
between phases and neutral, were used to simulate different
load types. Impulse currents with amplitudes of about 5 kA
20
20
were applied to the arrester; the voltages were measured
10
10 not only at the transformer terminals (low-voltage side) but
0 also at the end of the line. The arrester was connected
-10
-10
0 4 8 12 16 20 either to the HV terminals short-circuited or only to the
time [us] terminal H1, as shown in Fig. 9. Fig. 10 shows the current
(a)
injected into the circuit.
 impulse generator
line length
L = 19 m 8

X3 d = 0,11 m L3
X2 6
surge X1 L2
arrester X0 L1 L0 no load
45 ohms
Line height 4
R1 h = 1,1 m R2

shunt kV 2

reference
0

Fig. 9 - Test configuration. Shunt resistor = 2.62 mΩ; -2

R1 = 4.5 Ω; R2 = 2.5 Ω.
-4
5 time [5 us/div]
(a)
4 [kV]
8

3
6
kA 2
4

1
2

0
0

-1
time [5 us/div] -2

Fig. 10 - Current injected into the test circuit. -4

0 5 10 15 20 25 30
NO LOAD

(b)
Figs. 11 to 14 present some comparisons between measured
and calculated phase-to neutral voltages at the loads, in Fig. 12 - Phase-to neutral voltages at L1 (VL1 - VL0), for two conditions:
different situations. For the calculations, the ATP without loads and with resistive loads of 45 Ω. Arrester connected only
to terminal H1. a) measured b) calculated
(Alternative Transients Program) was used.
15
15

10 1 nF
10 VL1-VL0 40 nF
VL2-VL0
VL3-VL0 5
5
kV

kV
0
0

-5
-5

-10
-10 tempo [5 us/div]
time [5us/div]
time [5us/div]

(a) [kV] (a)

[kV]
15
15

10
10

5
5

0
0

-5
-5

-10
-10
0 5 10 15 20 25 30
0 5 10 15 20 25 30

(b) (b)

Fig. 11 - Phase-to neutral voltages at the loads (capacitances of 40 nF). Fig. 13 - Phase-to neutral voltages at L1 (VL1 - VL0). Loads represented
Arrester connected only to terminal H1. by capacitances of 1 nF and 40 nF. Arrester connected only to
a) measured b) calculated terminal H1. a) measured b) calculated
 15 In this topology the primary and secondary lines are
partially coupled, and the transformer is placed at the end of
PR’s
SA’s em H1
in H1
10
SA’s em
PR’s in H1,H2,H3
H1H2H3 the primary line.

5 The possibility of the occurrence of flashovers in the

kV
primary and secondary circuits, as well as from the neutral
0
conductor to the ground, was taken into account by means
of the switches shown in Fig. 15. They were placed at every
-5
pole; the Vxt characteristic curves were calculated using the
standardised lightning impulse waveform (1,2/50 µs), with
-10
tempo
time [5us/div]
[5us/div]
the integration method presented in [2].
(a)
[kV] The points labelled as D1, D2, ... and E1, E2, ..., in Fig.15,
15
represent the poles, with a span distance of 30 m. Rc and Rs
10 represent the ground resistances of the consumers. Different
values were assumed for the ground resistances of the
5 poles: Rt (effectively grounded) and Rp (not effectively
grounded). The values of Rp were defined as twice the
0 values of Rt, as obtained in tests reported in [3].

-5
-10
0 5 10 15 20 25
(b)
Fig. 14 - Phase-to neutral voltages at L1 (VL1 - VL0). Loads represented
by capacitances of 40 nF. Arrester connected to terminals
H1, H2 and H3 (short-circuited) and only to terminal H1.
a) measured b) calculated
The results show that the proposed model gives a
reasonable representation of the distribution transformer in
terms of voltages transferred to the secondary. As shown in
the comparisons relative to the simplest test circuit, the
differences between measured and calculated voltages tend
to increase as the impedance connected to low-voltage
terminals decreases. The relatively good results obtained
for the more complex test configuration indicates that the
model is also adequate to more realistic situations.
LIGHTNING SURGES TRANSFERRED TO THE LV
NETWORK
The consumers loads, connected between phase and neutral,
were represented by a resistor of 30 Ω in parallel with a
30
capacitance of 4 nF. The impedance modulus of this
association lies in the range of the values presented in [4].
The lightning current, injected in the primary line at the D3
point (90 m distant of the transformer), was represented by
a triangular waveform with a peak value of 31 kA, time to
crest of 2 µs and time to half-value 50 µs. A typical VxI
curve was used for representing the characteristic of the
ZnO primary arrester.
For illustration purposes, Figs 16 to 18 show some
examples of calculated voltages transferred to the
secondary (each phase-to-neutral) in the case of a direct
strike on the MV line at the point D3. In Fig. 16 the
voltages were calculated at the point D1 (located 30 m from
the transformer). Fig. 17 refers to voltages at point D5 (150
m far from the transformer) while the Fig.18 shows the
voltages at the transformer terminals. Ground resistances of
100 Ω were assumed for Rt and Rc.

With the transformer represented by the model previously Another application of the transformer model for
obtained, simulations were then performed with the ATP, in calculating surges in LV networks as well as associated
order to calculate voltages transferred to the secondary in currents and charges in protective devices is presented in
some typical situations, such as in that presented in Fig. 15. [5].

Fig. 15 – Line configuration used in the simulations

 [kV]
CONCLUSIONS
15 15

A simple model for representation of distribution

10
10
transformers was developed, aiming at the calculation of
surges transferred to the secondary. The results of the
5
5 comparisons between measured and calculated voltages
indicate that the proposed model can be an useful tool for
0
the evaluation of transferred surges, taking into account the
0
compromise between accuracy and simplicity.
-5
-5

-10
ACKNOWLEDGEMENT
-10
0 5 10 15 20 25
0 5 10 15 20 25
(5) BFDA 1 - ND1 (6)BFDB1 - ND1 (7)BFDC1 - ND1
The authors greatly acknowledge the Center of Excellence
in Distribution of Electrical Energy (CED) for supporting
this work.
Fig. 16 - Phase-to-neutral voltages on the secondary line (at point D1).

[kV]
REFERENCES
[kV]
40
40
[1] Vaessen, P.T. "Transformer model for high
30
30
frequencies". IEEE Trans. on Power Delivery, v. 3,
no 4, pp. 1761-1768.
20
20
[2] Darveniza, M.; Vlastos, A. E. "The Generalized
Integration Method for Predicting Impulse Volt-
10
10
Time Characteristics for Non-Standard Wave
Shapes - a Theoretical Basis". IEEE Transactions
0 on Electrical Insulation, vol. 23, no. 3, June 1988,
0

pp. 373-381.
-10
- 10
5 10 15 20 25
[3] Sekioka,S.; Yamamoto, K.; Yokoyama, S.
0 5 10 15 20 25
(21) BFDA 5 - ND5 (22)BFDB5 - ND5 (23)BFDC5 - ND5 "Measurements of a Concrete Pole Impedance with
an Impulse Current Source". In: International
Conference on Power Systems Transients, Lisbon,
Fig. 17 - Phase-to-neutral voltages on the secondary line (at point D5).
September 3-7, 1995.
[kV]
[kV]
15
15
[4] Hoidalen, H. K. "Lightning-induced voltages in low-
voltage systems and its dependency on voltage line
terminations". In: International Conference on
10
10
Lightning Protection, 24, Birmingham, 1998.

5
5 [5] W. Bassi, N. M. Matsuo, A. Piantini, "Currents and
charge absorbed by low-voltage SPDs in overhead
0
distribution systems due to lightning". (Submitted to
the 11th International Symposium on High Voltage
0

Engineering, London, 1999)

-5
-5

-10
- 10
5 10 15 20 25
0 5 10 15 20 25
(1)BFA0 - N0 (2)BFB0 - N0 (3)BFC0 - N0

Fig. 18 - Phase-to-neutral voltages on the secondary line

(at the transformer terminals).