Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

ISBN : 978-1-5090-2690-6 Bali, 6 - 8 October 2016 ICSGTEIS 2016

Transformer Inrush Transients Using Jiles-Atherton


Model in PSCAD/EMTDC
H. Abdull Halim1,2, Thinh Dao1, B.T. Phung1 and J.E. Fletcher1
1 School of Electrical Engineering and Telecommunications,
The University of New South Wales, Australia
2 School of Electrical Systems Engineering,
Universiti Malaysia Perlis (UniMAP), Malaysia
hana.abdullhalim@student.unsw.edu.au

Abstract — To investigate the effect of transformer the proposed solutions are too costly and not suitable for all
energisation inrush, accurate modeling of the transformer is vital transformers: the inductive loads are only applicable for
where a mathematical model must be established for numerical transformers with air-gap; the use of series impedance is
calculations. This paper presents a detailed transformer undesirable as it will cause voltage distortion, and also lower
modelling for the slow transients study. The PSCAD/EMTDC
classical transformer combined with the Jiles-Atherton non-
the efficiency due to inherent power loss.
linear hysteresis core model is used for modelling. The developed As the power grid is moving toward accommodating more
model enables determination of the B-H curve including distributed generations, the need to investigate the effect of
consideration of the saturation effects in the magnetic material. inrush is crucial. More participants in the network will bring
For validation, laboratory work has been carried out. A 16kVA, more frequent changes and lead to more switching [1]. The
250V/11kV single phase distribution transformer is used in installation of distributed generation from an independent
simulation and experimentation to obtain its magnetising inrush power provider may compromise several parallel generation
response. Both PSCAD simulation and experimental units. More complex inrush events are expected to happen.
measurements show good agreement and thus demonstrates the When several transformers are connected in parallel, the
capability of the model to accurately represent the energisation
transient. The model shows it can generate the hysteresis loop
sympathetic inrush current may reduce the peak of the inrush
representative of the magnetic core material. For further but decay very slowly and thus presents another problem.
validation, MATLAB simulation is also developed for result In order to study the switching inrush transients, a detailed
comparison. transformer model is developed in this paper, complete with
the B-H curve and the magnetising inrush is then performed
Keywords—transformer energisation; inrush; transformer and presented to check the reliability of this transformer
modelling; PSCAD modelling model.

I. INTRODUCTION
II. NON-LINEAR TRANSFORMER MODELLING FOR SLOW
In power system networks, power transformers are the most TRANSIENT STUDIES
essential and expensive component. Energisation of the
transformer is considered a critical event as it often results in A. Transformer Modelling for 50/ 60 Hz Frequency
high magnitude inrush current. This transient disturbance will
In modelling the transformer for energisation transients
adversely affect not only the transformer itself but also other
study, the iron core and the representation of windings are to
connected devices on the network. It may cause many power
be focused. As the slow front transient is defined from 50/ 60
system issues; for example voltage sags, mal-operation of the
Hz to 20 kHz, the line energisation is considered. In slow
relay, harmonic over-voltages and mechanical stress on the
transient studies, the short- circuit impedance data, saturation
transformer windings.
data and all losses are very important [4].
Various researches have been conducted on the inrush
calculation, its mitigation and also provided a few solutions to
minimize these inrush currents. Some of the proposed B. Challenges in Transformer Modelling
solutions include [1-3]: synchronously connecting an To achieve a good model, the challenges include [5]:
inductive load near a supply voltage peak (opposite with the ƒ The accurate representations of the core
zero voltage switching) that can have the maximum inrush ƒ The flux initialization
transients, placing a series impedance in the supply, using a ƒ The core losses and hysteresis losses with their non-
pre-load circuit with the input stage of the converter, linearity and dependence on the frequency
following a specific soft-start to control the power switching. ƒ Complicated parameter estimation
However, all the solutions are depending on the particular ƒ The piecewise- linear curve characterization of the non-
characteristics of the transformer itself where its design and linear inductance/ resistance.
core material are the main factors to be considered. Some of

978-1-5090-2690-6/16/$31.00 ©2016 IEEE 38


Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on July 01,2023 at 12:20:33 UTC from IEEE Xplore. Restrictions apply.
III. HYSTERESIS CHARACTERISTICS B = μo .( H + M ) (1)
where B is the flux density, H is the magnetic field intensity
A. Theory of the Hysteresis Characteristics and M is the total magnetisation. The total magnetisation is the
One of the important elements of transformer modelling combination of two components, irreversible magnetisation
for inrush simulation is the representation of the core. Prior to and reversible magnetisation.
the simulation modelling, an accurate hysteresis magnetic loop For anhysteretic magnetisation and the effective magnetic,
has to be obtained. the equations are given by:
The magnetic hysteresis loop (known as the B-H curve)
shows the relationship between the flux density B against the M an = M s . f .( H e ) (2)
magnetic field strength H of the ferromagnetic core. By He = H + α M (3)
plotting the hysteresis, it defines the delay of the magnetic The improved Langevin type function can be applied to
material, which is shown by the behavior between B and H produce the sigmoid type curve for f(He) as below:
when magnetised and de-magnetised.
By referring to the B-H curve as shown in Fig. 1, it can be ª 1 1 º
seen that the relationship traces a closed loop a-b-c-d-e-f-a as M an = M s « − » (4)
the magnetising current flows through the coil. The de- ¬« tanh( H e A) ( e A) ¼»
H
magnetised point would be at center, 0. The gradual increasing In (2)-(4), Ms, He and Man are respectively, the saturation
and decreasing curves show the magnetised and de- magnetisation, the effective magnetic field and the
magnetised behavior. On magnetizing, B and H increase in the anhysteretic magnetisation. Į is a Jiles-Atherton parameter
positive direction and reach saturation (point a). However, representing the inter-domain coupling factor. A denotes the
when the magnetising current reduces, it will not reach zero shape parameter.
because of the residual, as shown from point a to b. Differentiating (4) with respect to He:
Opposite to point a, the reverse of it is point d. Point b and
e correspond to the residual magnetism (also called remanence dM an M s ª 1 1 º
= «1 − + » (5)
or remanent magnetisation). This is the magnetisation left in dH e A «¬ tanh ( H e A) ( H e A ) ¼»
2 2

the core after the applied magnetic field has been removed, i.e.
the magnetising current reduced to zero. Point c and f Based on (5), the final set of equations can be obtained to
correspond to the coercive force, i.e. the opposing magnetic construct M-He and B-H loops by introducing another Jiles-
intensity required to remove the residual magnetism. Atherton parameter, c:
dM an M an − M
c +
dH e δ k α ( M an − M )

dM μo 1− c
= (6)
dH dM an
1−αc
dM e
where į is introduced as a directional parameter which is
either 1 or -1 based on the sign of dH dt . Noted that į is also
another Jiles-Atherton parameter. By omitting some
redundancy, equation (6) can be simplified as [8]:
dM an
c
dM dH e
= (7)
Fig. 1. The relationship of B and H shown by the hysteresis loop. dH dM an
1−αc
dH e
B. The Jiles-Atherton Model At every step during the computation, the changes in the
The ferromagnetic hysteresis model developed by Jiles and magnetising current should be calculated based on the changes
Atherton is one of the most commonly-used models. Its in the magnetising flux. The magnetising flux is also
parameters are determined by series of differential evaluation calculated by integrating the magnetising branch voltage. The
algorithms and a few reasonable assumptions. The simulation process shown below can be applied to solve for the unknown
technique describes the nonlinear hysteretic transformer values for ¨H and ¨M [9]:
characteristics based on the proposed model. The Jiles-
Atherton subroutine is called in order to produce the hysteresis § 1 · § l ·

loop models.
³ → λ ⎯⎯⎯⎯
V ⎯⎯

© NA ¹
¸
⎯ → B JA
⎯⎯ → H
x¨ ¸
⎯⎯⎯
©N¹
→ I m ag
λ = N φ ;φ = BA Hl = NI
The Jiles-Atherton hysteresis model is based on the
magnetic quantities B, H and M. Converting the M-H loop to a
B-H loop, it is expressed as [6, 7]:

39
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on July 01,2023 at 12:20:33 UTC from IEEE Xplore. Restrictions apply.
IV. TRANSFORMER MODELLING TABLE I. TRANSFORMER PARAMETER
Parameter Value
A. Modelling in PSCAD and Parameter Calculations Rating 16 kVA
HV Voltage 11 kV
LV Voltage 0.25 kV
In PSCAD, the transformer models are in the Transformers Frequency 50 Hz
Library Group of the PSCAD Master Library. There are two HV Current 1.45 A
fundamental types of transformer models available: classical LV Current 64 A
and Unified Magnetic Equivalent Circuit (UMEC).
Since PSCAD 4.6 was released, the classical transformer has
been added with the hysteresis algorithm. The algorithm C. Simulation of Inrush Current
includes two modelling methods for hysteresis which are the The inrush transient is observed by inputting the calculated
basic hysteresis and the Jiles-Atherton model. and measured parameters. Then, the switching angle is varied.
All the parameters in PSCAD are specified in per unit The angle is started with 0o, and then increased. The measured
system. For this research, several parameters are estimated. input peak currents are shown in Table II and later compared
Besides the basic parameters given from the transformer name with the laboratory experiment results for validation analysis.
plate and basic calculation as described in [10], other formulae
to calculate the model parameters are:
2
§ %I z · 2
Winding leakage reactance (pu): x = ¨ ¸ −r (8)
© 100 ¹
Psc [ kW ]
Load losses resistance (pu): r = (9)
MVASCtest [ MVA ] × 1000
2
and the copper loss, which corresponds to I1 R1 in the
2
primary side and I 2 R2 in the secondary side.
Fig.3. Magnetising inrush when switching angle= 0o.
B. The PSCAD Transformer Model

A detailed PSCAD model of the transformer was created,


with the same parameters as the transformer used in the
laboratory experiments. The classical transformer with the
Jiles-Atherton model presented in section III has been
implemented in this environment. The accepted flux value is
estimated in the modelling process.
The test object used is a single-phase distribution
transformer, 16 kVA, 11 kV/250V. To perform the experiment,
the transformer is energised on the low voltage side from the
mains supply of 240 V, 50 Hz. The transformer energisation
simulation is shown in Fig. 2 and the transformer parameters
are listed in Table I. Fig.4. B-H curve for PSCAD transformer modelling.

Fig. 3 shows the peak inrush in the first energisation (zero


phase angle). When the phase angle is varied by increment of
45o, the peak currents are shown in Table II. From this PSCAD
modelling, the B-H curve is plotted. Fig. 4 shows the B-H
curve generated by the modelling.

TABLE II. INRUSH CURRENT MAGNITUDES

Switching Angle (degree) Peak Current (A)


0 717.67
45 602.28
Fig.2. PSCAD simulation for inrush current. 90 321.75
135 -200.35
180 -393.44

40
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on July 01,2023 at 12:20:33 UTC from IEEE Xplore. Restrictions apply.
the dashed green line denotes the experimental result. Fig. 7 is
the result from the oscilloscope, showing the magnetising
V. VALIDATION WORKS AND ANALYSIS inrush for the zero switching angle.

A. Laboratory Works

To verify the transformer inrush modelling, the experiment


was set up in the laboratory to measure the transients and
compare the results to those obtained from simulation.
The experimental setup is shown in Fig. 5. A single phase
11kV/250V tap changeable distribution transformer
manufactured by Tyree Industries is chosen as the test object.
During inrush current experiment, the low voltage side of the
transformer is energised via the mains supply (240V 50Hz) and
the high voltage side is left open. The input current is measured
with a 300 A current probe and a differential voltage probe is
also used to measure the input voltage. An oscilloscope is used Fig 6. Comparison of experimental data vs simulation data.
to monitor the probe signals. A point-on-wave switching setup
is custom-designed and built in the laboratory, using
microcontroller MSP 430 and interfaced to a desktop computer
for switching control by software. A Celduc S0745090 triac is
used for the point-on-wave switch.

Fig. 7. Magnetising inrush for switching angle = 0o

When a single transformer is energised, the inrush current


amplitude could reach ten times its rated current. Since the
experiment energised the low voltage side and the rated current
is 64 A, it can be seen that the energised current amplitude can
reach about 640 A. Fig. 3 shows that the large current
amplitude at the beginning, and then it decays substantially,
similar to a typical inrush current. The decaying is mainly due
to the effective resistance in the power circuit. It should be
Fig. 5. Experimental setup for magnetising inrush current experiment. mentioned that only the peak inrush is noted in this laboratory
work, as most of the time, the circuit breaker will trip right
B. Experimental Measurement Results after that.
By comparing with the previous work in [10], the
simulations in the current study using the Jiles-Atherton model
In order to verify the model, the transformer was energised
provides closer agreement to the experimental values. Since the
under varying the switching angles. The transformer inrush
laboratory experiments and the PSCAD/EMTDC simulations
current level is recorded for energisation angle from 0 degree
give similar results, it demonstrates the capability of the model
with 45 degrees increment until 180 degrees. The input peaks
to accurately represent the energisation transient of a
of the magnetising inrush current captured on the oscilloscope
distribution transformer.
are measured and compared with results obtained from the
PSCAD transformer model.
C. Results comparison with MATLAB
As the energisation angle varies, the peaks of the inrush are
recorded. Fig. 6 shows comparison between magnetising inrush To gain further confidence on the validity of the proposed
experiments in the laboratory and simulated results. The B-H curve and its model, MATLAB is used for simulating the
straight red line is the simulation result with Jiles-Atherton same rated transformer energisation. The results are shown in
hysteresis algorithm taking into account the residual flux while Figure 8.

41
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on July 01,2023 at 12:20:33 UTC from IEEE Xplore. Restrictions apply.
REFERENCES

[1] W. G. C4.307, "Transformer Energisation in Power Systems: A


Study Guide," vol. 568, ed. France: CIGRÉ Publication, February
2014, pp. 8-73.
[2] R. Harchandani and R. Kale, "Selection of Effective Mitigation
Method for Inrush Current in Power Transformer," International
Journal of Advanced Technology in Engineering and Science, vol.
02, May 2014 2014.
[3] P. Jinsheng, "Assessment of Transformer Energisation Transients
and Their Impacts on Power Systems," Doctor of Philosophy,
Faculty of Engineering and Physical Sciences, The University of
Manchester, 2013.
[4] International Conference on Large High Voltage Electric Systems
Working Group Internal Overvoltages, Guidelines for
Representation of Network Elements when Calculating Transients:
CIGRÉ, 1990.
[5] N. Chiesa, "Power Transformer Modeling for Inrush Current
Fig.8. B-H curve from MATLAB. Calculation," Degree of Philosophiae Doctor, Department of
Electric Power Engineering, Norwegian University of Science and
Technology, 2010.
As seen from the B-H curve, MATLAB shows a smooth [6] D. Zhang and B. T. Phung, "A Large-Pulsating-Current Generation
and its Application to Identifying Parameters in Jiles-Atherton
fully-closed loop and reveals hysteresis loss within the loop Model for Current Transformer " Asian Journal of Engineering
whereas in PSCAD, the shape fitting is not good. Note that in and Technology vol. 03, pp. 1-15, 2015.
the PSCAD, users are limited to the Jiles-Atherton model [7] U. D. Annakkage, P. G. McLaren, E. Dirks, R. P. Jayasinghe, and
implemented with some fixed parameters while in the A. D. Parker, "A Current Transformer Model Based on the Jiles–
Atherton Theory of Ferromagnetic Hysteresis," IEEE Transactions
MATLAB, the coding with all the Jiles-Atherton parameters on Power Delivery, vol. 15, pp. 57-61, January 2000.
are identified. The difference between these implementations [8] D. C. Jiles, J. B. Thoelke, and M. K. Devine, "Numerical
may occur since there are unknown parameters assumed in the Determination of Hysteresis Parameters the Modeling of Magnetic
process. This issue will be examined in future investigations. Properties Using the Theory of Ferromagnetic Hysteresis " IEEE
Transactions on Magnetics, vol. 28, pp. 27-35, January 1992.
[9] "Transformer Magnetic Hysteresis," in PSCAD On-line Help
System, ed: Manitoba HVDC Research Centre, Last updated 2015-
VI. CONCLUSION 06-05.
[10] H. A. Halim, T. Dao, and B. T. Phung, "Modeling Distribution
Transformers for Inrush Transients Study," ARPN Journal of
In this paper, a single-phase distribution transformer model Engineering and Applied Sciences vol. 11, 2016.
is developed and presented for the study of magnetising inrush
currents. The modelling is based on the use of
PSCAD/EMTDC software which provides the Jiles-Atherton
hysteresis model embedded in the PSCAD classic saturation
transformer. The electrical parameters are still adopted from
the previous study by the authors but the latest PSCAD 4.6
now includes the Jiles-Atherton hysteresis in its classical
model.
By comparing with the experimental results, the PSCAD
simulation results of inrush currents show good agreement. All
in all, the simulations with the Jiles-Atherton model provide
closer magnitude to the experimental values as compared to the
previous model (without the Jiles-Atherton model). The
magnitude difference between PSCAD/EMTDC simulation
and laboratory experiment results is acceptable. Hence, the
satisfactory results demonstrate the capability of the model to
accurately represent the energisation transient of a distribution
transformer. The model shows it can actually form the
hysteresis loop to follow the characteristics of the magnetic
core material.
ACKNOWLEDGMENT
The authors would like to acknowledge the assistance by
Mr. Zhenyu Liu with the experimental work and Dr. Daming
Zhang with the MATLAB coding.

42
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on July 01,2023 at 12:20:33 UTC from IEEE Xplore. Restrictions apply.

You might also like