Beruflich Dokumente
Kultur Dokumente
sin sin
2
) (
0
t t
s w
t
m
e t K K
Z
V
t i
(1)
with V
m
maximum applied voltage ; Z
t
total impedance
under inrush, including system; energization angle; t
time; t
0
point at which ferromagnetic core saturates;
time constant of transformer winding under inrush conditions;
function of t
0
; K
w
coefficient which accounts for a 3
phase winding connection; K
s
accounts for the short-circuit
power of the grid.
An analytical method for determining the inrush current for
a single transformer, having taken into account the saturation
of the core, concludes that the inrush current can be
determined by [7]:
( ) ( )
+ +
+
=
t e
R
L
R
V
t I
t
L
R
p
sin sin
1
) (
2
(2)
where
=
R
L
1
tan
(3)
and - core flux; R winding phase resistance; L
winding phase inductance.
A summary of known analytical solutions is presented in
[6], [7], with expressions correlated with constructive
parameters of the transformer:
2 2 2
2
) (
core air W
L R
U
t i
( )
( )
( )
s
t t
L
R
t e t
s
core air
w
sin sin
(4)
=
N
R N S
core air
B
B B B
L
U
t i
~
1
2
) (
(5)
with
( )
cos sin 2 ) (
~
) (
~
=
core air
w
N R R
L
R
B old B new B
(6)
n
t
N
S R N
core air w
e
B
B B B
L R
U
t i
=
2 2
) (
2 2 2
(7)
The focal point of the above examples is the study of the
reenergization of a single transformer, and the requirement
for constructive details such as the remanent and saturation
flux of the core.
The data collected for the current paper covers events
which occur on a regular basis in medium voltage grids, but
only comprises of the variation curves of the phase currents,
sampled through digital protection relays.
Our case studied medium voltage distribution grid consists
of 9 power stations, 750 substations and a number of around
800 active 10/0.4 kV distribution transformers, with the rated
power ranging from 250 kVA to 1600 kVA.
This type of problem requires identifying an expression
which describes a variation curve of the inrush rms values, so
it may be regarded as an analytical curve fitting and
parameter identification problem.
III. INRUSH CURRENT MODELING AND SIMULATION
A. Identification and numerical analysis
We started from a post-event study regarding the
reconnection of a cascaded section which contains 5
substations, having a number of 4 power transformers with a
total rated power of 1.91 MVA.
From the monitoring of a delayed overcurrent protection
which initially consisted of a pickup value of 280A and a trip
delay of 0.2 sec, repeated unexpected tripping were observed.
The reconnected section is shown in Fig.2.
These settings were adopted after retrofitting the cubicles,
targeting a faster disconnection of the faulty line and ensuring
selectivity conditions with the upstream relays.
By conducting a basic analysis of the oscillograms,
followed by a raw numeric extrapolation, we found out a
convenient time-saving solution of raising the delay from 0.2
sec to 0.4 sec, applicable to all the digital protection relays.
In order to study this undesired phenomenon, multiple
Fig. 2. The reconnected section for the first event
samples were extracted.
Next, it was necessary to determine an equation to model a
realistic inrush current for a group of transformers, taking into
consideration the configuration of the reconnected section, by
correlating the inrush to the total rated power of the
transformers, the number of transformers, the total length of
the cables between the substations and the duration of the de-
energization.
The sampled inrush currents are shown in Fig. 3.
After extracting multiple samples, it resulted that the
maximum value of the peak inrush coefficient of any group of
medium/low voltage distribution transformers is:
4 . 7
_
_
= =
downstream n
inrush peak
inrush
I
I
k
(7)
n
n
i
n
downstream n
U
S
I
tr
i
=
3
1
_
(8)
It is worth mentioning that the effect of the switching angle
cannot be evaluated due to the lack of relevant data.
It was also determined that the inrush current is best
described by a decaying exponential function.
The function is described by the peak inrush current
reached within the first two cycles, and the damping
coefficient k
d
.
The damping coefficient is correlated to the configuration
of the reconnected group of transformers. The use of well-
known analytical solutions mentioned before often implies an
error percentage of 30%-40%, which is unacceptable when
targeting an optimization of the overcurrent protection delay.
We determined that the inrush current can be modeled as:
t k
downstream n inrush inrush
d
e I k I
=
_
(9)
The peak inrush coefficient k
inrush
is obtained via
identification of the peak value of the inrush current and
dividing it to the value obtained by using the corresponding
expression, (8).
The damping coefficient was identified by a trial-and-error
method for each of the three initially studied events, thus
obtaining the expression:
downstream n
I
d
e k
_
0068 . 0
11
= (10)
B. Further field measurements and determination of coefficients
In order to improve the above-mentioned expressions and
model the realistic inrush more accurately, several sets of data
were extracted, and the samples were post-processed using
Mathcad, considering the identified peak inrush coefficient
determined by (7), the equivalent rated current calculated as
shown in (8), and the value of k
d
identified by the trial and-
error method, targeting to improve its variation curve (10).
In order to conclude each studied event with adequate
settings for the protection relays, we simulated a worst-case
scenario by using k
inrush
=7.4, as identified in (7).
For this paper, a total of nine events were studied, which
are summarized in Table 1, and for which the corresponding
simulation results are shown from Fig. 4 to Fig. 12.
TABLE 1
SUMMARY OF THE STUDIED EVENTS
No.
Number of
transformers
Total rated
power
(MVA)
De-energization
duration (min)
Identified
kd
1. 4 1.91 5 3.90
2. 7 2.96 28 3.25
3. 3 1.89 2 3.28
4. 10 5.02 77 4.10
5. 14 8.06 1 6.80
6. 9 5.40 18 6.95
7. 9 5.43 50 4.70
8. 10 5.77 30 5.80
9. 5 2.15 20 3.85
Fig. 3. Sampled inrush current for the initial case of unexpected tripping
Fig. 4. Simulation results for the first case
Fig. 5. Simulation results for the second case
Furthermore, since the damping coefficient does not have
an obvious variation correlated with the reconnected sections
parameters, we defined the damping coefficient as a product
between two other coefficients, one connecting to the number
of refeeded transformers, and the other connecting to the total
rated power of the section:
pow tr d
k k k = (11)
The transformer coefficient, k
tr
, is expressed as:
rs transforme refeeded of number
k
tr
_ _ _
1
=
(12)
The power coefficient, k
pow
, has been computed using (11)
and (12), and the values obtained for each studied event are
presented in Table 2.
Fig. 6. Simulation results for the third case
Fig. 7. Simulation results for the fourth case
Fig. 8. Simulation results for the fifth case
Fig. 9. Simulation results for the sixth case
Fig. 10. Simulation results for the seventh case
Fig. 11. Simulation results for the eighth case
Fig. 12. Simulation results for the ninth case
TABLE 2
POWER COEFFICIENT DETERMINATION
No.
Number of
transformers
Total rated
power
(MVA)
kd ktr kpow
1. 4 1.91 3.90 0.25 15.60
2. 7 2.96 3.25 0.14 22.75
3. 3 1.89 3.28 0.33 9.84
4. 10 5.02 4.10 0.10 41.00
5. 14 8.06 6.80 0.07 95.20
6. 9 5.40 6.95 0.11 62.55
7. 9 5.43 4.70 0.11 42.30
8. 10 5.77 5.80 0.10 58.00
9. 5 2.15 3.85 0.20 19.25
The power coefficient follows a linear variation, suggesting
a dependence of the total rated power of the refeeded
transformers in agreement with the previous observations.
In order to establish a numerical model of the power
coefficient variation curve, the convergence of the
k
pow
=f(S
n
_
total
) function has been verified by:
( )
( ) ( )
=
= = = =
= = =
8
0
2
8
0
2
8
0
2
8
0
2
8
0
8
0
8
0
8
1
8
1
8
1
i i
pow pow
i i
n n
i i
pow
i
n pow n
C
i i i i
i i i i
k k S S
k S k S
a (13)
For the data extracted from Table 2, the correlation factor
is a
C
=0.997 (ideally 1).
In order to determine a function which describes the
dependence of the power coefficient in relation with the total
rated power of the refeeded transformers, we used Mathcad
integrated curve fitting tools, with a comparison in Figure 13.
It resulted that the power coefficient k
pow
can be accurately
described by a linear function:
=
+ =
tr
i
n
i
n pow
S k
1
853 . 12 269 . 14
(14)
As a result, we now have a set of expressions which allow
for an accurate evaluation of the inrush current of a group of
transformers.
IV. CONCLUSIONS AND AUTHORS CONTRIBUTION
The most important aspect when targeting any optimization
of a protection system is that the overcurrent pickup is
restricted by the upper thermal current limit of the weakest
cable in the protected area.
The feeder protection system is designed to protect even
the weakest element from degradation caused by external
fault currents.
Considering the cross-section and the degradation of the
cables in the studied grid, the pickup settings of 250-300 A
are commonly used. This is generally applicable to medium
voltage grids which mainly consist of service-aged
underground cables with a cross-section of 150 mm
2
.
As a conclusion, this newly developed relation offers
valuable and feasible results, as the 0.4s initial delay proved
adequate, not too long nor too short, depending on the
possible modifications in the protected sections
configuration.
Furthermore, simulated results demonstrate that the
optimization involves only sensible alterations of this delay,
which also proves the applicability of the developed
expressions.
As demonstrated earlier, the worst-case scenario involves
the initial peak inrush being 7.4 times larger than the total
rated current of the downstream transformers. Since
implementing the general 0.4s delay, there has been no
unexpected tripping in the protected sections, over a period of
24 months.
The inrush current can be therefore modeled by
determining the configuration of the section(s) which can be
reconnected at some point, determine the corresponding
transformer coefficient by using (12), the power coefficient
by using (14), and then computing the damping coefficient
and modeling the worst-case inrush with k
inrush
=7.4 and
substituting these terms into (9).
An important and potential application of this function is
represented by a possibility to enhance the automatic real-
time computing and control of the overcurrent protection in a
smart-grid, eliminating the need for a human operator to
calculate and upload the adequate settings to each protective
relay [6].
V. FUTURE PERSPECTIVES
To the best of our knowledge, there are no similar
numerical evaluations of the inrush current of a group of
transformers, other papers being focused on the protection
systems of power transformers located in power stations.
The results presented in the current paper can and will be
further investigated for improving the inrush expression, as
Fig. 13. Power coefficient numerical approximation
we observed that the sampled inrush currents drop to around
60% of the peak inrush within the first three or four cycles.
It would also be interesting to study the influence of other
phenomena disregarded in our current analysis, such as ferro-
resonance, sympathetic inrush, capacitive reactive current of
the underground cables, and load ratio of the secondary
windings.
ACKNOWLEDGEMENTS
The work on this paper is supported by the project: "TE
253/2010 Modeling, prediction and design solutions, with
maximum effectiveness, for reducing the impact of stray
currents on underground metallic gas pipelines." No.
34/2010, project funded by the Romanian Ministry of
Education.
REFERENCES
[1] L. Czumbil, Denisa te, D.D. Micu, V. opa and L. Anc, Stream Gas
Pipeline in Proximitity of High Voltage Power Lines. Part II - Induced
Voltage Evaluation, 47th Internatonal Universities' Conference on
Power Energy, UPEC, London, September 4-7, 2012.
[2] A. Abou-Safe, G. Kettleborough, Modeling and Calculating the In-Rush
Currents in Power Transformers, Damascus University Journal, Vol.
21, pp. 75 92, 2005.
[3] M. Jamali, M. Mirzaie, S.A. Gholamian, Calculation and Analysis of
Transformer Inrush Current Based on Parameters of Transformer and
Operating Conditions, Electronics and Electrical Engineering
Kaunas: Technologija no. 3(109), pp. 17 20 , 2011.
[4] Robert M.Del Vecchio, Bertrand Poulin, Pierre T.Feghali, Dilipkumar
M.Shah, Rajendra Ahuja - Transformer design principles with
applications to core-form power transformers, CRC Press, Boca Raton
London New York Washington, D.C., 2002, pag. 39-66
[5] L. Prikler, G. Banfai, G. Ban and P. Becker, Reducing the Magnetizing
Inrush current by means of Controlled Energization and de-
Energization of Large Power Transformer, International Conference on
Power System Transients, IPST, 2003.
[6] D.D. Micu, G.C. Christoforidis and L. Czumbil, Analytical Difficulties in
the Interfstud Software Development, 10th International Conference on
Applied Electromagnetics, ISBN: 978-86-6125-042-2, Nis, Serbia,
September 25-29, 2011.
[7] N. Chiesa, Doctoral Thesis: Power Transformer Modeling for Inrush
Current Calculation, Norwegian University for Science and
Technology, 2010.