Professional Documents
Culture Documents
Abstract—The paper aims at defining a methodology to evalu- For this reason, especially in low-insulation systems, they
ate the lightning performance of a system of overhead distribution represent one of the most significant causes of lightning-related
lines. As is well known, the evaluation of the lightning performance outages. As a consequence, the literature has been focused on
relies generally on a Monte Carlo approach that generates many
different lightning events and calculates the corresponding over- the evaluation of the indirect lightning performance which is,
voltages. To perform such calculation, the power system computer unfortunately, a very hard task from a computational point of
aided design (PSCAD) module developed in the companion paper view, as it requires us to solve the problem of the field-to-line
is adopted. However, as any call to PSCAD implies a nonnegligi- coupling a significant number of times in order to account for
ble computational effort, here an efficient approach is presented all the possible values that the stochastic variables can assume.
in order to limit the number of simulations as much as possi-
ble. The obtained results are compared with the ones provided by To overcome this problem, first a simplified approach has been
other available techniques both on simple structures in order to proposed [2] that basically adopts the Rusck formula for the
highlight the main features of the method and on a typical Ital- calculation of the maximum amplitude of the lightning-induced
ian distribution network segment. The comparative analysis shows voltages on the line. Then, such an approach has been improved
good agreement between the methods and the effectiveness of the in order to account both for the finite length of multiconductor
proposed technique in terms of computational effort reduction.
line and for the ground finite conductivity [3].
Index Terms—Monte Carlo methods, overvoltage protection, The complexity of the power distribution networks character-
power distribution lines, power system lightning effects. ized by a number of lines (main feeder with laterals) and by the
presence of several power components (e.g., transformers and
I. INTRODUCTION surge arresters) has been accounted for in [4], where a heuris-
tic technique has been applied in order to reduce the number
HE increasing attention toward power quality problems
T has increased the interest on the assessment of lightning
impact on power systems by researchers and engineers [1].
of calls to the coupling code. The recently released IEEE Std.
1410-2011 [5] employs that method for the evaluation of the
induced-voltage flashover rate of overhead distribution lines.
In this context, the evaluation of the lightning performance More recently, the influence of direct strokes on lightning per-
of distribution networks is very important when dealing formance of distribution networks has been analyzed [6], where
with the design of their lightning protection systems. The it was shown in particular that, if the rate of flashovers along the
lightning performance typically consists of curves reporting whole line is of interest, independently from their locations, the
how many lightning faults per year the system may experience effect of the direct strikes can be accounted for by simply adding
as a function of its insulation level, and so it expresses the the rate of direct strikes to the indirect lightning performance, as
probability that the line is subject to an overvoltage greater than each direct strike to the line is likely to cause a flashover. More
its critical-flash-over (CFO) voltage. detailed analysis of the effect of direct strikes has been carried
Lightning may cause flashovers on distribution lines from out in [7] and [8]. In [7], guidelines for the optimal spacing
both direct strikes and induced voltages from nearby strikes. between arresters are given and in [8], the direct and indirect
Direct lightning strikes to power distribution lines cause strike flashover rates are characterized from a statistical point
insulation flashover in a great majority. On the other hand, of view. However, based on these studies, the lightning perfor-
indirect strikes are more frequent and, in the presence of tall mance of a medium voltage (MV) distribution network in its
structures, can occur even at very close distances to the line. inherent complexity cannot be directly evaluated.
The aim of this paper is to evaluate the lightning performance
Manuscript received May 16, 2016; revised July 8, 2016; accepted August
10, 2016.
of a power distribution network considered in its complexity
M. Brignone, F. Delfino, R. Procopio, and M. Rossi are with the Naval, and accounting both for direct and indirect strikes.
ICT, and Electrical Engineering Department, University of Genoa, Genova In order to reduce the computational effort, a methodology
I-16145, Italy (e-mail: massimo.brignone@unige.it; federico.delfino@unige.it;
renato.procopio@unige.it; mansueto.rossi@unige.it).
is proposed in order to limit as much as possible the analysis
F. Rachidi is with the EMC Laboratory, Swiss Federal Institute of Technology, performed with the algorithm presented in the companion paper
Lausanne CH-1015, Switzerland (e-mail: farhad.rachidi@epfl.ch). [9]. Moreover, the importance of the instant in which the light-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
ning occurs is highlighted, showing that, for very low insulation
Digital Object Identifier 10.1109/TEMC.2016.2601657 levels, the maximum overvoltage depends on such instant, but
0018-9375 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
The procedure for the evaluation of the lightning performance A = [xm in , xm ax ] × [ym in , ym ax ] (1)
of power system has been proposed in [3] and [4] and is briefly
where xm in = −L/2 and xm ax = 3L/2 (see [3], [4]). ym in and
recalled in what follows.
ym ax can be chosen as follows:
1) A large number of lightning events ntot is randomly gen-
1) ym in can be chosen as the value under which, for any
erated. Each event is characterized by the point of impact
peak current [10], [11], the power system will always ex-
PF = (xF , yF ) and the channel-base current amplitude
perience a direct strike. Recalling the electrogeometric
I0 . According to [10] and [11], the current is assumed to
criterion described in the companion paper [9], one can
follow a log-normal probability density function, while
express the lateral distance as a function of the peak cur-
the points of impact coordinates are uniformly distributed
rent. As can be expected, such a function monotonically
into a striking area containing the power system of interest
increases; so, the value corresponding to the minimum
and all the possible lightning event that can cause CFOs
current can be selected as ym in . In the following, ym in
[3], [4].
will be set as 15.00 m, which corresponds to about 2 kA
2) The application of the electrogeometric model (EGM)
(that has a probability smaller than 0.016% according to
[12] allows us to choose whether the selected event is
the current lognormal probability density function);
a direct or an indirect stroke. In any case, the event is
2) ym ax can be chosen as the horizontal distance at which,
processed as outlined in the companion paper in order to
no matter what the peak current is, the resulting overvolt-
evaluate the maximum overvoltage generated.
age will always be smaller than the CFO. According to
3) Such overvoltage is compared with the line CFO and a
the Rusck–Darveniza formula [12], one can set ym ax as
counter n is increased by one unit if the overvoltage is
follows:
greater than the CFO.
4) The ratio between the final value of the counter n and ntot 38.80 0.15
ym ax = h+ √ max {I0 } . (2)
is computed in order to evaluate the probability of having CFO σg
a dangerous overvoltage.
In the following, ym ax is assumed to be 1005.0 m, correspond-
ing to the most critical situation characterized by a peak current
B. Reduction of the Computational Effort of 100.0 kA (that has a probability smaller than 0.0038% ac-
The proposed statistical approach requires a very large num- cording to the current lognormal probability density function),
ber of calls to the code that calculates the indirect overvoltages, a ground conductivity of 0.001 S/m and a CFO of 50 kV.
thus requiring a prohibitive computational effort. In this sec- 2) Evaluation of the Maximum Overvoltage: Setting rm in =
tion, an efficient technique to reduce the number of such calls is ym in and rm ax = x2m ax + ym 2
ax , one can construct the light-
presented. ning electromagnetic field database through steps 1–3 of the
1) Choice of the Domain for the Points of Impact: As out- algorithm presented in the companion paper [9]. As the appli-
lined in the previous subsection, first of all a suitable area A has cation of steps 4–9 of the same algorithm allows us to find out,
to be chosen containing the points of impact of all the possible for any i = 1, . . ., M , the function Vi (x, t) for a given lightning
events that can cause dangerous overvoltages. peak current and point of impact, with a simple postprocessing
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
BRIGNONE et al.: EVALUATION OF POWER SYSTEM LIGHTNING PERFORMANCE—PART II: APPLICATION TO AN OVERHEAD 3
Fig. 2. Illustration of the interactive process for the grid definition: (a) lossy ground and Case-1, (b) PEC ground and Case-1, (c) lossy ground and Case-2, and
(d) PEC ground and Case-2.
Fig. 3. Comparison of the proposed approach with the ones in [3] and [4]: (a) lossy ground and Case-1, (b) PEC ground and Case-1, (c) lossy ground and Case-2,
and (d) PEC ground and Case-2.
to 9%, (for low-insulation levels) while the lossy case needs Monte Carlo replications, the overall number of code calls will
more computational effort, as the percentage to obtain a reliable always be 105.
result for a CFO equal to 50 kV reaches 23%. The next test is relevant to the analysis of the effect of direct
It should be observed that, in principle, for each Monte strikes. The number of events that has led to direct strikes has
Carlo replication, the analysis conducted with the method pro- a mean value μ = 312 with a variance σ = 19 for case (a),
posed in [4] should require new calls to the coupling code, μ = 314 and σ = 22 for case (b), μ = 315 and σ = 14 for case
which increases the overall computational effort, while the (c) and μ = 315 and σ = 17 for case (d). As described in the
proposed approach instead relies on the interpolation of the companion paper [9], the developed code is able to automatically
predefined grid, which means that, no matter the number of distinguish between direct and indirect events; as a consequence
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
BRIGNONE et al.: EVALUATION OF POWER SYSTEM LIGHTNING PERFORMANCE—PART II: APPLICATION TO AN OVERHEAD 5
Fig. 4. Accounting for direct strokes in the lightning performance evaluation: (a) lossy ground and Case-1, (b) PEC ground and Case-1, (c) lossy ground and
Case-2, and (d) PEC ground and Case-2.
Fig. 5. The ratio between the probability of having an indirect overvoltage greater than CFO over the probability of having an overvoltage greater than CFO is
presented: (a) lossy ground and Case-1, (b) PEC ground and Case-1, (c) lossy ground and Case-2, and (d) PEC ground and Case-2.
it can be used in order to analyze the effect of direct strokes on the four cases. Examining the figure, one can conclude the
the overall lightning performance. following.
In all the situations, it is apparent that assuming that each 1) As expected (see [8]), all the curves show a decreasing
direct strike causes flashover [6] is effective, while neglecting trend as a function of CFO, which means that the effect
the direct strokes results in an underestimation of the lightning of indirect strikes is more meaningful for low-insulation
performance (see Fig. 4). levels.
In order to check whether the influence of direct strokes 2) The lossy ground enhances the effect of the indirect strikes
depends on the insulation level, the ratio between the prob- that provide a nonnegligible contribution even for very
ability that a direct event will cause a dangerous overvolt- high CFOs, while in the case of a PEC, for insulation
age and the overall probability is plotted in Fig. 5 for levels greater than 150 kV, only direct effects matter.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
The aim of this section is to evaluate the lightning perfor- fRES = 325A0.275
N /VN0.95 (8)
mance of the MV distribution network represented in Fig. 6, fRES = 220A0.36 0.95
N /VN (9)
where the HV/MV and MV/LV transformers are protected by
surge arresters. From a geometrical point of view, the MV line where AN is the (phase) machine rating (expressed in MVA)
is the same as the one analyzed in the previous section (see and VN is the rated (L-G) voltage (expressed in kV).
Table I). The two laterals are, respectively, at 250 and 750 m So, one can draw the Bode diagram of the transformer in-
from the HV/MV transformer and consist of two MV/LV iden- put impedance as a function of the unknown capacitances and
tical transformers with two 300-m long LV cables. The HV/MV adjust their values in order to obtain a resonance frequency as
transformer is fed by a 132 kV network with infinite short-circuit close as possible to the values predicted by (8) or (9). Start-
power (i.e., an ideal voltage source). First, some considerations ing from the classical typically available data of a transformer
on the transformers and arresters modeling are presented. (i.e., short-circuit voltage, copper losses, no-load current, and
no-load losses), one can evaluate the short-circuit resistance rcc
A. Arresters and reactance xcc (at industrial frequency fr ) as well as the
iron core resistance rm and the magnetizing reactance xm (at
According to the indications reported in [14], the surge ar- industrial frequency). As a consequence, one obtains
resters can be modeled using a nonlinear V–I characteristics,
thus not considering a frequency-dependent model. In the fol- 1 xcc
Z1 (ω) = Z2 (ω) = rcc + jω (10)
lowing, such characteristics have been chosen according to a 2 2πfr
typical ABB 15 kV arrester [15]. jωrm xm
Zm (ω) = . (11)
2πfr rm + jωxm
B. Transformers
Examining Fig. 7, it is apparent that the transformer
For the evaluation of the lightning overvoltages, the imple- impedance (seen at the primary side with the secondary one
mentation of an adequate high-frequency model for HV/MV short circuited) is given by
and MV/LV transformers is of crucial importance [16]. The
Z1 (ω) + Zp (ω)
most complex and accurate models have been proposed in ZT 1 (ω) = (12)
[16] and [17] and basically perform a rational approximation 1 + jω (C10 + C12 ) (Z1 (ω) + Zp (ω))
of the frequency response of the transformer admittance ma- where
trix. However, such models require specific measurements in Z2 (ω) Zm (ω)
order to obtain the suitable entries for the matrix. A more sim- Zp (ω) = . (13)
Z2 (ω) + Zm (ω)
plified approach consists of either substituting the traditional
low-frequency model with a π-circuit made of capacitances, Thus, one can adjust the values of C10 + C12 such that the
or to combine the low-frequency model with suitable capaci- peak of ZT 1 occurs at the frequency predicted by (8) or (9).
tors. As shown in [16], the second option provides better results Repeating the procedure for the secondary side, it is possible to
and, most of all, also allows us to represent correctly the initial get an estimate of C20 + C12 . As an example, let us consider the
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
BRIGNONE et al.: EVALUATION OF POWER SYSTEM LIGHTNING PERFORMANCE—PART II: APPLICATION TO AN OVERHEAD 7
Fig. 9. Comparison of the proposed approach with the ones in [3] and [4] in
the evaluation of the lightning performance of a complex distribution network.
TABLE II
MAXIMUM OVERVOLTAGE VERSUS PRECONTINGENCY MAXIMUM VOLTAGE
As far as the lightning performance is concerned, in principle, [5] Guide for Improving the Lightning Performance of Electric Power
the lightning strike time instant should be considered as an Overhead Distribution Lines, IEEE Standard 1410, 2011.
[6] F. Tossani, F. Napolitano, A. Borghetti, and C. A. Nucci,“Estimation of
additional variable of the Monte Carlo simulations. This allows the influence of direct strokes on the lightning performance of overhead
us to conclude that the influence of the instant of the occurrence distribution lines,” in Proc. IEEE Eindhoven Powertech 2015, 2015.
of the lightning discharge is substantially negligible for all the [7] J. Chen and M. Zhu, “Calculation of lightning flashover rates of overhead
distribution lines considering direct and indirect strokes,” IEEE Trans.
values of CFO of practical interest. Electromagn. Compat., vol. 56, no. 3, pp. 668–674, Jun. 2014.
Finally, a comment on the computational aspects of the pro- [8] P. N. Mikropoulos and T. E. Tsovilis, “Statistical method for the evaluation
posed approach is in order. Using a Microsoft Windows 8.1 PC of the lightning performance of overhead distribution lines,” IEEE Trans.
Dielectr. Electr. Insul., vol. 20, no. 1, pp. 202–211, Feb. 2013.
equipped with 8 Gb of RAM and Intel Core i7-2600 CPU at [9] M. Brignone, F. Delfino, R. Procopio, M. Rossi, and F. Rachidi, “Evalua-
3.4 GHz, the proposed algorithm requires about: 420 s for the tion of power system lightning performance—Part I: Model and numerical
field database creation (it should be underlined that such step solution using PSCAD–EMTDC,” IEEE Trans. Electromagn. Compat.,
vol. 59, no. 1, in press.
has to be performed only once for a given network), and less [10] K. Berger, R. B. Anderson, and H. Kröninger, “Parameters of lightning
than 1 s for the overvoltage computation (with a time step of flashes,” Electra, vol. 41, pp. 23–37, Jul. 1975.
0.01 μs and a simulation length of 0.01 ms). [11] Cigré Working Group 01 of SC33, Guide to procedures for the lightning
performance to transmission lines, Cigrè, Paris, France, Oct. 1991, Ref.
no. 63.
V. CONCLUSION [12] M. Darveniza, “A practical extension of Rusck’s formula for maximum
lightning induced voltages that accounts for ground resistivity,” IEEE
A methodology to evaluate the lightning performance of an Trans Power Del., vol. 22, no. 1, pp.605–612, Jan. 2007.
MV distribution network was presented, starting from the plat- [13] F. H. Silveira and S. Visacro, “On the lightning-induced voltage ampli-
tude: First versus subsequent negative strokes,” IEEE Trans. Electromagn.
form developed in the companion paper. First, a technique to Compat., vol. 51, no. 3, pp. 741–747, Aug. 2009.
reduce the overall computational effort was presented and vali- [14] A. F. Imece et al., “Modeling guidelines for fast front transients,” IEEE
dated against other methods available in the literature. Then, the Trans. Power Del., vol. 11, no. 1, pp. 493–506, Jan. 1996.
[15] ABB Division Surge Arresters, ABB Application Guidelines, 5th revised
method was applied to a realistic MV network characterized by ed., “Overvoltage protection: Metal oxide surge arresters medium voltage
a complex topology and by the presence of nonlinear devices. systems,” Wettingen/Switzerland, (2011, May). [Online]. Available:
Finally, the effect of the instant of the occurrence of the lightning https://library.e.abb.com/public/70e9fd6933c8c644c12578d200333cb5/
952_abb_awr_mittelspannung_E_low.pdf
discharge was analyzed both on the system overvoltage related [16] A. Borghetti, A. S. Morched, F. Napolitano, C. A. Nucci, and M. Paolone,
to a specific event and on the overall lightning performance. “Lightning induced overvoltages transferred through distribution power
transformers,” IEEE Trans. Power Del., vol. 24, no. 1, pp. 360–372,
Jan. 2009.
ACKNOWLEDGMENT [17] F. Delfino, R. Procopio, and M. Rossi, “High-Frequency EHV/HV au-
totransformer model identification from LEMP test data,” IEEE Trans.
The authors would like to thank Prof. M. Paolone from DESL Power Del., vol. 26, no. 2, pp. 714–723, Apr. 2011.
Laboratory, Swiss Federal Institute of Technology, for his valu- [18] W. J. McNutt, T. J. Blalock, and R. A. Hinton, “Response of transformer
able comments and suggestions to improve the quality of the windings to system transient voltages,” IEEE Trans. Power App. Syst.,
vol. PAS-93, no. 2, pp. 457–467, Mar. 1974.
paper. [19] PSCAD-EMTDC version 4.5, The Professional’s Tool for Electromag-
netic Transients Simulation, Manitoba HVDC Research Centre Inc.,
REFERENCES Manitoba, Canada.
[1] C. A. Nucci, “A survey on Cigré and IEEE procedures for the estimation Massimo Brignone’ photograph and biography not available at the time of
of the lightning performance of overhead transmission and distribution publication.
lines,” in Proc. Asia-Pacific Symp. Electromagn. Compat., Apr. 2010,
pp. 1124–1133.
[2] Guide for Improving the Lightning Performance of Electric Power Over-
head Distribution Lines, IEEE Standard 1410, 2004. Federico Delfino’ (M’XX) photograph and biography not available at the time
[3] A. Borghetti, C. A. Nucci, and M. Paolone, “An improved procedure for of publication.
the assessment of overhead line indirect lightning performance and its
comparison with the IEEE Std. 1410 method,” IEEE Trans. Power Del., Mansueto Rossi’ photograph and biography not available at the time of publi-
vol. 22, no. 1, pp. 684–692, Jan. 2007. cation.
[4] A. Borghetti, C. A. Nucci, and M. Paolone, “Indirect-lightning perfor-
mance of overhead distribution networks with complex topology,” IEEE Farhad Rachidi’ (F’XX) photograph and biography not available at the time
Trans. Power Del., vol. 24, no. 4, pp. 2206–2213, Oct. 2009. of publication.