1

Medium-Voltage Arc Flash in Switchgear

and Live-Front Transformers
M. L. Eblen, Member, IEEE, T. A. Short, Senior Member, IEEE and W.-J. Lee, PhD, Fellow, IEEE
engage fully with the stationary primary contacts inside the
Abstract—Arc flash on distribution circuits is a safety issue that cubicle. For breakers that “rack up,” the racking mechanism
can impact work practices for line and substation workers. This picks up the breaker and moves the breaker farther up
paper shows test results for arc flash events in medium-voltage
vertically inside the cubicle until the movable primary
switchgear cubicles and live-front transformer enclosures. Both
scenarios can result in more incident energy than the prediction contacts on the breaker engage fully with the stationary
of IEEE 1584-2002. Much higher than expected energies were primary contacts inside the cubicle.
measured from rack-in style switchgear breaker cubicles due to
A. Test Setup
the orientation of the electrodes. Live-front transformer cubicles
showed wide variability that was not directly related to the Testing was performed at the PG&E high-current testing
enclosure size or arc directionality. Unexpectedly high energies facility in San Ramon, California. The fault source was
were measured in multiple configurations. For selection of the created from a 12-kV ungrounded delta circuit fed by a direct
appropriate clothing systems, incident energy multipliers are
utility connection with a maximum available fault current of
suggested based on IEEE 1584-2002.
9.1 kA.
Index Terms—Arc flash, power distribution, safety, personnel For measuring incident energies and heat rates, copper slug
protection. calorimeters were installed. These are thin copper plates with
a thermocouple attached. The temperature change measured
I. INTRODUCTION by the thermocouples before and after the arc flash event is
Arc flash analysis and protection on distribution circuits is still used to estimate the incident energy to the copper plate.
evolving. More test data, better understanding of arc physics, Calorimeters were built and calibrated according to ASTM
and more industry experience with arc flash will help utilities specifications [1, 6]. Incident energy measurements were
protect workers. In this paper, we discuss results of arc flash made at 36, 48, or 58 in (91, 122, or 147 cm) away from the
testing of, switchgear enclosures, both rack-in (horizontal arc initiation plane.
electrodes) and rack-up (vertical electrodes) styles. Five B. Horizontal, Rack-in Style
different styles of live-front transformers were tested,
Three older-style rack-in switchgear cubicles were obtained.
including single-phase and three-phase units. The severity of
These units were single breaker full size cubicles with the
an arc flash event is normally quantified as the incident energy
breaker going in the bottom half of the switchgear cubicle.
that would reach a worker, normally given in terms of cal/cm2.
The stationary primary contacts (electrodes) inside the cubicle
Flame-resistant (FR) clothing systems have an arc thermal
were surrounded by a ceramic insulating bottle. The bottle
performance value (ATPV) rating or an energy break-open
assemblies are normally covered by a phenolic bottle barrier
threshold (Ebt), based on ASTM test standards [1]. The
and shutter assembly that shields the normally energized parts
ATPV rating is the incident energy in cal/cm2 on the clothing
from direct contact when the breaker is removed from the
surface that has a 50% probability of causing a second-degree
cubicle. The shutter assembly was removed for testing, but the
skin burn, and Ebt is the 50% probability of break-open. The
bottle barriers were left in place for the first test shot in each
goal of an arc flash analysis is to ensure that workers have an
cubicle section. See Figure 1. The interior dimensions of the
arc rating sufficient to handle the incident energy that might
cubicle were 33×69.5×39 in (83×142×149 cm) with the line
be expected in a given work scenario.
side stationary contacts pointing out of the back wall of the
To estimate incident energies on enclosed distribution
cubicle towards the front with 10 in (25 cm) of separation
equipment, most utilities follow the assumptions and approach
between phases and approximately 30 in (76 cm) up from the
used in IEEE 1584-2002 for arc flash in equipment [2].
bottom of the cubicle. The “rack-in” style was also tested with
II. ARC FLASH IN MEDIUM VOLTAGE SWITCHGEAR a simulated breaker inserted into the cubicle.
Metal-clad switchgear is manufactured so that the circuit C. Vertical, Rack-up Style
breakers are rolled into the cubicles until they engage with the Three full-size cubicles with the breaker going in the bottom
racking mechanism. Metal-clad switchgear can be separated of the switchgear and racking up after insertion were tested.
into two main categories. For breakers that “rack in,” the The shutter assembly was removed for testing. See Figure 2.
racking mechanism picks up the breaker and as the racking The interior dimensions of the cubicle were 36×56×58.5 in
gear turns, the breaker moves farther into the cubicle (91×142×149 cm) with the line side stationary contacts
horizontally until the movable primary contacts on the breaker
 2
pointing down from the top of the cubicle at approximately 24 cubicle instead of the breaker side of the cubicle. This is
in (61 cm) from the front edge. There was 9.5 in (24 cm) of distinctly different from the rack-in style. See Figure 3 and 4.
separation between phases. Incident energy measurements were made in front of the
cubicle

Line Side Primary

Stationary Contacts
with Shutter
Assembly

Fig. 3. Horizontal, rack-in style; stationary contacts shutter

removed

Fig. 1. Horizontal, rack-in style switchgear cubicle tested

Fig. 4. Vertical, rack-up style; stationary contacts shutter

removed
E. Test Results: Horizontal, Rack-in Style
To initiate the test, a solid #10 AWG copper wire was
wrapped from A to B to C phase on the line-side stationary
Line Side contacts. Current and durations were varied. In addition, a
Stationary calorimeter sensor head was used simultaneously with the
Contacts
with Shutter calorimeter array to evaluate arc flash boundary questions.
The sensor head was placed in the back of the test chamber,
making the face sensors approximately 10 ft. (3 m) away from
the arc and approximately 6 ft. (1.8 m) off the ground, which
is a good approximation of where another worker might be.
When the calorimeter array was spaced 36 in (0.9 m) from
the back wall of the enclosure, it was about 3 in (7.6 cm)
inside the front plane of the opening. During the longer
events—30 cycles or longer—a large arc plasma (fireball)
projected out of the enclosure and enveloped the calorimeters.
See Figure 5.
Fig. 2. Vertical, rack-up style switchgear cubicle tested
Arcing for all events was very similar. As shown in Figure
The stationary primary contacts were potted into a single
6, the arc starts between phases with the fuse wire but quickly
metal flange for each of the three phases for both the line side
moves to the case and arcs around the bottom of the CTs and
and the load side. The majority of the ceramic bottles
to the side wall.
surrounding the stationary contacts were on the bus side of the
 3
less to be ejected out of the cubicle onto the calorimeters. Due
to the limitations of available hardware and destructiveness of
the testing, the three samples were used to their maximum
limitations i.e. much of the testing was conducted with broken
bottles. See Figure 7.

Fig 7. Bottles intact TOP vs bottles broken BOTTOM

Measured heat rates were fairly linear with time and with
magnitude of arcing current. However, the Figure 8 shows
that measured incident energies significantly exceeded the
IEEE 1584-2002 switchgear formula. Multipliers of 3.0 to 4.0
would be needed to accurately predict resultant incident
Fig. 5. Test ID 6: 5.7 kA, 30 cycles, 13 cal/cm2 max, 10.4
energies. See Figure 8. Longer events would be more likely to
cal/cm2 avg., 25.6 cal/cm2/sec heat rate
be at the higher end multiplier (a 3.0 multiplier encompasses
all the events that were 20 cycles or less in duration). See
Table 1 for the overall probability that each multiplier
encompasses all the test data points.

Fig.6. Video frames from test ID 40 for the first six frames
(300 fps) viewed through an infrared-passing filter
When the ceramic bottles were in place, the arcing pattern
was similar, but the arc lengths were extended. Even with
extended arcs, the measured incident energies were reduced. Note 1: Bubble size reflects arc durations of 5.8, 10, 12, 20, 30, 45, 75
This is due to the absorption of the energy by the ceramic cycles respectively. Lines represent multipliers to IEEE 1584-2002.
bottles. The ceramic always broke, and in the process, some of Fig 8. Horizontal switchgear: measured vs. predicted energy
the energy was used to heat and break the ceramic, leaving
 4
TABLE 1
PROBABILITIES OF ENCOMPASSING TEST DATA
Multiplier to IEEE 1584-2002 Probability
3.0 65%
3.5 89%
4.0 98%

Figure 9 shows that the heat rises significantly above the

sensor head (located 10 ft. [3 m] away from the arc). This is
true for all cases. The measured energy at the sensor head was
a fraction of what IEEE 1584-2002 predicts. This will
significantly reduce the arc flash boundary predictions from
those previously done. Actual measurements at 10 ft. (3 m)
varied from 40% to 90% of what IEEE 1584 would predict, Fig 10: More energy from sides and bottom, test ID #48
with the bulk at 50% to 60%. This is an indicator that the
“distance factor” relationship is finite at some boundary point. The presence of an object in the cubicle reduced the amount of
Beyond 5 ft. (1.5 m) away from the arc, a different formula energy ejected. However, the values generally exceeded the
would need to be developed to determine the real arc flash IEEE 1584-2002 predictions but with smaller multipliers. As
boundary (the point at which incident energy would fall below shown in Table 2, the multipliers varied from 1.0 to 1.8. This
2 cal/cm2, the OSHA limit for FR clothing protection) appears to be a function of how far the simulated breaker was
[Paragraph (l)(8) of §1910.269, 2014 and Paragraph (g) of inserted into the cubicle. More energy is ejected with the
§1926.960]. breaker farther out of the cubicle. This confirms the
conclusions with the laboratory arc-in-a-box mock up done
Sensor
during previous testing [4].
Head

TABLE 2
MULTIPLIERS REQUIRED
(SWITCHGEAR WITH SIMULATED BREAKER)
Multiplier to IEEE 1584-2002 Test ID
1.8 41
1.8 42
1.6 43
1.5 46
0.8 47
1.2 48
1.0 49

Fig. 9. Test ID #134- 3.3 kA, 45 cycles, calorimeter array at F. Test Results: Vertical, Rack-up Style
58 in (147 cm) from arc, image at or very near end of the event
To initiate the test, #16 AWG solid copper wire was wrapped
from A to B to C on the line side of the stationary contact.
Current research by the IEEE/NFPA collaboration project has
The bottle assemblies were only able to withstand one test
confirmed the effect of electrode configuration and enclosure
shot each, so test data was limited to the 12 available bottle
geometry on increasing incident energy beyond the values
predicted by IEEE 1584-2002. IEEE Std. 1584 is currently in assemblies.
the revision process to develop new more representative As expected, with a vertical electrode configuration, the
formulas for these conditions. Higher incident energies with emission of arc plasma (fireball) out of the cubicle took longer
horizontal electrodes matches other test results [4, 7]. than for the horizontal electrode configuration. For 5.7 kA, it
Previous testing [4] concluded that if an obstacle (circuit took six cycles for the fireball to reach 36 in (91 cm), as
breaker) was inside the switchgear, much of the incident energy compared to about three cycles for the horizontal electrodes.
would be diverted or absorbed by the obstacle. To verify this Higher currents also eject energy faster than lower currents.
conclusion, a simulated breaker was constructed. As shown in For 5.7 kA, it took three cycles to reach 24 in (61 cm) and six
Figure 10, the presence of the object forced most of the cycles to reach 36 in (91 cm). However, for 9.1 kA, it took
incident energy out the sides and bottom of the cubicle.
only one cycle to reach 24 in (61 cm) and only three cycles to
reach 36 in (91 cm). See Figure 11 for a side-by-side
comparison of Test ID #52 and Test ID #55. Figure 11 also
 5
demonstrates higher overall energy. The end of the 9.1-kA
event appears “brighter” than the end of the 5.7-kA event.
5.7-kA Event 9.1-kA Event

Test ID #52: End of Cycle 3 Test ID #55: End of Cycle 1

Test ID #52: End of Cycle 6 Test ID #55: End of Cycle 3

Fig. 12. Rack-up style switchgear: IEEE 1584-2002 predicted

versus measured incident energy

III. ARC FLASH IN MEDIUM VOLTAGE LIVE FRONT

Test ID #52: End of Cycle 20 Test ID #55: End of Cycle 20
TRANSFORMERS
Fig. 11. Side-by-side comparison, 5.7 kA on left, 9.1 kA on Live-front transformers vary greatly in size and configuration.
right Five different types of enclosures and transformer sizes were
chosen that span from the smallest to the largest that are likely
Figure 12 shows that measured incident energies were below to be encountered in the field. The selected transformers were
the predicted values of the IEEE 1584-2002 switchgear drained of oil and the windings were disconnected from the
formula. The trend line is showing more conservatism as fault bushing compartment. Fault initiation was single-line-to
currents increase. However, additional testing would be needed grounding the primary bushing compartment to replicate the
to confirm this trend. most likely scenario of an actual field initiated arc. Multi-
IEEE 1584-2002 has a 20% difference between grounded phases were energized to determine if the arc would propagate
and ungrounded switchgear predictions. This is done from the on its own beyond the single phase started. Not all
original laboratory testing that suggested 20% of the energy transformers were rated for 21 kV line to line, but the bushing
returned via the ground connection and was not available for sizes and spacing allowed the primary bushings to be
heat production. Figure 12 was overlaid with the IEEE 1584- energized with a 21-kV grounded-wye source. Results are
2002 grounded switchgear predicted energy to show the expected to apply for similar equipment for voltages from 4
relative difference between the ungrounded and grounded kV to 35 kV.
predictions. Even though this switchgear was tested with an
ungrounded source, the IEEE 1584-2002 grounded prediction
envelopes all but one measured value. A. Test Setup
While the existing IEEE formulas were reasonably accurate The transformers tested are shown in Fig; 13 through 17 by the
for 24 and 36 in (61 and 91 cm), measurements at the sensor order they were tested.
head (92 and 100 in [2.3 and 2.5 m]) were less than 20% of
what IEEE 1584-2002 predicts for that distance. Beyond 5 ft.
(1.5 m), using a distance factor of 2 might be more predictive
of the arc flash boundary.
 6

Fig. 13. Specimen #1: 3Φ 12kV-120/208V 75 kVA (small

one) style IIA
Fig. 17: Specimen #5: 3 Φ 12kV x 21kV-277/480V 750kVA
(large one) radial style
Specimens #1 and #5 had pin terminals that pointed
vertically while the others had pin terminals facing
horizontally. The pins were bent at a right angle to left to fit
into the horizontal terminal. See Figure 18.

Fig. 14: Specimen #2: 1Φ 4160/7200 X 12000V-120/240V

50-kVA clamshell with 3Φ cabinet

Fig. 18: Horizontal (left) specimen #3 vs vertical (right)

specimen #5 pin terminals
Specimen #1 had a set of standoff bushings installed from
the top down in front of the primary bushings coming out of
the transformer on the back wall. These were left in place to
see if they had a deflecting influence on the heat. See Figure
19.

Fig. 15. Specimen #3: 1Φ 4160 X 12000V-120/240V Standoff

Bushing
37.5kVA OLD style with 3Φ cabinet

Primary
Bushing

Fig. 16: Specimen #4: 1 Φ 12kV x 21kV-120/240V 50-kVA

box style with 1Φ cabinet Fig. 19: Specimen #1 primary bushing compartment:
showing standoff bushing in front of primary bushing
 7
The dimensions and phase spacing for each specimen is
shown in Table 3.

TABLE 3
TEST SPECIMEN DIMENSIONS
Spc Box Dim PH Space Pin
# (inches) top/back/ph Config
WxHxD (inches)
1 30x41x24 14/8/10 Vert
2 28x29x18 9.5/8/10 Horiz
3 26x26x17 8.5/9/9 Horiz Specimen #1: Test ID 64 end of event Specimen #2: Test ID 74 End of event
4 17x27x18 8.5/6.75/9 Horiz
5 41x69x32 24/8.5/11.5 Vert

A #20 AWG copper wire was used to start a single phase arc
for all bushing compartments. The wire was wrapped around
the pin terminal and attached to the nearest grounded metal,
preferably near the front of the enclosure. If no readily
available spot was near the front of the cubicle, the wire was
connected to a bolt head on the back wall.
Fault current levels of 2.3 kA and 6.7 kA were chosen at the
beginning with durations of 20 and 40 cycles. After the Specimen #3: Test ID 85 end of event Specimen #4, Test ID 96 end of event
second specimen was tested, we discovered that the test values
were exceeding laboratory limits. The test plan was altered to
have 4.5 kA as the highest available fault current with a goal
of gathering some data at 60 cycle durations. All incident
energy measurements were made at 33 in (84 cm) from the tip
of the bushing.
B. Test Results
All transformers behaved very differently. The transformers
that had pin terminals pointing straight up tended to start with
arcing to the back wall of the enclosure, but some went to the
top. Both 3Φ transformers had this style of terminal, but both Specimen #5: Test ID 106 mid-event Specimen #5: Test ID 106 end of event
exhibited extremely different incident energy behavior. The Fig 20: Comparison of energy pattern from each specimen
largest cubicle had much higher incident energy expelled than
the smaller cubicle (opposite of expectations). The Recommendations in previous EPRI reports [4, 5] were to
transformers that had pin terminals pointed to the left tended use IEEE 1584 for all arc-in-a-box scenarios for voltages from
to start with the arc bouncing off the left side wall and 1 to 35 kV. For 25- and 35-kV class equipment, enter a voltage
attaching to either the roof or to a back wall protrusion. For of 10 kV in the IEEE 1584 spreadsheet, so the Lee Method is
these, the smallest did expel the most energy, but the values not triggered (the Lee Method gives incident energies that are
were higher than expected. unrealistically high). The measured values exceed predicted
Most specimens propagated to multiphase arcs on their own values using the IEEE 1584-2002 formula, as shown in Figure
very early in the test, except for the smallest and the largest 21. See Table 4 for the overall probability that each multiplier
enclosures, specimen #4 and specimen #5. These did not to IEEE 1584-2002 encompasses all the test data points.
propagate to multi-phase until the bushings were very
contaminated and beginning to erode. However, even with
single-phase arcing, these two expelled more incident energy
than any of the other specimens even when they went three
phase. The fireball ejected from the cubicle was not only
different in magnitude but also direction. Every specimen had
a first test of 2.3 kA for 20 cycles. Figure 20 shows the
comparison of incident energy patterns (fireball) for each
specimen.
 8

IV. CONCLUSIONS
The test results discussed here show that it is important to
evaluate the specific equipment or arc flash scenario to better
refine arc flash analysis approaches.
With horizontal bus configurations (rack-in style), incident
2. energies ranged from 1.5 times that predicted by IEEE 1584-
0 2002 up to 4.7 times the predicted value. 89% of the
measurements had energies less than 3.5 times the IEEE 1584
formula. Only 2% exceeded 4.0 times IEEE 1584 predictions.
With vertical bus configurations, incident energies were
between 0.5 and 1.0 times what IEEE 1584 predicts.
Rack-in switchgear with horizontal electrodes had much
No Multiplier higher incident energies than the switchgear with vertical
electrodes. With horizontal electrodes, the magnetic forces
push the arc and the fireball out the front of the enclosure.
With horizontal electrodes, the fireball goes downward, and
Note 1: Bubble size reflects arc durations 20, 40, 60 cycles respectively. Lines more energy stays in the enclosure. Figure 22 and Figure 23
represent multipliers to IEEE 1584-2002. compare similar events on the two types of switchgear tested.
Fig 21: Live-front transformers: IEEE 1584-2002 prediction Both events were done with Iarc = 8.4 kA and duration = 31
vs measured incident energy cycles. IEEE 1584-2002 predicts 6.9 cal/cm2 for these
conditions.
TABLE 4
PROBABILITIES OF ENCOMPASSING TEST DATA
Multiplier to IEEE 1584-2002 Probability
1.0 34%
2.0 87%
3.0 98%

Additionally, when separated by specimen, the multipliers

vary widely, as shown in Table 5.

TABLE 5
MULTIPLIER RANGE FOR EACH SPECIMEN
Multiplier Range to IEEE 1584-2002 Specimen
0.2 – 1.6 #1 Fig 22: Horizontal racking configuration: 22.1 cal/cm2
0.4 – 1.3 #2
0.4 – 1.5 #3
1.4 – 3.6 #4
1.1 – 2.5 #5

The measured energy at the sensor head was a fraction of

what IEEE 1584-2002 predicts. This will significantly reduce
the arc flash boundary predictions from those previously done.
Actual measurements at the sensor head (91 to 101 in or 2.3 m
to 2.5 m from the arc initiation point) varied from 10% to 50%
of what IEEE 1584-2002 would predict, with the bulk under
25%. Only specimen #4 had values that exceeded 25%. As
with the switchgear, this is an indicator that the IEEE ‘distance
factor’ relationship is finite at some boundary point.
Fig 23: Vertical racking configuration: 6.4 cal/cm2
 9
There was wide variability between incident energies for Distribution Lines as the working group chair. He is currently leading a series
of research projects on distribution arc flash. Email: [email protected].
different types and sizes of live-front transformers. To predict
the incident energy using the existing IEEE 1584-2002 Wei-Jen Lee received the B.S. and M.S. degrees from National Taiwan
University, Taipei, Taiwan, R.O.C., and the Ph.D. degree from the University
switchgear formula, a multiplier should be used. Measured
of Texas, Arlington, in 1978, 1980, and 1985, respectively, all in Electrical
multipliers ranged from 0.2 to 3.6. However, only 13% of the Engineering. He is currently a professor of the Electrical Engineering
measurements exceed 2.0, and only 4% exceeded 2.5. Department and the director of the Energy Systems Research Center. Prof. Lee
Additionally, the largest enclosure and the smallest enclosure has been involved in research on arc flash hazards and electrical safety, utility
deregulation, renewable energy, smart grid, microgrid, load forecasting, power
had incident energies that were typically more than the quality, distribution automation and demand side management, power systems
medium-sized enclosures; and with single-phase initiation, it analysis, online real time equipment diagnostic and prognostic system, and
is very likely that the arc will propagate to the other phases in microcomputer based instrument for power systems monitoring, measurement,
control, and protection. Since 2008, He has also served as the project manager
all but the largest enclosure.
for the IEEE/NFPA Arc Flash Research Project. Prof. Lee is a Fellow of IEEE
Future work could include: and registered Professional Engineer in the State of Texas. Email:
 Tests on other types of horizontal switchgear especially [email protected].
switchgear with non-ceramic electrode barriers.
 More testing on livefront transformers to determine if
there is any commonality to predict which ones would
need the higher multipliers
 Testing to determine arc flash boundary prediction
methods
If anyone would like copies of the test data reported here,
please request it from one of the authors.

ACKNOWLEDGEMENTS
In support of the PG&E laboratories, we would like to thank
the dedicated work of Ralph Seban, Dan Kaufman, and Ryan
Sparacino.
REFERENCES

[1] ASTM F1959, Standard Test Method for Determining the Arc Rating of
Materials for Clothing: ASTM International, 2006.
[2] IEEE Std 1584-2002, IEEE Guide for Performing Arc-Flash Hazard
Calculations.
[3] EPRI 1018693, Distribution Arc Flash: Analysis Methods and Arc
Characteristics: Electric Power Research Institute, Palo Alto, CA, 2009.
[4] EPRI 1022697, Distribution Arc Flash: Phase II Test Results and
Analysis, Electric Power Research Institute, Palo Alto, CA, 2011.
[5] Kinectrics, "State-of-Art Report on Arc Hazard Modeling," in
Distribution Arc Flash: Analysis Methods and Arc Characteristics:
Electric Power Research Institute, Palo Alto, CA, 2008.
[6] ASTM E 457 – 08, Standard Test Method for Measuring Heat-Transfer
Rate Using a Thermal Capacitance (Slug) Calorimeter: ASTM
International, 2008.
[7] Short, T. A. and Eblen, M. L., "Medium-Voltage Arc Flash in Open Air
and Padmounted Equipment," IEEE Transactions on Industry
Applications, vol. 48, no. 1, pp. 245-253, Jan.-Feb. 2012.

Marcia L. Eblen graduated from University of Colorado–Boulder in 1982 with

a BSEE degree. She is recently retired from Pacific Gas & Electric where she
worked for almost thirty years. Since 2002 she served as the Principle
Grounding and Arc Flash Engineer. She is a member of the IEEE Substation
Safety, IEEE ESMOL subcommittee, IEEE 1584 subcommittee, ASTM F18
Committee, and has been a voting member to the NFPA70E technical
committee since 2010. She is a registered professional engineer in the state of
California. She currently consults in arc flash and grounding through MLE
Engineering, Inc. Email: [email protected].
Tom A. Short is with EPRI in Burnt Hills, NY. Before joining EPRI in 2000,
he worked for Power Technologies, Inc. for ten years. Mr. Short has a Master’s
of Science degree in Electrical Engineering from Montana State University
(1990). Mr. Short authored the Electric Power Distribution Handbook (2ed,
CRC Press, 2014). In addition, he led the development of IEEE Std. 1410-
1997, Improving the Lightning Performance of Electric Power Overhead