Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Downloaded from https://iranpaper.

ir
https://www.tarjomano.com/order

A REVIEW OF COMMONLY USED DC ARC MODELS

Tammy Gammon, P.E. Wei-Jen Lee, P.E. Zhenyuan Zhang Ben Johnson
Senior Member, IEEE Fellow, IEEE Student Member, IEEE Life Fellow, IEEE
John Matthews & Associates University of Texas - Arlington Thermon
P. O. Box 3229 Energy Systems Research Center P. O. Box 609
Cookeville, TN 38502 Arlington, TX 76019 San Marcos, TX 78666


Abtract - The DC arc hazard is a great concern to most popular DC arc model is often believed to be the DC
industry. Quantitative arc-hazard assessments are equivalent of Ralph Lee’s theoretical three-phase AC arc
performed on DC systems to determine a nearby worker’s model; however, this model is not a DC replication of Ralph
potential incident-energy exposure during an arcing Lee’s work. The original Ralph Lee model is more
event. Four viable DC assessment methods are reviewed sophisticated than determining burn hazard directly from the
in this paper. The most widely used model for predicting electrical arc energy, which is the approach taken by the
DC incident energy is based on Lee’s theoretical arc widely used DC model and Ammerman’s DC arc model,
model; the electrical arc power is determined from the addressed in Section III. Ammerman’s method for
maximum power transfer theorem and the arc is depicted determining electrical arc energy is based on an empirically
as a spherical radiant source with uniform heat derived, previously published arc current-power relationship.
transmission in all directions. Like Lee’s model, Ammerman includes two equations for determining incident
Ammerman’s model assumes complete conversion of energy density – one for open air and one for enclosures. In
electrical arc energy into thermal energy, but arc power is Section IV, DC arc testing conducted at the Kinectrics High
determined from an iterative technique constrained by Current Laboratory and a published set of equations for 600-V
arc power and circuit characteristics. Ammerman DC transit systems are discussed. Additional examples,
incorporates multiplying factors which account for the including one on photovoltaic systems, are presented in
higher incident energies associated with arcing in Sections V and VI.
enclosures. Based on DC arc testing, the applicability of
an existing software package has been extended to DC
systems through multiplying factors and equations for II. RALPH LEE’S WORK AND THE SPHERICAL,
DC rail and transit systems and equations have also been RADIANT ARC MODEL
developed. Model derivation is examined in this paper for
suitability to arcing in general and DC specifically. Model One of the most widely used DC models is based on the
performance is assessed using the available, limited data maximum power transfer theorem. Ralph Lee used the
(AC or DC). Example calculations are provided. maximum power transfer theorem in developing a theoretical
arc model to predict the distances for “curable” and “incurable”
burns sustained from three-phase AC arcing faults [1]. The
Index Terms – DC arc model, Ralph Lee model,
“curable” burn distance is analogous to the “arc flash
Ammerman’s model, DC transit systems
boundary” distance used in IEEE 1584 and NFPA 70E. Lee
assumed that the arc behaved as a passive linear resistance
and the system impedance was purely reactive; therefore, by
I. INTRODUCTION the maximum power transfer theorem, the maximum AC arc
power is 50% of the bolted-fault VA. The maximum power
Industry is working at a fast pace to address the DC arc transfer theorem states that the maximum power is
hazard exposure for workers. Unfortunately, the demand to transferred to the load when the load impedance equals the
assess worker risk has outsped the development of accurate source impedance. For a DC circuit, the maximum power
assessment methods for DC power systems. At present the transferred to an arc is 25% of the bolted-fault VA, as
only quantitative methods, which have been verified or are illustrated in Fig. 1. Lee did not use the arc current to
based on representative DC arc testing, have been developed determine the response time of the overcurrent protective
at a well-known high current laboratory. The popular, widely device. His original “curable” and “incurable” burn distances
used theoretical model, based on Ralph Lee’s work is were based on an arc duration of 100 ms.
addressed in Section II. The section includes a brief review of Lee conceptualized the arc as a perfect sphere radiating
radiant heat projection and thermal energy absorption by heat, essentially a blackbody radiator. Thermal radiation is
human skin. With the theoretical model, the arc power is the emission of electromagnetic waves in the range of 100 nm
determined by the maximum power transfer theorem. The to 100 um, covering the infrared, visible and part of the
ultraviolet spectrums [2], as shown in Fig. 2. For an ideal
blackbody at rest, the specific distribution of the thermal
wavelengths only depends on its temperature. The rate of
radiant heat, q, is a function of temperature as defined by the

978-1-4799-2066-2/14/$31.00 ©2014 IEEE 34


Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Stefan-Boltzmann law [3] in (1) where A and T are the surface 100 ms, would sustain a “curable” arc burn [1, Table IV].
area and temperature of the transmitting body, respectively. (Please note that the duration of elevated skin temperature
does not necessarily coincide with the duration of heat
∙ ∙ , 5.6697 ∙ 10 (1) exposure. For an illustration, refer to Fig. 7 in companion
paper [6].) Lee produced quite remarkable work.
In contrast, the widely used DC arc model in (2) based on
the maximum power transfer theorem simply assumes that all
electrical arc energy is converted directly to an incident
2
energy density (J/cm ) absorbed by a surface at a distance, d.
, ⋅
(2)

Referring to the DC model in (2), the NFPA 70E, 2012 Edition,


states “it would be prudent to use a multiplying factor of three
for the resulting incident energy value [7].” Doan’s paper, the
premier reference on the subject, suggests that it would be
“prudent to consider using a multiplying factor” for equipment
enclosures [8]. The recommendations are based on three-
Fig. 1. Maximum power transfer to ac and dc circuits. phase, 600-V arc testing conducted by Doughty [9]; according
to [9, Fig. 9], the “results indicate that with the 22” W x 20” H x
21” D box, the incident energy is multiplied by a factor of 3 as
compared to the incident energy produced by open arcs.”
The tests were conducted using a 36.25-kA bolted-fault
current and vertical electrodes spaced 1.25 inches apart.
Enclosure tests are likely to result in higher incident energy
density than open air arc testing, but the actual energy
increase is also dependent on additional factors such as
system type and voltage, bolted fault current, gap width,
electrode orientation and enclosure dimensions.
Not all of the arc’s electrical power is converted to radiant
Fig. 2. Thermal radiation in the electromagnetic spectrum heat transmitted uniformly in all directions. (Even open air
as a function of wavelength (in meters) [2, Fig. 11-1]. arcs do not generate heat uniformly in all directions.) The arc
power is also dissipated in several ways. One reference
identified that roughly 40% of the arc energy heats the
Lee states that the temperature of the metal electrodes
surrounding air (which includes pressure rise [10]), 50%
associated with electric arcs have been reported at 20,000K
causes a temperature rise and UV radiation, and 10% is
(35,000°F). However, he continues, “The intermediate
transmitted to the electrodes [11]. Results from testing in
(plasma) part of the arc, the portion away from the terminals,
closed configurations with copper or aluminum electrodes
the “shank” of the dogbone, figuratively, is reported to have a
indicated that 40% was transformed to pressure rise and 10-
temperature of 13,000K.” The radiant energy absorbed
12% in heat conducted to the electrode materials [12]. It is
depends on the coefficient of absorption of the receiving
also interesting for the two tests involving copper electrodes in
surface and on the energy incident on that surface. Lee
[12], 27-34% of the arc energy was identified with conversion
further reported that the coefficient for clean, light-colored skin
to optical radiation, but researchers attributed the 17-24% of
is around 0.5 and approaches one for dirty or dark-colored
unaccounted arc energy to copper electrodes transferring
skin. However, it has also been reported by a more recent
short wave radiation into long wave radiation [13]. The
engineering guide that the skin’s absorption coefficient for
researchers stated that optical radiation was associated with
infrared radiation is between 0.94 and 0.99 [4]. Lee assumed
heating materials close to the arc [14]. Just as the human eye
that the absorption of the skin was unity. He also clearly
has varying sensitivity to the wavelength of visible light, the
stated that his “calculations do not take into account heat
absorptivity of the human skin is a nonlinear function of the
reflected from surfaces near the arc,” “the ignition of
wavelength of the thermal radiation. Skin absorptivity drops to
clothing…[which ignites] from 400°C to 800°C,” or droplets of
roughly 38% around 600 to 800 nm near the visible light
molten material which “at temperatures of 1000°C or more,
spectrum, but increases in both the ultraviolet and infrared
will ignite clothing instantly, and cause spot burns on contact
spectrums. Skin absorptivity is about 90% at 300 nm in the
[1].”
UV range; in the infrared range, it is about 90% at 1.6 um and
Lee’s derivation was based on the temperature rise in
increases to about 97% at 2 um (Fig. 12.23 in [15]).
human skin. According to his extrapolation of the Artz [5]
The electrical arc power is ordinarily less than the
curve (see Fig. 6 in companion paper [6] for Lee’s original
maximum power which can be transferred to the arc. Fig. 3
figure), Lee determined that raising the skin temperature to
illustrates that the arc power estimated from Hall’s DC rms arc
80°C (176°F) for 100 ms results in a “curable” skin burn.
current and voltage data [16] is less than the theoretical 25%
Since normal skin temperature is approximately 34°C (93°F),
maximum of the DC power supplied. (Note: the true arc
the temperature rise is 46°C. Based on Lee’s derivation, a
power is equal to time-integrated average power.) Fig. 4
person standing 20 inches away from a 5.2 MW arc lasting
shows that Hall’s measured DC arc currents are significantly

35
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

higher than 50% of the bolted fault current, as specified by the The IEEE 1584, three-phase AC, arc test data has also
maximum power transfer theorem. Furthermore, the been analyzed to determine if the power delivered to the arc
measured DC arc voltages are significantly lower than 50% of approaches the theoretical 50% maximum for AC systems
the supply voltage. It should be noted that the data in the [18], [19]. The average ratios of arc power to available VA
figures represent one limited set of tests conducted using a were 0.332 and 0.214, respectively, for 165 low voltage (“LV,”
325-V DC supply, a 3/8-in (9.5mm) electrode gap width and <1 kV) and 148 medium voltage (“MV,” >1 kV) tests [18]. The
series vertical electrodes in open air. Different DC supply medium voltage ratios were lower because the arc current
voltages, bolted-fault currents, gap widths, electrode approaches the magnitude of the bolted fault current and the
orientations and configurations are expected to result in a arc voltage is a relatively small percentage of the system
wide range of arc currents, arc voltages and incident energy voltage over the gap widths tested.
densities. It is very unwise to use the maximum power Lee’s original equation to determine the curable burn
transfer theorem to predict DC arc current and incident energy distance for three-phase AC arcs was modified to calculate
density. Accurate estimation of the arc current is essential in incident energy density (J/cm2) in the IEEE Standard 1584-
determining the expected response time of the overcurrent 2002 [20]. For the 1584-modified form in (3), the system
protective device. The DC time constant, which may delay voltage, bolted-fault current, time and distance to the arc are
the arcing fault current in reaching its full magnitude, also expressed in kV, kA, seconds and mm, respectively.
needs consideration in the operation of the protective device;
it has been addressed in [8], [17]. 2.142 ∙ 10 ∙ ∙ ∙ (3)

The “theoretical” equation in (2), attributed to Ralph Lee,


and Lee’s equation in (3) as presented in 1584 were both
used to predict the incident energy density for the 1584 test
results database (300+ LV and MV tests). The incident
energy density predicted using the “theoretical” equation in (2)
is 3.2 times higher than equation (3) for each test entry listed
in the database. A comparison of the “theoretical” incident
energy density in (2) with the maximum incident energies
measured during the 1584 testing and predicted by the 1584
equation is presented in Table I. The basis of comparison,
the average higher percentage of the theoretical equation
results, is defined in (4).


% (4)

The number of tests averaged in (4) is listed as the number of


test entries, “# Test Entries,” in the table. The last column in
the table specifies the number of test entries that are
Fig. 3. Hall Data and predicted values for Parc. represented in the average where (2) predicts lower incident
energy than the 1584 equation or data. Table I shows that the
“theoretical” solution is significantly higher than the 1584 data,
the 1584-modified Lee equation and the equations developed
for the 1584 Standard. The 1584 equations were
conservatively formulated to achieve a 95% confidence level
of calculated incident energy values falling below the correct
PPE hazard risk category ratings of 1.2, 8, 25, 40 and 100
cal/cm2 [20].
Based on Table I’s comparison of the theoretical model with
the 1584 AC arc data, multiplying the theoretical model in (2)
by a factor of 3 to predict the DC arc incident energy projected
from enclosures would unnecessarily elevate the DC incident
energy predictions.
TABLE I
IE COMPARISON - “THEORETICAL” & IEEE 1584-2002
# Test Avg % # Tests, Eqn (2)
IEEE 1584-2002
Entries Higher Predicts lower IE
R. Lee (in 1584) 314 222% 0
1584 LV data 166 175% 2
1584 LV eqn. 166 67% 0
Fig. 4. Hall Data and predicted values for Iarc and Varc. 1584 MV data 148 1769% 0
1584 MV eqn. 148 1600% 0

36
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

III. AMMERMAN’S WORK IN DC ARC MODELING TABLE II


MULTIPLIERS FOR DC ENCLOSURE INCIDENT ENERGY
Ammerman [21] developed a DC arc model based on the E1 = B ∙ Es, where B equals:
work of Stokes and Oppenlander [22], Fisher [23] and Wilkins Eqmt. a, B, at distances (in) from the arc:
[24]. Stokes and Oppenlander conducted over 200 open air, type mm k 18 24 30 36 42 48 60
single-phase AC and DC arc tests. They used a 6-kV power LVpanel 100 0.127 1.523 1.554 1.569 1.577 1.582 1.585 1.589
supply to generate 0.1 to 20-kA arc currents between series,
LVswgr 400 0.312 2.221 2.741 3.074 3.291 3.437 3.540 3.668
copper and aluminum electrodes spaced 5, 20, 100 and 500
mm apart. Stokes and Oppenlander observed that the MVswgr 950 0.416 0.983 1.525 2.047 2.514 2.916 3.253 3.765
instantaneous arc power tended to be more constant than the
instantaneous arc current or arc voltage; their empirical formulas for predicting the AC arc current and incident energy
power characteristic [22] in (5) was developed for AC and DC density generated from a three-phase arcing fault. The
arcs where the instantaneous arc power and current, and the accuracy of the developed equations was confirmed by
gap width are expressed in W, A and mm, respectively. statistical analysis. The equations in (7) are an intermediate
calculation in the Wilkins’ method [24]; (8) is the final equation
20 0.534g .
(5) for determining the incident energy density for a three-phase
AC arcing fault.
Ammerman developed a method for determining the DC
. . .
arc current, similar to the iterative technique developed by 114.9 ∙ , ∙ ∙ (8)
Fisher for AC arcing faults. The non-linear DC arc resistance
was derived from the arc power-current relationship in (5). In the previous equation, g is the gap width (mm), Vll is the
The DC rms “steady-state” arc current and resistance were line-to-line voltage (V) and the incident energy density is
calculated from the set of constraining equations in (6). calculated in cal/cm2. Before the incident energy density
component, Es or E1 from (7) can be calculated, the AC arc
.
20 0.534g (6) current and arc resistance is determined in an iterative
manner similar to Fisher [23]. Wilkins calculated phase-to-
neutral arc resistance with (9).

. . .
The DC arc current, voltage and power determined by ∙ 1.757 ∙ ∙ ∙ ,
Ammerman’s model, displayed in Figs. 3 and 4, more closely 0.821 , 1 (9)
match Hall’s data than the “theoretical” values previously
discussed. However, it must be remembered that the Stokes The ac arc current (line) is constrained by the phase-to-
equation was formulated for series (opposing) electrodes in neutral voltage, system impedance and arc resistance.
open air under fairly stable arcing conditions. Arcing at lower Equation (9) was formulated from a multiple regression fit of
voltages, wider gap widths, higher bolted-fault currents or with the available data set. The three-phase AC electrical arc
parallel electrodes may result in a lower Iarc/Ibf ratio and a energy was calculated and the spherical incident energy
different V-I characteristic. The presence of an enclosure may density component, ES, in (7) was defined; then, the maximum
tend to increase the Iarc/Ibf ratio. incident energy density equation in (8) for open air, ES, was
Ammerman used the equations in (7) based on Wilkins’ formulated from a “best-fit” statistical analysis. Finally the
work to determine the DC incident energy density, where d is values for a and k were determined for the enclosure incident
the worker distance, and the factors k and a account for the energy density component, E1, by a “least-squares best fit”
arc energy that is reflected from the enclosure walls and statistical analysis. The statistical analysis is based on an AC
projected to the front of the enclosure. data set which may not uniformly represent a wide cross-
section of electrical parameters or typical configurations
∙ (7) where an AC or DC arcing fault may occur. The determined a
and k values are based on a progressive, three-stage
, statistical analysis and a specific set of three-phase AC arc
4 data.
, The 1584 test database includes 54 LV “panel”
(14inx12inx7.5in) tests, 87 LV “switchgear” (20inx20inx20in)
tests and 78 MV “switchgear” (45inx30inx30in) tests. The
Inspired by Ammerman’s work, Fontaine [25] further calorimeter distance was 600-610 mm for 43 of 54 LV panel
extended Wilkins’ iterative technique and more tests, 78 of 87 LV switchgear tests and 56 of 78 MV
comprehensively presented the application of the Wilkins’ a switchgear tests. The 22 MV switchgear tests conducted at
and k factors in determining the incident energy projected other calorimeter distances involved only one bolted-fault
from enclosures. Wilkins’ constants and the effective current level at 15 kA. Indeed, the low-voltage multipliers at
multipliers used to calculate the DC enclosure incident energy 18 in to 24 in and the medium-voltage multiplier at 24 in may
density, E1, from the DC open air incident energy, Es, are be reasonable for vertical electrodes and for the three-phase
provided in Table II for various worker distances. AC 1584 arc data. However, the increase in the LV and MV
Wilkins used the 1584 data set (310 entries) and 37 switchgear multipliers as the distance from the arc increases
additional tests performed at Ferraz-Shawmut to develop is a concern. Data to validate the use of these multipliers,

37
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

even in three-phase AC systems, is sparse if not inadequate. incident energy density than Ammerman predicts for a panel,
There is no data to validate the use of these multipliers over but less than Ammerman predicts for switchgear. In
the wide range of typical applications in DC systems. This comparison with the incident energy densities predicted by the
paper does not support or endorse the use of these multipliers DC Ammerman model and the three-phase AC 1584 models,
at any arbitrary distance in DC arc modeling. As the Wilkins’ a 3x factor for enclosures, suggested for the DC theoretical
constants suggest, the effect of the arc distance and the model appears high. Since the 1584 equations have been
enclosure size certainly impact the level of incident energy developed to conservatively predict the incident energy
impressed on a surface. A deeper understanding is needed density for a three-phase AC arcing fault, the higher 1584
of the thermodynamic projection of heat and the relationships incident energy predictions are expected. The 1584 incident
between enclosure dimensions, electrode gap widths, energy predictions should be considerably higher than a
electrode-to-back-panel distance and electrode orientation, as single-phase DC arcing fault.
well as bolted-fault current and system voltage. Heat transfer
from the arc is dependent on the system type – DC and AC
single-phase or AC three-phase. The total heat generated IV. WORK IN DC ARC TESTING & MODELING
and the directional projection of the arc’s thermal energy are
impacted by the physical presence and combustive properties Kinectrics has conducted a DC arc testing program in their
of real world electrical equipment. High Current Laboratory. The range of test parameters
Like the widely used “theoretical” DC arc model, covered is listed in Table IV [26]. Based on the DC arc
Ammerman’s DC arc model assumes 100% conversion of testing, multiplying factors for DC systems have been included
TM
electrical arc energy into incident energy absorbed by the in ArcPro and verified for the range of bolted fault currents
receiving surface. In reality, the electrical arc energy is also from 2 to 25 kA covering gaps up to 0.5 in for 130 Vdc and up
converted to additional forms of energy (refer to Section II, to 2 in for 260 Vdc. The technical basis of the arc model used
paragraph 6). Furthermore, not all forms of thermal energy, in the commercial software is described in Appendix A.
such as visible light, are absorbed by the receiving surface The 600-Vdc arc testing was conducted with series
with the same efficiency. Human skin has a greater electrodes in open air and with parallel electrodes directed
coefficient of absorption for infrared wavelengths than visible either downward or outward (i.e., vertical or horizontal) in an
light (refer to Section II, paragraph 6). enclosure. The enclosure tests resulted in higher incident
Table III presents a comparison of results predicted by the energy measurements than the open air testing. The
“theoretical” DC and the Ammerman DC arc models, as well enclosure incident energy measurements were similar for the
as the 1584 three-phase AC arc equations. The Ammerman vertical and horizontal electrode orientations. The equations
DC model more closely matches the measured arc current, in (10) were developed [26] and apply to open air DC arcing
arc voltage and estimated arc power for the 1800-A, bolted- [27]. In (10), the bolted fault and arc currents are expressed
fault DC arc test conducted by Hall [16] than the theoretical in kA; the electrode gap, G, and calorimeter distance, D, are
model. The theoretical DC model predicts an incident energy expressed in inches. The arc duration time, t, is expressed in
density which is 66% higher than the Ammerman DC model seconds.
for open air. The theoretical model predicts a slightly higher
. .
0.9063 ∙ 0.1051 ∙ ∙ 1 (10)
6
TABLE III 0.9694 0.0589 0.4793 ln 1.0027
COMPARISON OF HALL DATA WITH DC & AC MODELS 0.1
a
Iarc Varc Parc IE
The sustainability of the arc at gap widths up to 6 inches
A V kW cal/cm2 depended on the supply voltage, the bolted fault current and
Hall DC Arc Data for Tests, g = 3/8 in, Vs = 325 V the electrode configuration. For all of 600-V tests conducted,
Ibf = 1800A 1500 50 75 the arcing fault current ranged between 64% and 97% of the
DC Arc Models bolted fault current. The 600-V equation could be used to
"Theoretical" 900 162.5 146.3 1.33 conservatively predict incident energy for 125-V and 250-V
DC systems in areas without overcurrent protection or with
"Theoretical" x 3 3.99
overcurrent protective devices not anticipated to respond.
Ammerman open air 1467 60 88.3 0.80 However with responsive circuit protection in place, accurately
Ammerman panel 1467 60 88.3 1.22
Ammerman swgr 1467 60 88.3 1.78
600-V open air 1612 0.89 TABLE IV
DC ARC TEST RANGE AT HIGH CURRENT LAB [26]
Three-Phase AC (for grounded & ungrounded systems)
DC voltage (V) 125, 250, 600
1584 open air grd 1234 2.14
Bolted fault current (kA) 1 - 25
1584 open air ug 1234 2.77
Electrode gap width (in) 0.2 - 6 (6-152 mm)
1584 panel grd 1404 3.82 Arc duration (s) 0.01 - 2
1584 panel ug 1404 4.96 Arc distance (in) 6, 12, 22, 34 (152-864 mm)
1584 switchgear grd 1404 3.64 Electrode/enclosure Vertical series electrodes in
1584 switchgear ug 1404 4.72 configuration open air; vertical & horizontal
a
IE predicted at 18 in (457 mm) from arc, duration = 1s. parallel electrodes in enclosure

38
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

estimating the arc current is a critical step; the incident energy incident energy density predicted for an arc duration ranging
depends on the response time of the protective device, from 0 to 2 s is displayed in Fig. 5 for both the single- and
activated by the magnitude of the fault current. double-string case studies.
Table III includes the arc current and incident energy Tables VI and VII show that the incident energy (0.5-s arc
predicted by (10), the 600-Vdc open air model. It is not duration) predicted by the Ammerman, 600-V open air and
surprising that the model predicts an arc current that is slightly theoretical DC arc models lie in fairly close proximity for the
higher (7.5%) than the arc current initiated by a 325-Vdc system parameters given. For 260-Vdc arcing across a 1-in
supply. It also predicts an incident energy density slightly gap, the 2007 report [28, Fig. 7] suggests an available bolted
higher (11%) than the value predicted by Ammerman for open fault current of 18.7 kA would result in a 9.2-kA arc current, a
air. Equation (10) was derived from testing bolted fault magnitude significantly lower than Ammerman’s model
currents between 2 and 25 kA, a range higher than the bolted predicts but probably more representative of actual arcing in
fault current for the example data in Table III; gap widths for DC power systems. Arc duration and therefore incident
the 600-V testing were also probably wider than 3/8-in arc gap energy are governed by the arc current when responsive
used. Although not published in [26], enclosure multiplying protective devices are present. The arc current and incident
factors (based on 20inx20inx20in test enclosures) were energy density predicted by the 1584 three-phase AC
formulated for (10) [27]. equations do not and are not expected to match the values
A summary of the 130-V and 260-V DC arc testing program predicted by the DC arc models. A 250-V, three-phase Vac
was published in [28]. Although detailed test data and system would be associated with low magnitude, possibly
formulations were not included, [28] contains some valuable non-sustaining, arc currents.
observations about the testing. A 130-Vdc supply was
capable of sustaining arcs across 0.25-in and 0.5-in gap TABLE VI
widths; however, 0.25-in gap arcing could not be sustained SINGLE-STRING BATTERY SYSTEM
using a single-phase, 125-Vac supply and similar bolted fault System Ibf, Vdc, gap 18.7 kA 258 V 20 mm
current levels. Arcing was sustained across 1-in and 2-in Iarc Iarc/Ibf IE (cal/cm2)
gaps using a 260-Vdc supply. To achieve sustainable AC Arc models (kA) (%) 18”, t=0.5 s
arcs across 1-in and 2-in gaps, the single-phase AC supply
Ammerman open air 11.8 63 5.1
voltage had to be increased to 480 or 600 Vac [28]. Unlike
AC arcs, DC arcs do not pass through current zero and do not Theoretical 9.4 50 5.5
require re-ignition each half cycle. A single-phase, AC arc 600-V open air 12.6 67 6.0
current oscillates at the same fundamental time-varying 1584 open air, gnda 6.8 36 6.9
frequency as the AC supply voltage; the arc plasma column 1584 open air, ugb 6.8 36 9.0
(and radiated heat) will also expand and contract. Under a
“gnd” - grounded system; b “ug” - ungrounded system
stable arcing conditions, the magnitudes of DC arc current are
much more constant over time. The DC arc plasma column TABLE VII
also exhibits much less variation [26]. DOUBLE-STRING BATTERY SYSTEM
For average DC and AC rms arc currents over a range of 2 System Ibf, Vdc, gap 39.0 kA 258 V 20 mm
to 10 kA, the heat flux (the rate of incident energy) is Iarc Iarc/Ibf
2
IE (cal/cm )
approximately 1.25 times higher for DC arcing [28]. Rough arc Arc models (kA) (%) d=18 in, t=0.5 s
current and heat flux interpretations of figures presented in
[28] are provided in Table V. Ammerman open air 23.5 60 11.0
Theoretical 19.5 50 11.4
TABLE V 600-V open air 25.4 65 12.1
260-V DC ARC TEST RESULTS IN [28, FIGS. 7 AND 8] a 1584 open air, gnda 11.7 30 12.5
Gap width (in) 1 2 1584 open air, ugb 11.7 30 16.2
a
Bolted fault current (kA) 2 – 25 2 – 31 “gnd” - grounded system; b “ug” - ungrounded system
2
Arc current (kA) 1¼ – 11½ /3 – 9½
2
Heat flux (cal/cm -s) at 12 in ⅟3 – 12½ ⅟3 – 18
a
Ranges estimated by rough visual interpretation of graphs.

V. BATTERY POWER EXAMPLES

In [21] Ammerman presented arc flash case studies of a


DC distribution system fed by a 250-Vdc bus supplied by a
double-string battery system. For the single-string case study,
one battery was removed. A more comprehensive discussion
of the single- and double-string case studies is discussed in
[21] where Ammerman predicted open air incident energy
density for the two cases. Tables VI and VII provide the arc
current and open air incident energy predicted using several
arc models; the system bolted fault current, open circuit
voltage and gap width are also listed in the tables. The Fig. 5. IE comparison of arc models for case studies in [21].

39
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

In [8] Doan presented an example for a 135-V battery with results in Fig. 5 illustrate the potential to severely burn or kill
a short-circuit current of 1338 A. A 2-s arc duration was set, when high fault currents are available and sensitive
the maximum value used in arc-flash studies. The incident responsive devices are not protecting the system.
energy predicted by the Ammerman and the theoretical DC
arc models is listed in Table VIII. Fig. 6 displays the predicted
incident energy as a function of time. For 135-V arcing across VI. PHOTOVOLTAIC INSTALLATIONS
an assumed 0.5-in gap width, the Ammerman and the
theoretical DC models predict arc currents in closer proximity The maximum power transfer theorem states that maximum
than the previous example. The incident energy predictions power is transferred to a linear load when the load impedance
for the open air Ammerman and theoretical DC models are matches the source impedance. As shown in Fig. 1, 25% of
almost identical. The proximity in the results lends confidence the DC short-circuit power is the maximum power transferred
in the performance of each model for 130-Vdc systems and to the DC arc; the DC arc current and arc voltage are equal to
open air arcing. However, the incident energy values 50% of the short circuit current and open circuit voltage. The
predicted by the Ammerman panel and switchgear models are arc is assumed to be a passive linear resistance and any non-
considerably higher than the theoretical DC model, but linearity of the DC power source is not addressed.
considerably lower than three times the DC theoretical model, Photovoltaic cells are highly nonlinear elements. The voltage
which is sometimes applied to enclosure arcing. These two and current at maximum load power may even be estimated
battery examples suggest that the Ammerman and theoretical as 80% of the open circuit voltage and 94% of the short circuit
DC arc models may be adequate for predicting open air current for utility-grade photovoltaic cells [29, Table II].
incident energy on 125 to 250 Vdc bus when protective Therefore, the maximum load power might be estimated as
devices are not expected to interrupt the arcing fault. 75% of the short circuit power – three times higher than the
However, no arc testing has been done in the public domain maximum power transfer theorem for linear DC loads.
to verify either model’s performance for these conditions. However, usage of either the maximum power transfer
Furthermore, arcing occurs in a wide range of DC systems theorem or of the photovoltaic maximum power point to
(and system voltages) and often occurs in some type of predict maximum arc power fails to address the potential
electrical enclosure. current limiting effect of the arc resistance.
The open air incident energy values predicted by the DC Large photovoltaic arrays are becoming increasingly
arc models shown in Figs. 5 and 6 do provide an excellent common on commercial and industrial sites to reduce the
perspective of the wide range of potential incident energy consumption of utility power. The number of solar farms
exposures in the 125-V to 250-V DC range. Low short-circuit connected to the utility grid increases annually to decrease
capacity certainly limits the potential heat exposure, even for a the dependence on fossil fuel in electric power generation.
325-Vdc supply as shown in Table III. But fairly low voltage The differences between photovoltaic cells and batteries may
DC supplies do not necessary limit the risk of injury. The need to be addressed in arc hazard assessment. Two recent
papers provide additional information about photo voltaic
TABLE VIII systems [29], [30] and are excellent references on the subject.
MODEL RESULTS WITH 135-V BATTERY, ISC = 1338A Reference [30] discusses an arc-flash hazard study
Arc Models Iarc Varc IE (cal/cm2), t = 2 s, conducted for a 1.3-MW photovoltaic system.
A V d = 18 in (457 mm) An example solar farm study presented in depth in [29] will
Ammerman open air 751 59 0.81 be briefly covered in this work; the arc hazard is being
Ammerman panel 751 59 1.23 analyzed for a recombiner panel. The photovoltaic system
Ammerman switchgear 751 59 1.80 parameters are summarized in Table IX. The arc currents
Theoretical 669 67.5 0.82 and incident energies predicted by the various DC arc models,
Theoretical (3 x IE) 2.5 as well as the Standard 1584 three-phase AC arc models, are
600-V open air model “blows up” (invalid for Ibf < 2000 A) provided in Table X for comparison. Without conducting arc
tests supplied by photovoltaic cells, it is difficult to even guess
which DC model is the best predictor of arc current. The DC
2
incident energy predictions range from 1.61 to 38.1 cal/cm .
The 600-V open air equation is the only DC model based on
DC arc testing, but it was formulated from open air arc tests
which were supplied by a lower voltage, controlled DC power
source, not eleven photovoltaic cells connected in series [29].
It seems plausible that the potential DC incident energy
exposure should not exceed the potential exposure from a
three-phase, AC arcing fault. Therefore, the conservative
maximum incident energy exposure is assumed to be less
than or equal to 9.28 cal/cm2, the value predicted for the
three-phase AC low voltage, ungrounded panel enclosure
(highest value predicted by any form of the 1584 equation). It
should also be kept in mind that the available, short circuit
current is very low and the 2-s arc duration used in this
example is long. Heat loss occurs as heat is radiated,
Fig. 6. IE comparison of DC arc models for system in [8]. convected and possibly conducted (molten material) to a

40
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

distance 18 inches (for this example) away from the arc. Also, extended to DC arcing, and 600-V open air equations
as previously addressed in Section II (paragraph 6), not all of developed for rail and transit systems. Only the commercial
the energy generated by the arc is converted to incident software package and 600-Vdc open air equations have been
energy, although the DC theoretical, Ammerman and PV verified by or are based upon representative DC arc testing.
maximum power point models are based on this assumption. While it is agreed that industry needs to address the arc
Long duration, low current arcing may not pose a potential hazards posed by DC systems, accurate methods of
2
incident energy risk even close to 10 cal/cm at an 18-in quantifying the arc hazard are critical. Without accurate
distance from the arc. assessment, any arc flash hazard study performed on a DC
Reference [29] predicts a maximum incident energy of 52 power system has limited value. Inaccurate assessment is
2
cal/cm from a source with an available short circuit of less counterproductive to the effective protection of workers.
than 1000 A. The incident energy prediction in [29] is based Furthermore, inflating potential risks not only adds to the cost
on the maximum power point of a photovoltaic cell, a safety of protection, but also taxes workers with additional personal
multiplier of 3, and a temperature compensation factor for the protective equipment (PPE), which may be more cumbersome
solar cells (not addressed here). Certainly, implementing or hot than necessary. PPE is a necessary means of
such PPE requirements would be counterproductive to protecting workers from arc hazards in energized systems, but
industry, counterproductive to worker safety, and wearing it in some circumstances may also increase the risk
counterproductive to the acceptance and adoption of safe of heat stroke or some form of occupational injury. Industry
work practices. The publication of the results in [29] may needs an acceptable method of DC arc hazard assessment
have been partially motivated to raise awareness in the solar developed in a nationally recognized standard. Support is
photovoltaic industry on the urgent need for arc testing so that needed for research and testing of the DC arc hazard.
the potential arc hazard risk to workers can be assessed
accurately.
VIII. AUTHORS’ INFORMATION
TABLE IX
ELECTRICAL PARAMETERS FOR EXAMPLE SYSTEM [29] Tammy Gammon earned a bachelor, a master of science
Circuit Current Voltage and a Ph.D. in electrical engineering from the Georgia
A V Institute of Technology in the 1990s. She is licensed as a
Open circuit -- 981.2 Professional Engineer in North Carolina. Since 2003, Tammy
Short circuit 940.2 -- has worked as senior electrical engineer for John Matthews
PV maximum power point 879.8 794.2 and Associates. She performs research and analysis in
power and power quality issues, in fires of electrical origin, in
electrical arc and shock injuries, and in product design and
TABLE X manufacturing. Since 2006, Tammy Gammon has also
ARC MODEL RESULTS - EXAMPLE SYSTEM IN TABLE IX served as the research manager for the IEEE/NFPA Arc Flash
DC & 1584 Arc Models Iarc Varc IE cal/cm
2 Research Project. Email: tgammon@tds.net.
(assumed a gap of 4 in) A V d=18in,
t=2s Wei-Jen Lee (S’85-M’85-SM’97-F’07) received the B.S. and
Ammerman open air 782 165 2.35 M.S. degrees from National Taiwan University, Taipei,
Taiwan, R.O.C., and the Ph.D. degree from the University of
Ammerman LV panel 782 165 3.58
Texas, Arlington, in 1978, 1980, and 1985, respectively, all in
Ammerman LV swgr 782 165 5.22
Electrical Engineering. He is currently a professor of the
Ammerman MV swgr 782 165 2.31
Electrical Engineering Department and the director of the
Theoretical 470 491 4.20 Energy Systems Research Center.
Theoretical (3 x IE) 12.6 Prof. Lee has been involved in research on arc flash and
PV max power point 880 794 12.7 electrical safety, utility deregulation, renewable energy, smart
PV max power (3 x IE) 38.1 grid, microgrid, load forecasting, power quality, distribution
600-V open air 508 1.61 automation and demand side management, power systems
1584 LV open, uga 936 5.19 analysis, online real time equipment diagnostic and prognostic
1584 LV panel, uga 1,064 9.28 system, and microcomputer based instrument for power
1584 LV swgr, uga 1,064 8.84 systems monitoring, measurement, control, and protection.
1584 MV open, uga 950 3.52 Since 2008, he has also served as the project manager for the
1584 MV swgr, uga 950 4.51 IEEE/NFPA Arc Flash Research Project.
a
1584 - Higher IE are predicted for ungrounded systems. LV IE Prof. Lee is a Fellow of IEEE and registered Professional
are higher due to the 1.5 crest factor for LV systems. Engineer in the State of Texas. Email: wlee@uta.edu.

Zhenyuan Zhang received the B.S. degree from Chang’an


VII. CONCLUSION University, Xi’an, China in 2007. He is now pursuing a Ph.D.
in electrical engineering at the Energy Systems Research
Four available DC arc models for estimating incident Center, Department of Electrical Engineering, the University of
energy density have been discussed and explored in this Texas at Arlington, Arlington, Texas, USA. His focus lies in
paper, namely: the theoretical model attributed to Lee, arc flash research, but he has also been involved in hybrid
Ammerman’s model, a commercial software package [31] energy storage, smart grids, renewable energy, electrical

41
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

safety analysis and power systems analysis. Since 2010, T 1   T 


  cp     E 2  enet  r   
Zhenyuan has served as the project associate for the
t r r  r 
IEEE/NFPA Arc Flash Research Project. Email:

zhenyuan.zhang@mavs.uta.edu.
I (t )  E (t )     2 r  dr
0
Ben C. Johnson is presently Senior Consultant for Thermon
Manufacturing Company. His career spans a broad range of Varc (t )  E (t )  lengtharc  Vanode  Vcathode (11)
industrial experience, including 44 years with Thermon and eight
years in the petrochemical industry with the Ethyl Corporation
and the Diamond Shamrock Corporation. Mr. Johnson was TABLE XI
Thermon’s Vice President of North American Sales for five years VARIABLES FOR ARC ENERGY BALANCE [31]
and Thermon's Vice President of Engineering for twelve years, Variable Definition
responsible for product application design, field and construction  Gas density, kg/m
3
services. He was previously Thermon's Vice President of T Time, s
Research and Development. He is the holder of eight patents in  Gas electrical conductivity, S/m
the field of surface heating, and is responsible for numerous new
cp Gas specific heat at constant pressure, J/(K-kg)
product innovations. He has authored or co-authored 19 papers r Radial distance from the arc, m
for various societies. As United States delegate to the λ Gas thermal conductivity, W/m-K
International Electro-technical Commission (IEC), he is the T Temperature, K
Convener for TC31 Maintenance Team 79-30, Electrical enet Net radiation heat transfer, W/m3
Equipment in Flammable Atmospheres, Electrical Resistance E Electrical field in the arc column, V/m
Trace Heating and US Technical Advisor for IEC TC27, Safety in
Electroheat Installations. He is also a member of the US
Technical Advisory Committee for IEC TC31. Mr. Johnson is a
A finite difference method is used to solve for temperature
Life Fellow of the IEEE and is Co-Chair of the IEEE/NFPA
and electric field. The arc is assumed to be isothermal, and
Collaboration on Arc Flash Research. Email:
the temperature field is one-dimensional inside the arc. The
Ben.Johnson@thermon.com.
electrode vapor and the melting and shortening of electrodes
are not considered. The arc length is assumed to be at least
twenty greater than the arc diameter. The original, single-
phase AC model was verified for the following parameter
APPENDIX A ranges: ac arc currents from 3.5 to 21.5 kA, gap widths from 1
TECHNICAL BASIS OF ARCPROTM SOFTWARE [31] to 12 inches, arc durations from 4 to 30 cycles, and arc
distances from 8 to 24 inches. Based on arc testing
The model was developed to be most applicable to single- conducted at the company’s high current test laboratory, the
phase AC arcs longer than 2.5 cm [31]. The arc is defined as software has been modified to include multiplying factors to
the conducting core where the gas temperature is higher than extend its applicability to DC and three-phase AC arcing.
6000K. The set of energy balance equations is provided in
(11) and the equation variables are identified in Table XI.

REFERENCES
[8] D. Doan, “Arc flash calculations for exposures to dc
[1] R. Lee, “The other electrical hazard: electrical arc blast systems,” IEEE Trans. Ind. Appl., vol. 46, no. 6, pp. 2299-
burns,” IEEE Trans. Ind. Appl., vol. IA-18, no. 3, pp. 246- 2302, Nov./Dec. 2010.
251, May/June 1982. [9] R. L. Doughty, T. E. Neal, T. A. Dear and A. H. Bingham,
[2] D. Pitts and L. Sissom, Schaum’s Outline of Heat Transfer, “Testing update on protective clothing & equipment for
2nd ed. New York: McGraw-Hill Companies, Inc., 1998. electric arc exposure,” IEEE Ind. Appl. Mag., vol. 5, no. 1,
[3] Y. A. Çengel and M. A. Boles, Thermodynamics: An pp. 37-49, Jan./Feb. 1999.
Engineering Approach, 3rd ed. New York: McGraw-Hill [10] H. Wu, email communication, Feb. 18, 2008.
Companies, Inc., 1998. [11] H. Schau and D. Stade, “Impacts of internal arcing faults in
[4] Engineering Guide for Predicting 1st and 2nd Degree Skin low-voltage switchgear assemblies and power systems,” in
Burns from Thermal Radiation, Society of Fire Protection Proc. 28th Universities Power Engineering Conf., Stafford,
Engineers, 2000. United Kingdom, 1995.
[5] C. P. Artz, J. A. Moncrief and B. A. Pruitt, Burns. [12] A. Dasbach, “Untersuchungen zur leistungsbilanz von
Philadelphia, PA: W. B. Saunders Company, 1979. storlichtbogen im hinblick auf die druckbeanspruchung von
[6] T. Gammon, W.-J. Lee, Z. Zhang and B. Johnson, “‘Arc schaltagen,” Ph.D. dissertation, RWTH Aachen Univ.,
flash’ hazards, incident energy, ppe ratings and thermal Aachen, Germany, 1987.
burn injury – a deeper look,” unpublished work. [13] H. Wu, unpublished discussion of Dasbach’s work, Feb.
[7] Standard for Electrical Safety in the Workplace, 2012 ed., 28, 2008.
NFPA Standard 70E, 2011. [14] H. Wu, unpublished key points of Dasbach’s work (in
English), Nov. 18, 2007.

42
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

[15] F. P. Incropera and D. P. DeWitt, Fundamentals of Heat


and Mass Transfer, 5th ed. New York: John Wiley & Sons,
2002.
[16] P. M. Hall, K. Myers and W. S. Vilcheck, “Arcing faults on
direct current trolley systems,” in Proc. Fifth WVU Conf.
Coal Mine Electrotechnology, Morgantown, WV, 1978.
[17] M. Fontaine, unpublished work (with plans for publication).
[18] T. Gammon and J. Matthews, “IEEE 1584-2002: incident
energy factors and simple 480-V incident energy
equations,” IEEE Ind. Appl. Mag., vol. 11, no. 1, pp. 23-31,
Jan./Feb. 2005.
[19] T. Gammon and J. Matthews, "IEEE 1584-2002 arc
modeling debate," IEEE Ind. Appl. Mag., vol. 14, no. 4,
pp.61-69, July/Aug. 2008.
[20] IEEE Guide for Performing Arc-Flash Hazard Calculations,
IEEE Standard 1584, 2002.
[21] R. F. Ammerman, T. Gammon, P. K. Sen, and J. P.
Nelson, “DC arc models and incident energy calculations,”
IEEE Trans. Ind. Appl., vol. 46, no. 5, pp. 1810-1819,
Sept./Oct. 2010.
[22] A. D. Stokes and W. T. Oppenlander, “Electric arcs in
open air,” J. Phys. D: Appl. Phys., vol. 24, no. 1, pp. 26-
35, Jan. 14, 1991.
[23] L. E. Fisher, “Resistance of low-voltage ac arcs,” IEEE
Trans. Ind. Gen. Appl., vol. IGA-6, pp. 607-616,
Nov./Dec.1970.
[24] R. Wilkins, “Proposal for improved equations,” posted by
Mike Lang to the IEEE Electrical Safety Forum,
https://www.ieeecommunities.org/ieee.esafety, on Aug.
30, 2004. [No longer online.]
[25] M. Fontaine and P. Walsh, “DC arc flash calculations –
arc-in-open-air and arc-in-a-box using a simplified
approach (multiplication factor method)”, in Conf. Rec.
2012 IEEE IAS ESW, Jan. 31-Feb. 03, 2012, Daytona
Beach, FL, pp. 97-104.
[26] K. S. Y. Cheng, S. L. Cress and D. J. Minini, “Arc hazard
assessment for dc applications in the transit industry,” in
Conf. Rec. 2011 American Public Transportation
Association (APTA) Rail Conf., Boston, MA, June 12-15,
2011.
[27] K. Cheng, email communication, December 04, 2013.
[28] C. Keyes and C. Maurice, “DC arc hazard assessment
phase II,” Kinectrics, Ontario, Canada, Rep. K-012623-
RA-0001-R00, July 7, 2007.
[29] E. H. Enrique, P. N. Haub and T. P. Bailey, “DC arc flash
calculations for solar farms,” in Conf. Rec. 2013 1st IEEE
Conference on Technologies for Sustainability, Aug. 1-4
2013, Portland, Oregon, pp.97-102.
[30] D. R. Doan and R. M. Derer, “Arc flash calculations for a
1.3MW photovoltaic system,” in Conf. Rec. 2014 IEEE IAS
ESW, Feb. 4-7, 2014, San Diego, California.
[31] User’s Guide for ARCPRO, Version 2.0, Kinectrics,
Toronto, Canada, 2000.

43

You might also like