See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/337486267

Arcing Persistency Measurement for IEC62606 Arc- Fault Detection Device

Certification

Conference Paper · November 2019

DOI: 10.1109/SPCE47297.2019.8950836

CITATIONS READS
0 295

5 authors, including:

Costin Vasile
Schneider Electric
14 PUBLICATIONS   36 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Electrical Transient Detection (Partial Discharge and DC Electrical Arc Fault) View project

Ultrasound Inspection of Heterogeneous Metallic Structures View project

All content following this page was uploaded by Costin Vasile on 25 November 2019.

The user has requested enhancement of the downloaded file.

 Arcing Persistency Measurement for IEC62606 Arc-
Fault Detection Device Certification
Vasile Costin Vivian D’Orchymont Jérôme Meunier-Carus Jean-François Rey
Advanced Protection & Earth Global Labs Advanced Protection & Earth Standardization Officer
Leakage Schneider Electric Leakage Schneider Electric
Schneider Electric Grenoble, France Schneider Electric Grenoble, France
Grenoble, France [email protected] Grenoble, France [email protected]
[email protected] [email protected]

Hynek Raisigel
Advanced Protection & Earth Leakage
Schneider Electric
Grenoble, France
[email protected]

Abstract—Arc Fault Detection Devices (AFDD’s) or Arc Fault in mind that there is a strong connection between the two
Circuit Interrupters (AFCI’s) are mandatory electrical protection standards.
devices in panel boards all over the world, to mitigate the risk of
fire due to hazardous electrical arc faults. By its nature, the arc The two types of arc faults which are considered dangerous
fault is an erratic phenomenon, with variable temporal persistency are the series and parallel arc faults. As per the IEC62606
and variable ability to ignite a fire. The two main standards for standard, a parallel arc fault is an arc fault where the arc current
AFDD certification are UL1699 & IEC62606, which require is flowing between active conductors in parallel with the load of
similar arcing tests. However, there is no current requirement for the circuit”; similarly, the series arc fault is “an arc fault where
a minimal arcing persistency during a standard series arc test, the current is flowing through the load(s) of the final circuit
neither a clear guideline on how to measure the arcing temporal protected by an AFDD” [1]. The two types of arcs are depicted
persistency and confirm the test yielded a successful series arc in Fig. 1.
fault. This work aims to enhance the standards, by proposing a
simple method to precisely determine the arcing persistency
during a series arc fault test. Using only the arc current and arc
voltage as inputs, a standard series arc signal can be accepted or
rejected, based on the measured level of temporal arcing
persistency.

Keywords—AFDD, AFCI, series arc fault, IEC62606, electrical

arc, residential electrical protection, arc voltage, arc current.

I. INTRODUCTION
The world of residential electrical protection is presently a
mature market, in terms of the involved actors and most notably
in terms of the technology used. Products such as Miniature
Fig. 1. Series & parallel arc faults: occurrence in the electrical circuit.
Circuit Breakers (MCB’s, which typically protect the electrical
wiring) or Residual Current Devices (RCD’s, used to protect The IEC62606 standard gives a clear definition of what
people from the risk of electrocution) are ubiquitous today in amount of arcing is necessary during a parallel arc fault test, as
electrical panel boards across the world. A rather new electrical shown in Table 1 [1].
protection device is called the Arc Fault Circuit Interrupter
(AFCI) or Arc Fault Detection Device (AFDD), and it is TABLE I. MAXIMUM ALLOWED NUMBER OF ARCING HALF-CYCLES
intended to mitigate the risk of an electrical fire caused by a WITHIN 0.5S
hazardous (from a fire risk perspective) arc fault [1]. This type
Test
of protection has been introduced in the US in 2002 as arc 75 A 100 A 150A 200A 300A 500A
mandatory for residential panel boards, and it is recommended current
for the IEC market, since 2014 [2]. (rms)
N 12 10 8 8 8 8
The two main standards today used for product certification N = number of half-cycles at the rated frequency
are the UL1699 (North America) and the IEC62606 (rest of the
world). For this work we will focus on IEC62606, while keeping
 While not as clearly defined in terms of mandatory arcing test conditions and non-repeatable results (in terms of arcing
duration, there is, however, a trip time requirement for series arc persistency/duration), when performing tests across different
faults, in function of the test arc current [1]: various laboratories.

TABLE II. LIMIT VALUES OF BREAK TIME FOR UN=230V AND 400V II. PROBLEM OUTLINE & EXPERIMENTAL SETUP
AFDD’S
A. Problem Outline
Test arc
current 2.5 A 5A 10 A 16 A 32 A 63 A For a better understanding of the present-day limitations of
(rms) the IEC62606 standard (and its UL counterpart), let us inspect
Maximim the recorded arc current and arc voltage during a standard arcing
break 1s 0.5 s 0.25 s 0.15 s 0.12 s 0.12 s test (blue – arcing signals, red – trip time window):
time

For more context, the most common method for reproducing

electrical arc fault in accordance to the IEC62606 (and UL1699)
is the carbonized wire sample arcing method, as shown in Fig. 2
[1][3].

Fig. 3. Arc current & arc voltage, with arcing starting at 0s. Red line –
conformance window. Stable series arc fault.

Arc fault tests involve introducing a series arc fault (the

carbonized cable sample from Fig. 2) into an electrical circuit
branch, interacting with different kind of load types (more in
section II-B). Fig. 3 shows what is considered to be a
“persistent” series arc fault, with conduction at each network
half-cycle (network frequency = 50Hz), non-zero arcing current
(no open circuit in the wire sample), non-zero arc voltage (no
short circuit in the wire sample) and showing the arc-specific
Fig. 2. Carbonized wire sample arcing method. shape at each half-cycle, that contains the ignition peak,
conduction (arc voltage plateau) & extinction peak phases. Also,
The insulators of the parallel conductors are cut transversally the values for the arc plateau voltage are around 50V, which is
and the incision is wrapped in PVC electrical tape. Next, three consistent with known values [1][5][7][8][11]. Before adding
sequences of voltages are applied to break the insulation the series arc into the circuit, the carbonized wire sample current
between the conductive wires and generate the carbonized path is 0A (before t = 0s).
(hence the name “carbonized cable specimen”). Once carbon is
deposited in the incision, the conditions for arcing are obtained
and the connections shown in the lower part of Fig. 2 allow
placing the arc fault in series with the load(s) [1][3]. At this step,
the cable is considered ready for testing, in that it was
conditioned for series arc fault reproduction, similar to the real
word unintentional arc [12], allowing for the AFDD tripping
evaluation.
The IEC62606 allows for the reproducibility of generating
series arc faults for certification testing, but it does not and
cannot control the arcing phenomenon itself, which can generate
arc faults with different durations (correlated to different fire
probabilities) [4]-[8][11] in the exact same test conditions. Fig. 4. Arc current & arc voltage, with arcing starting at 0s. Red line –
conformance window. Short-lived series arc fault.
In this paper, we will address the lack of a defined minimum
series arcing persistency (e.g. stability, duration) during the In Fig. 4, however, there is a short-lived series arc fault.
standard test. A method for measuring arcing persistency inside After ~0.15s the arc fault becomes an open-circuit, indicated by
the allowed trip time will be presented, in order to help IEC the 0A arcing current & the arc voltage which becomes the
certification laboratories identify which series arc fault tests are network voltage at 230Vrms. The series arc fault has a low
successful or not. Not having a definite requirement on a intensity and duration, when compared to the allowed trip time
minimum amount of arcing and how to measure this information (red line). As such, the total dissipated energy and the associated
represents a grey zone in the IEC standard and it leads to unclear fire risk are reduced [6][11].
 The IEC62606 standard requires nonetheless that the AFDD features and the result for the test arc signal is the arcing
trips on this type of signal, although the risk factors for series arc persistency inside the allowed 1000ms: Persistency = 40 arcing
faults are minimal for very short durations, such as in the above half-cycles / 100 allowed half-cycles = 40%.
figure. By proposing an arcing persistency measurement, this
works aims at helping IEC62606 compliant laboratories to To find the proper threshold values for the numerical
perform successful & repeatable series arc fault tests needed for features, we use two approaches:
IEC certification of AFDD-type products. - Thresholds are chosen by limiting certain features due
to physics-based knowledge of arcing: arcing cannot
B. Experimental Setup & Methodology
occur under certain electrical conditions.
For this work, the captured signals consisted of the arc
current (the current passing through the conditioned wire - Other thresholds are chosen empirically, by visually
sample) and the arc voltage (the voltage across the wire sample). inspecting and statistically analyzing our signal
These were acquired with a Lecroy HDO 6034 scope, offering database. The values that best classify arcing vs non-
12bit vertical resolution. The sampling frequency was set at arcing half-cycles are selected.
1Ms/s, allowing the possibility of later high frequency spectral
analysis if needed. A diagram with the experimental setup can III. ARCING PERSISTENCY MEASUREMENT
be seen in Fig. 5. A. Objectives
As stated before, the main objective is to propose a method
for measuring the arcing persistency during a IEC62606
standard series arc test. This would help with invalidating some
tests which exhibit very short arcing (with lower fire ignition
probability [6][11]) and make it non-mandatory for the AFDD
to trip on these types of signals.
A secondary yet still important objective is to attempt to
keep this method as simple and easy to implement as possible,
to increase the chance that an IEC certification laboratory would
adopt this methodology. This translates to using the smallest
possible number of arcing features, a simple algorithm that
combines these features (and decides “arcing” or “non-arcing”
Fig. 5. Experimental configuration with data acquisition system & sensors. for each half-cycle) and typical hardware for any electrical
measurement laboratory.
The following sensors were used to measure the signals:
The arcing criteria & persistency algorithm do not need to be
- Lecroy CP030 current clamp, with a DC-50MHz implemented in real-time processing, as the signals can be
bandwidth; processed offline, immediately after they are recorded and the
- Lecroy HVD3106 differential voltage probe, with a series arc test has been performed.
120MHz bandwidth. B. Arcing Criteria
The IEC62606 standard, section §9.9.4.2, requires validating Firstly, some minimum arcing requirements are needed, in
an AFDD with multiple electrical appliances, in what is called a terms of minimum arcing current & voltage values. In [4][5] a
“masking test with inhibition loads”. The mandatory loads are: minimal half-cycle arcing voltage of 10V is proposed, which is
a vacuum cleaner, an electronic switching mode power supply in line with Schneider Electric previous experience. Also, in [9]
(SMPS), an air compressor, electronic lamp dimmers, two 40W & [10] Shea indicates that 10V and lower voltages are more
fluorescent lamps + one 5A resistive load, 12V halogen lamps + typical for glowing contacts. The first criteria is, for each
one 5A resistive load and an electric hand tool, such as a power network half-cycle:
drill (with at least 600W power) [1]. Each appliance has been 𝑝𝑒𝑎𝑘
𝑉𝐴𝑅𝐶 > 10𝑉
tested in the four different electrical configurations required by
the standard (IEC62606, Figure 11 [1]). Additional section The IEC62606 standard requires a minimum arcing current
§9.9.2.2 testing has been performed at several current levels, of 2.5A to solicit a trip command from the AFDD. Following
with resistive loads in series with the arc fault. In total, 220 our experience, a minimum value of 1A for the peak arc current
arcing files were collected for this work, offering sufficient is chosen (consistent with [4] and [8]). Consequently, our second
insight into the variability of the series arc fault phenomena. arcing criteria is, for each network half-cycle:
𝑝𝑒𝑎𝑘
The methodology of this work consists of identifying 𝐼𝐴𝑅𝐶 > 1𝐴 (2)
numerical features that indicate, by thresholding, the presence of
arcing in each half-cycle. Then, each arcing half-cycle is Again, these first two conditions act as minimum arcing
counted inside the standard trip window and a persistency is requirements, yet not sufficient to indicate series arcing at a half-
computed for each signal. Example: in an allowed trip time of cycle level. For this purpose, a third criterion has been
1s = 1000ms we have 100 total network half-cycles (50Hz introduced: the rms (root mean square) value of the arcing
network frequency yields 10ms/1half-cycle); a total number of power, at a half-cycle level:
40 arcing half-cycles is counted by using various numerical
 𝐴𝑅𝐶 𝐴𝑅𝐶
𝑃𝐴𝑅𝐶 = 𝑟𝑚𝑠(𝑈1:𝑁 × 𝐼1:𝑁 )

Where N is the sample length for one half-cycle of the arc

voltage & arc current, respectively and rms represents the root
mean square value. The criteria itself is to expect, in case of
arcing inside the half-cycle, the following condition:
Fig. 7. IEC62606 masking load configuration C.
𝑃𝐴𝑅𝐶 > 45𝑊 (4)

The 45W threshold value has been chosen based on visually

inspecting all arcing current & voltage waveforms and finding
the value that successfully classified arcing & non-arcing half-
cycles. Analytically, this value can be explained by two factors:
- Arcing can still occur at 1A, albeit less intense & less
likely to ignite a fire [4][5][8].
- In [5], for arc current values of 0.5-1.5A a mean arc
voltage of 56.2V has been observed. Thus, an expected
minimum PARC of 45W would be consistent with these
results.
C. Arcing Persistency Measurement Algorithm
The previously presented arcing criteria are applied for each
half-cycle of the arc current & voltage signals and a final
persistency value is given as its output.
This very easy to implement algorithm is depicted in Fig. 8. Series arc test with a vacuum cleaner.
diagram form in Fig. 6.
The series arc is very short lived: after some stable half-
cycles after the ignition, the arc starts to sputter and extinguish
itself. The arcing presence algorithm captures this behavior and
the persistency is measured at 0.2 (20%) inside the allowed trip
window.
The next example is for a series arc masking test with a PC
power supply, in configuration A. The IEC62606 standard
configuration A for a masking load is shown below [1].

Fig. 6. Block diagram of the arcing persistency measurement algorithm.

The arcing criteria are combined with a simple logical-AND,

to decide whether arcing is present or not in the given half-cycle.
This information is then accumulated and, by using the length of Fig. 9. IEC62606 masking load configuration A.
the conformance window (available trip time, given in network
half-cycles), the value of the arcing persistency is computed,
ranging from 0 to 1.
Thus, for every entry in the 220-signal database, we will
have a corresponding arcing persistency value.
D. Results
To prove the efficiency of this method in identifying arcing
presence (at a half-cycle level) and measuring its persistency (at
a file/arc test level), its performances will be shown on signals
with low arcing persistency.
First, a series arc test performed with a vacuum cleaner
masking load, in configuration C, as per IEC62606. A standard
configuration C for any masking load is shown below [1].

Fig. 10. Series arc test with a PC power supply.

 This type of behavior is typical for an active load, like the TABLE III. LOW PERSISTENCY SERIES ARC FAULT TESTS.
PC power supply, that tries to ensure constant output power, due Load Arc
to the series arc fault, which is seen as a network disturbance. Type Persistency
The series arc fault acts as time-varying voltage source, which Vacuum 20%
makes it hard for the power supply to regulate its output. Cleaner
Therefore, for some half-cycles (generally of the same polarity), SMPS 53.2%
SMPS 59.6%
the arc current is close to 0A and the arcing persistency
SMPS 53.2%
algorithm discards these half-cycles as non-arcing half-cycles. SMPS 66.1%
There needs to be a different study about the real danger of SMPS 69.3%
SMPS 60.9%
low-arcing persistency tests, in conjunction with PC power
Light 35.3%
supplies, such as in [13]. It could be also envisaged to have a Dimmer
different, lower minimum arcing persistency requirement for arc
fault detection on switch mode power supplies (SMPS’s). For An overall look on arcing persistency across the 220-test
this test, the final arcing persistency was measured at 53% (0.53 database shows the majority of the arc tests yielded a very high
in the figure). (>80%) arcing persistency, as per Fig. 12.
Another low arcing persistency signal is shown in Fig. 11, a
series arc masking test with a light dimmer, in configuration C.

Fig. 12. Histogram of arcing persistency across all 220 signals.

Out of the total 220 arcing tests, 210 have shown arcing
persistency higher than 80%. This shows the IEC62606 standard
methodology manages to produce mostly high persistency &
high energy series arc faults. The other low-persistency arc tests
still require their detection by an AFDD, for a valid IEC62606
certification, and this is regardless of their lower risk to the
electrical installations (less arcing typically involves lower risk
[6][11]).
The method for arcing persistency measurement has also
Fig. 11. Series arc test with a light dimmer. been tested with input signals (arc current & arc voltage) sample
at 10kS/s, without any loss in precision, nor any change in the
In this case there is another typical example of a short-lived, results.
sputtering arc-fault, but due to a different process.
IV. CONCLUSION
The carbonized wire sample becomes resistive, as can be
seen by the non-zero sinusoidal arc voltage, most notably in the The IEC62606 standard requires an AFDD to detect series
second half of the signal. The first two arcing conditions are true arc faults, in combination with several different loads, to obtain
(minimal peak IARC = 1A and peak VARC = 10V), but the the IEC62606 certification (the same is true for UL1699). The
instantaneous power requirement of 45W manages to filter this standard, however, does not define, nor demand a minimum
phenomenon out. The arcing persistency is at 35% (0.35), required arcing persistency (as opposed to the parallel arc fault).
considerably shorter than the allowed standard trip time. For the exact same test conditions, two (or more) successive
To summarize, out of the 220 series arc fault tests, only 8 arc faults can have totally different arcing durations, and,
arcs showed less than 70% arcing persistency, as shown in consequently, different dissipated energy levels, with different
TABLE III. associated risk factors. It is important to maintain the focus of
standard testing on temporally persistent arc faults, which
correlate to higher probabilities of igniting a fire in the electrical
circuit.
To enhance the standards in this direction, this work presents
a simple method of measuring arcing persistency during a series
arc fault test. It uses criteria related to the minimum peak arc
current, minimum peak arc voltage and a minimal instantaneous
arc power to decide if arcing is present or not inside each half- [10] J.J. Shea, “Material Effect on Glowing Contact Properties,” IEEE
cycle. This information is accumulated and, combined with the Transactions on Components and Packaging Technologies, Vol. 32, No.
4, November, 2009.
standard trip time for each test, the series arc persistency is
[11] D. Kolker, S.Campolo, N. DiSalvo, “A study of Time/Current
computed. Characteristics of the Ignition Processes in Cellulosic Material Caused by
Tested on a 220-series arc test database, it has shown to be Electrical Arcing for Application in 240V Arc-Fault Circuit Interrupters,”
Proceedings of the 53rd IEEE Holm Conference on Electrical Contacts,
precise and the additional advantage of being simple to pp. 105-114, Sept..2007.
implement. For this work, the processing has been performed [12] G. Parise, L. Parise, P. Nicolucci, “Localized fire ignition hazard in
offline, after recording the signals. Nevertheless, a real-time branch circuits, cords and connected equipment,” IEEE Industry
version can be imagined, with a 10kS/s sampling rate of the Applications Society Annual Meeting, 2011.
input signals, which is easily achievable by a wide range of lab [13] K. Takenaka, Y. Mizuno and A. Yoshida, “Condition Monitoring of
equipment. Damaged AC Power Supply Cord Using Voltage Waveform”, IEEE
Condition Monitoring and Diagnosis (CMD), 2018.
As a next step, a consensus needs to be reached as to which
minimum arcing persistency should be required. It also needs to
study in more detail the fire risk level of lower arcing persistency
signals with PC power supplies (SMPS) as masking loads. One
possible consequence of this study might be to accept less
persistent arcing for this type of load, if the fire risk proves to be
high.
The scientific literature already has some examples of
requirements for minimal arcing persistency [3], but the
measurements have been performed differently. Nevertheless, it
represents a starting point for this discussion and this work aims
to contribute to this industry-wide topic.
As a future development, the results of this work will be
validated on a secondary series arc signal database, which will
be collected in a secondary laboratory. The aim is to verify the
assertions made in this study, by performing a cross-laboratory
analysis.
ACKNOWLEDGEMENT
Thank you for the support of our colleagues from Schneider
Electric US: John J. Shea, Brett Larson, Robert Isaacson.
REFERENCES

[1] International Standard IEC62606, IEC, February 2017.

[2] IEC 60364-4-42:2010+Am.1:2014 Low-voltage electrical installations -
Part 4-42: Protection forsafety - Protection against thermal effects.
[3] UL Standard for Safety for Arc-Fault Circuit-Interrupters (UL 1699), Ed.
2017-05-03.
[4] J-M. Martel, “Series arc faults in low-voltage AC electrical installations,”
PhD Thesis, Technische Universität Ilmenau/Universitätsbibliothek,
2018.
[5] J-M. Martel, F. Berger, “A study of arcing fault in the low-voltage
electrical installation”, Proceedings of the 56th IEEE Holm Conference
on Electrical Contacts, 2010.
[6] W-S. Moon, J-C. Kim and A. Jo, “Ignition Characteristics of Residential
Series Arc Faults in 220V HIV Wires,” IEEE Transactions on Industry
Applications, vol. 51, No. 3, May/June 2015, pp. 2054–2059.
[7] J.J. Shea, “Comparing 240 Vrms to 120 Vrms Series Arcing Faults in
Residential Wire”, Proceedings of the 54th IEEE Holm Conference on
Electrical Contacts, 2008.
[8] J.J. Shea, “Conditions for Series Arcing Phenomena in PVC Wiring,”
IEEE Transactions on Components and Packaging Technologies, Vol. 30,
No. 3, August 2007, pp. 532-539.
[9] J.J. Shea, “Glowing Contact Physics,” Proceedings of the 52nd IEEE
Holm Conference on Electrical Contacts, 2006.

View publication stats