602 Tools For The Simulation of The Effects of The Internal Arc in Transmission and Distribution Switchgear

602
Tools for the Simulation of the Effects of the

Internal Arc in Transmission and Distribution
Switchgear
Working Group
A3.24
December 2014
TOOLS FOR THE SIMULATION OF THE
EFFECTS OF THE INTERNAL ARC IN
TRANSMISSION AND DISTRIBUTION
SWITCHGEAR
WG A3.24
Members
N. Uzelac, Convenor (US) M. Glinkowski, Secretary (US), L. del Rio (ES), M. Kriegel,
Former Convenor (CH), J. Douchin (FR), E. Dullni (DE), S. Feitoza Costa (BR), E. Fjeld (NO),
H-K. Kim (KR), J. Lopez-Roldan (AU), R. Pater (CA), G. Pietsch (DE), T. Reiher (DE), G.
Schoonenberg (NL), S. Singh (DE), R. Smeets (NL), T. Uchii (JP), L. Van der Sluis (NL), P.
Vinson (FR), D. Yoshida (JP)
Copyright © 2014
“Ownership of a CIGRE publication, whether in paper form or on electronic support
only infers right of use for personal purposes. Are prohibited, except if explicitly
agreed by CIGRE, total or partial reproduction of the publication for use other than
personal and transfer to a third party; hence circulation on any intranet or other
company network is forbidden”.
Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor
does it accept any responsibility, as to the accuracy or exhaustiveness of the
information. All implied warranties and conditions are excluded to the maximum
extent permitted by law”.
ISBN : 978-2-85873-303-3
Tools for the simulation of the effects of the internal arc in transmission and distribution switchgear
TOOLS FOR THE SIMULATION OF THE

EFFECTS OF INTERNAL ARC IN TRANSMISSION
AND DISTRIBUTION SWITCHGEAR
T ABLE OF CONT ENTS
TABLE OF CON TEN TS ....................................................................................................... 1
EXECUTIVE SUMMARY ............................................................................................................ 3
1 INTRODUCTION ................................................................................................................. 6
1.1 Overview ........................................................................................................................ 6
1.2 Abbreviations ................................................................................................................. 9
1.3 Definitions .................................................................................................................... 10
1.4 Referred standards....................................................................................................... 13
2 CALCULATION OF PRESSURE USING A BASIC MODEL.............................................. 14

2.1 Introduction .................................................................................................................. 14
2.2 Equations of the basic model ........................................................................................ 14
2.3 Input parameters .......................................................................................................... 16
2.4 Application limits of the basic model ............................................................................. 20
2.5 Application of the basic model to selected test cases .................................................... 21
2.6 Enhancements of the basic model ................................................................................ 33
2.7 Summary ..................................................................................................................... 34
3 SENSITIVITY ANALYSIS OF THE BASIC MODEL........................................................... 36

3.1 Introduction .................................................................................................................. 36
3.2 Model parameters ........................................................................................................ 36
3.3 Model results................................................................................................................ 37
3.4 Sensitivity analysis ....................................................................................................... 42
3.5 Summary ..................................................................................................................... 48
4 CALCULATION OF PRESSURE WITH CFD SOFTWARE................................................ 49

4.1 Introduction .................................................................................................................. 49
4.2 Motivations for CFD...................................................................................................... 49
4.3 Workflow in CFD .......................................................................................................... 50
4.4 Summary ..................................................................................................................... 56
5 INTERNAL ARC TESTING ............................................................................................... 58

5.1 Introduction .................................................................................................................. 58
5.2 Standardization and Test Experience ........................................................................... 59
5.3 Summary ..................................................................................................................... 70
6 EFFECT OF INTERNAL ARC ON STRUCTURES ............................................................ 72

6.1 Introduction .................................................................................................................. 72
6.2 Mechanical stress on switchgear due to the overpressure ............................................ 73
6.3 Mechanical stress on building walls due to overpressure .............................................. 83
6.4 Burn-through .............................................................................................................. 102
6.5 Summary ................................................................................................................... 111
7 INTERNAL ARC SIMULATION REVIEW ........................................................................ 114
Page 1
7.1 Introduction ................................................................................................................ 114

7.2 Information for the design review ................................................................................ 115
7.3 Comparison of actual and reference switchgear using design parameters................... 115
7.4 Design Evaluation ...................................................................................................... 118
7.5 Summary ................................................................................................................... 119
8 CONCLUSION ................................................................................................................ 120
ANNEX A EQUATIONS FOR PRESSURE RISE CALCULATION ....................................... 124

A.1 Introduction ................................................................................................................ 124
A.2 Equations for the basic model..................................................................................... 124
A.3 Basic model ............................................................................................................... 130
A.4 Enhancement of the basic model ................................................................................ 131
A.5 Summary ................................................................................................................... 138
ANNEX B PRESSURE SENSORS: TYPES AND ACCURACIES........................................ 139

B.1 Introduction ................................................................................................................ 139
B.2 Pressure sensors ....................................................................................................... 139
B.3 Accuracy of the measuring system ............................................................................. 140
B.4 Summary ................................................................................................................... 141
ANNEX C COTTON INDICATORS: ENERGY ABSORPTION AND FLAMMABILITY ......... 143

C.1 Introduction ................................................................................................................ 143
C.2 Testing for resistance against internal arcing .............................................................. 143
C.3 Apparatus overview .................................................................................................... 144
C.4 Overview of the test procedure ................................................................................... 144
C.5 Test results ................................................................................................................ 144
C.6 Summary ................................................................................................................... 150
ANNEX D ROOM PRESSURE CALCULATION USING CFD EXAMPLES.......................... 151

D.1 Introduction ................................................................................................................ 151
D.2 Pressure development and pressure relief .................................................................. 151
D.3 How to determine worst case scenarios for simulations. ............................................. 153
D.4 Summary ................................................................................................................... 156
ANNEX E EFFECTS OF NEUTRAL GROUNDING ON THE INTERNAL ARC .................... 157

E.1 Introduction ................................................................................................................ 157
E.2 Test Comparison ........................................................................................................ 157
E.3 Discussion.................................................................................................................. 159
E.4 Summary ................................................................................................................... 160
ANNEX F FIGURES FROM SENSITIVITY ANALYSIS ....................................................... 161

F.1 Introduction ................................................................................................................ 161
F.2 Sensitivity analysis for selective cases ........................................................................ 161
F.3 Pressure overshoot equation ...................................................................................... 174
F.4 Summary ................................................................................................................... 174
ANNEX G EFFECT OF REPLACING SF6 WITH AIR ON BURN-THROUGH ....................... 175

G.1 Introduction ................................................................................................................ 175
G.2 Arc voltage ................................................................................................................. 175
G.3 Heat of chemical reactions ......................................................................................... 176
G.4 Radius of Arc root....................................................................................................... 176
G.5 Velocity of Arc Motion................................................................................................. 177
G.6 Pressure exerted on the tank wall ............................................................................... 178
G.7 Summary ................................................................................................................... 178
Page 2
EXECUTIVE SUMMARY
Recognizing the increasing role of commercial and “homemade” modeling software in the power industry, CIGRE
Study Committee A3 established the former WG A3.20 to evaluate existing simulation tools and the extent to which
they can be used as verification tools. Using a case study based on dielectric design, WG A3.20 concluded that
simulation is a valuable development tool; it can accurately predict stresses and can provide good performance
extrapolation where test data is available on similar designs (interpolation). The “pure” performance prediction
(utilizing extrapolation) at current time is limited.
WG A3.24 has continued the analysis of the use of simulation as verification tools with a specific focus on internal
arc testing of Medium and High Voltage SF6 and air-filled equipment.
An arc fault is a high power discharge of electricity between one or more conductors. The unlikely event of the arc
fault can present a danger to the switchgear installation, personnel, even public. Since the possibility of the internal
faults can’t ever be dismissed, lots of effort has been dedicated to minimize its impact. The switchgear designers
design “arc resistance switchgear”, with reinforced constructions and utilizing safe controls to control the pressure
rise and expansion of the hot gasses through rupture discs and ducts. The system engineers design fast protection
systems to minimize the duration of the internal fault. The end users design installation rooms that with ducts and
openings to keep the hot gasses away from the personal and keep the pressure rise inside the room in the
acceptable range.
An important part to mitigate the hazard from the internal arc is to verify the internal arc withstand capability by
testing. The arc fault is initiated by a thin wire during such tests, resulting in the power arc and pressure rise inside
an arcing chamber. The goal of the test is to verify that personal in close proximity to the switchgear will not be
injured by indirect effects of the internal arc, such as flying parts or hot exhaust gasses.
There are number of existing arc resistance standards that define the test methodology and pass/fail criteria for the
arc resistance switchgear. Some standards cover both SF6 and Air-filled equipment like the international standard
for MV metal-enclosed switchgear, IEC 62271-200 (Ed. 2.0, 2011). This standard allows SF6 to be replaced by air
during test. Others, like IEEE Std C37.20.7 (2007) guide for internal arc testing does not address SF6 at all, as the
standard is only dedicated to air insulation switchgear. Lastly, there are standards that allows the extension of test
results by calculation methods, like the standard for HV Gas-Insulated switchgear, IEC 62271-203 (Ed. 2.0, 2011).
There is no doubt that conducting those tests is very important from safety perspective. On the other hand, it would
be economically and environmentally prohibitive to test every switchgear design variation. CIGRE A3.24 WG was
assembled to analyze the effects of the internal fault, research the available computational tools and models, and
advise if some tests can be replaced by computer simulations. Specifically, the motivation for A3.24 working was:
• To evaluate methods for pressure calculations and to benchmark with performed tests;
• To reduce the number of internal arc tests for environmental reasons by improvement of the switchgear
development process;
• To verify design modifications of the similar switchgear with simulations (by interpolation of the test results)
• To verify validity of replacing SF6 with air during arc resitance testing.
Page 3
Figure 0-1 shows the test setup for internal arc testing.
room simulation
indicator
Figure 0-1: Arc resistance test per IEC 62271-200 (Ed. 2.0, 2011).
WG A3.24 started the work by reviewing the existing literature (100+ white papers and applicable IEEE and IEC
standards, a number of which are referenced later in this technical brochure), and collecting the test data from
numerous Internal Arc Tests. Test data was collected for more than 80 different cases; with tank sizes ranging
from small 5 liters test tanks to large GIS tanks (1200 liters), with fault currents ranging from 12 kA to 63 kA, with
fault durations ranging from 10 ms to 1.2 s, including single compartment and multi-compartment equipment and
including both SF6 and air-insulated switchgear (Figure 0-2).
The WG then reviewed existing software tools for calculating the effects of an internal arc fault, focusing on 3 main
effects of an internal arc:
a. Pressure rise
b. Mechanical stress
• on switch enclosure
• on installation rooms and buildings
c. Burn-through
This software review included various “home made” software tools, ranging from simple spreadsheets in Excel that
most engineers could use with a little effort, to a complex 3-D Computational Fluid Dynamic (CFD) software
package whose application remain limited to small number of experts due to the complexity and cost of the
software.
 Pressure rise: After calculating the pressure rise in a number of cases with simple “home made” tools,
the WG realized that the calculated pressure peak was within 10-20% of the measured peak, which
indicated that usage of the simpler tool should be explored. Encouraged by that finding, the WG has
developed a set of “basic equations” and validated this mathematical model for all 70+ cases. It has been
found that calculation of pressure curves inside the arc compartment during an internal arc fault gives good
agreement between test and simulation as long as the input arc energy is known. These findings are
covered in Chapter 2, which lists the equations for the basic model and identifies its benefits and
Page 4
limitations. Also, it shows different applications of the basic model for both MV and HV switchgear. Detailed
set of equations for the basic model is provided in ANNEX A. This can be used to create one’s own “home
made” tool. In addition to the basic model, this Technical Brochure also covers an enhanced model and
pressure rise calculations with CFD software. For better understanding of the relations between the
parameters and their effects on the pressure rise, refer to Sensitivity Analysis covered in Chapter 3 and
ANNEX F.
Figure 0-2 : Snapshots of some Internal Arc Tests for which A3.24 WG collected data (current, voltage,
pressure) and compared the measured vs calculated pressure rise
 Mechanical Stress: Overpressure generated during the internal arc fault test causes mechanical stress on
the switchgear enclosures and on building walls. Chapter 6 provides guidelines for calculating mechanical
stress using Finite Element Analysis (FEA) and CFD software tools.
 Burn-through: This effect is caused by the arc which can burn on a surface of the metallic enclosure (like
a switchgear wall or panel, or GIS bus duct), and it melts and punctures walls. It is covered in Section 6.4.
Lastly, in effort to minimize Internal Arc tests, the working group created guideline for Internal Arc Simulation
review. Chapter 7 provides the guideline to replace the internal arc withstand test of the specific switchgear by
performing a design analysis based on tests of a similar design. A criteria for comparing the switchgear designs
and determining if the testing can be replaced by calculations is provided.
Page 5
1 INTRODUCTION
1.1 Ove rview
An internal arc fault is an unintentional discharge of electrical energy within an enclosure. When the internal arc
fault occurs, the available short circuit current will flow through the arc between phases (see Figure 1-1) and/or
from phase(s) to ground. The energy released from an electrical arc heats the SF6 gas or the air within the
enclosure, resulting in a pressure rise.
Figure 1-1: A 13 kA electric arc moving between two conductors, at 5 cm distance.

Two ongoing trends in the power industry are causing the possible damage from internal arc faults to increase.
One is the increase in the available fault current levels, resulting in an increase in the available arc energy. The
other is the evolution towards more compact switchgear, which results in smaller enclosures. These reduced tank
volumes result in a higher rate of pressure rise, higher temperatures and larger electro-magnetic forces on the
conductors. On the other hand, minimizing SF6 gas release to the environment is becoming a more and more
important issue and IEC 62271-200 (Ed. 2.0, 2011) actually permits SF6 to be replaced by air in an internal arc test.
Pressure rise, temperature rise and arcing behavior, however, might be quite different between SF 6 and air,
because the relevant physical properties of these gases, such as specific heat capacity, density, etc. are
significantly different. Therefore any reasonable way to achieve an equivalent demonstration of the performance of
the test object using air or alternative gases should be explored.
The incidence of internal arc faults in MV and HV switchgear is rare, but when an arc fault occurs in an electrical
installation it may seriously damage the electrical equipment and the switchgear buildings (see Fehler!
Verweisquelle konnte nicht gefunden werden.), and endanger personnel.
Figure 1-2: Internal arc fault test.
I typically occur due to following reasons:

Page 6
 external influences
 material or mechanical defects
 incorrect operation
The physical results of an internal arc fault are significant. An internal arc fault has the following physical impacts:
 Pressure inside a small closed enclosure (volume 200 liters) can accelerate to 12 bars in 4 cycles of power
frequency during a 25 kA fault.
 Arc temperature can exceed 10,000 °C.
 The arc energy inside the arc compartment from a 25 kA fault for ¼ second is comparable to the energy
released by exploding 2 kg of dynamite (around 8000 kJ).
 Sound levels can reach 160 dB. By comparison, shotgun noise is measured to be in the range of 150 – 160
dB.
 Debris may travel at speeds up to 1000 km/h.
 Vaporized copper expands to 67,000 times its solid volume (1 cm3 of copper vaporises into 67 liters of
vapor). For comparison: the conversion of water into steam has an expansion factor of 1670.
 The resultant force of the expelled gases following rupture may reach several tons on the walls of an
enclosure or walls of an installation room.
 The temperature of the hot gases streaming out of an arcing compartment may exceed 1000 °C.
If a fault arc occurs in an electrical installation, the electrical energy of the arc plasma is transferred to its
surroundings by various different mechanisms. The pressure rise is mainly due to heat transfer. The energy input
into the fault arc by Joule heat is balanced by the interactions of the arc column with the electrodes, the arc length
and by several energy exchanges. This includes heat conduction, radiation and gas convection inside the
compartment where the arc occurs, and also through relief openings in the enclosure of the compartment. The
convective transfer of heat and the mass of the gas cause a change in the internal heat of the surrounding gas, and
is therefore part of the overall pressure rise. Additionally, evaporated metal from the arc roots release additional
energy by exothermal reactions with the surrounding gas. As a result, during internal arcing a rather large amount
of energy is released in the cubicle and into the environment within a short period of time.
Although the physical processes of energy transfer from the arc to the surroundings can be described in a general
manner, it is difficult to describe them quantitatively.
The pressure rise resulting from an internal arc in a compartment can be calculated in a number ways as listed
below:
1. Basic models:
The calculation of gas pressure is based on gas temperature according to the general gas equation and
on mass flow balance through pressure relief openings. The compartment, where the arc is ignited, and
other connected rooms are described by their effective volumes and openings among them. Gas
properties are assumed to be independent of temperature and pressure.
2. Enhanced models:
Page 7
These models are based on the same basic equations, effective volumes and openings. Some of them
consider temperature and pressure dependent gas properties. They may be extended by including further
effects such as exothermic reactions, ablation of material and mixing of gases.
3. CFD models:
The calculation of gas pressure and temperature is based on the fluid-dynamic equations describing the
conservation of mass, momentum and energy of the gas in each finite volume element. The system of
equations is solved three-dimensionally with a computational fluid dynamics (CFD) solver.
The calculation approach in Chapter 2 describes the basic model in detail and provides a comparison with
measured data. Enhanced and CFD models are presented and discussed in the brochure (Section 2.6 and Chapter
4). Table 1-1 summarizes the application range and the limitations of the three approaches.
Approach /model Appropriate Application Limitations
 Doesn’t consider spatial non-uniformity of gas

parameters (pressure, temperature, density) in each
To quickly calculate uniform volume part.
pressure rise inside an arc  Not applicable if the relief opening area is too large
Basic compartment and the exhaust in relation to the compartment volume.

1)
(low complexity) volume in typical MV  Calculations are not reliable, when gas temperature
switchgear and HV GIS exceeds approx. 2000 K for SF6 and 6000 K for air.
applications.  Doesn’t consider gas mixtures in the exhaust
compartment.
3
 Not applicable for large installation rooms (>50m )
 Doesn’t consider spatial non-uniformity of gas

To calculate uniform pressure
parameters (pressure, temperature, density) in each
Enhanced rise as under 1) adding further
2) (medium volume part.
approximations to better match
complexity)  Limitations and applications depend on the
test results and calculation.
implemented approximations.
For calculating spatial  High effort for the modeling and meshing of the
CFD pressure distribution and gas rooms and switchgear
3)
(High complexity) flow in odd shapes geometry  Requires large computing power and time.
and large rooms.
Table 1-1: Model selection table.
Page 8
1.2 Abb re viations

a.c. alternating current
AIS air insulated switchgear
AISI American Iron and Steel Institute
Al aluminum
CAD computer-aided design
CB circuit breaker
CFD computational fluid dynamic
Cu copper
CV control volume
d.c. direct current
EEMAC Electrical Equipment Manufacturers Association of Canada
EMC electromagnetic compatibility
FDM finite difference method
FEA finite element analysis
FVM finite volume method
GIS gas insulated switchgear
HV high voltage
IAC internal arc classification
IP ingress protection
IR infrared
LV low voltage
MV medium voltage
NFPA National Fire Protection Association
PDE partial differential equation
PPE personal protection equipment
RMS root mean square
RMU ring main unit
SF6 sulfur hexafluoride
TO test object
WG working group
Page 9
WPI Worchester Polytechnic Institute
3-D 3-dimensional
1.3 Definitions
Ablation: Removal of material from the surface of an object by vaporization, or other erosive processes.
Arc: High current electric discharge between electrodes in a fluid (liquid or gas).
Arc absorber: Meshes, grids, perforated metal sheets or similar devices placed in one or more layers into the
exhaust gas flow in order to absorb some energy of the arc exhaust.
Arc blast: Direct pressure wave (gas pressure either of hot plasma gases or cold gases, air or a combination of
these) that can cause damage to humans, equipment, and surroundings
Arc compartment: Enclosed part of metal-enclosed switchgear, where an arc fault occurs. Openings necessary
for pressure relief, control, or ventilation may be present.
Arc duct: Channel connected to the arc-exhaust intended to lead the arc products to another place.
Arc exhaust: The expulsion of hot gases from an arc fault through the relief opening of an arc compartment.
Arc fault: A high power discharge of electricity caused by a breakdown of insulation or flashover generating
excessive heat.
Arc fault – single phase: Arc fault occurring between one conductor and ground.
Arc fault – three phase: Arc fault occurring between three conductors or between three conductors and ground.
Arc flash: Direct physical phenomenon such as flame due to the hot plasma expansion of an arc fault. This can
cause burns and fire, and impacts humans as well as equipment and surroundings.
Arc plasma: Thermal plasma generated by an arc.
Arc power: Active (electrical) power of an arc given by the product of momentary current and voltage measured at
the terminals of the test object.
Arc voltage: Voltage which appears between the electrodes of an arc.
Arcing time: The time elapsing from the ignition of an arc to the interruption of the current.
Available (prospective) current: The current that would flow in a circuit if each pole of the switching device was
short-circuited by a link of negligible impedance without any other change in the circuit or the supply.
Basic model: Mathematical approach for the calculation of pressure rise due to an internal arc using simplified
equations under basic assumptions.
Burn-through: A hole burnt through the walls of the equipment enclosure or compartment by an arc.
Computational fluid dynamics (CFD): A branch of fluid mechanics that uses numerical methods and algorithms
to solve and analyze problems that involve fluid flows.
Deflection: The degree to which a structural element is displaced or bent under a mechanical load. It may refer to
an angle or a distance.
Page 10
Deflectors: Plates placed in the flow of exhaust gas to deflect the stream of gas.
Discharge coefficient: Ratio of effective opening area of a pressure relief device to its geometric area.Considers
the effect of discharge of gases through real and ideal nozzles.
Dynamic pressure: The pressure on a surface at which a flowing fluid is brought to rest in excess of the pressure
of the fluid at rest (static pressure).
Enclosure: A surrounding case or housing used to protect the enclosed equipment and to prevent personnel from
accidental contact with live parts.
Enhanced model: Mathematical approach for the calculation of pressure rise due to an internal arc applying a
number of extensions to the basic model with respect to equations and assumptions.
Exhaust channel: See arc duct.
Exhaust compartment: Enclosed volume adjacent to the arc compartment which receives the arc exhaust.
Exothermic: In thermodynamics, the term exothermic ("outside heating") describes a process or reaction that
releases energy from the system.
Fault-shorting switch (arc killers): Eliminates arc faults by creating a metallic short circuit. It generally has a
sensor to detect the arc and an grounding device to extinguish it. The sensor can be either sensitive to the light
generated by the arc or to the pressure reached in gas sealed tank.
Finite element analysis (FEA): A numerical technique for finding approximate solutions of partial differential
equations (PDE) as well as integral equations.
Finite volume method (FVM): A method for representing and evaluating partial differential equations in the form of
algebraic equations. "Finite volume" refers to the small volume surrounding each node point of a mesh.
Heat capacity ratio: Ratio of the specific heat of a gas taken at constant pressure to that taken at constant volume
also termed “adiabatic index”.
Heat conduction: A mode of transfer of energy within and between bodies of matter, due to a temperature
gradient. Conduction means collisional and diffusive transfer of kinetic energy of particles of tangible matter.
Heat convection: Heat transfer by the flow of a fluid in regions with different temperatures. Convective heat and
mass transfer take place through both diffusion (the random Brownian motion of individual particles in the fluid) and
by advection, in which matter or heat is transported by the larger-scale motion of currents in the fluid.
Heat flux: Heat flux or thermal flux, sometimes also referred to as heat flux density or heat flow rate intensity is a
flow of energy per unit of area per unit of time.
Heat flux sensor: A transducer that generates an electrical signal proportional to the total heat rate applied to the
surface of the sensor. The measured heat rate is divided by the surface area of the sensor to determine the heat
flux.
Heat radiation: Emission and propagation of energy in the form of electromagnetic waves.
Heat transfer: Transfer of heat e.g. from an arc to its surroundings. Heat transfer is classified into various
mechanisms, such as heat conduction, convection, thermal radiation, and transfer of energy by phase changes.
Page 11
Infrared thermography: Thermal imaging cameras detect radiation in the infrared range of the electromagnetic
spectrum (roughly 9000–14,000 nanometers or 9–14 µm) and produce images of that radiation, called
thermograms. The amount of radiation emitted by an object increases with temperature; therefore, thermography
allows one to see variations in temperature.
Joule heating: Heating caused by an electric current through a resistive material.
-factor: Ratio of that part of the arc power (or energy) responsible for the heating of gases inside the arc
compartment to the total arc power (or energy).
Longitudinal wave: Also known as "l-waves", are waves that have the same direction of vibration as their direction
of travel, which means that the movement of the medium is in the same direction as or the opposite direction to the
motion of the wave.
Metal-enclosed switchgear: A switchgear assembly completely enclosed by sheet metal (except for ventilation
openings and inspection windows) containing primary power circuit switching or interrupting devices, or both, with
buses and connections, which may also include control and auxiliary devices. Access to the interior of the
enclosure is provided by doors or removable covers.
Metal evaporation: Heating a metallic part up to a temperature, at which a considerable amount of metal vapor is
released from its surface.
Navier-Stokes equation: The Navier–Stokes equations describe the motion of fluid substances. These equations
arise from applying Newton's second law to fluid motion, together with the assumption that the fluid stress is the
sum of a diffusing viscous term (proportional to the gradient of velocity), plus a pressure term.
Net compartment volume: Effective volume of a compartment after subtraction of the volume of all built-in
components.
Overpressure: Pressure above ambient pressure within a pressurized enclosure.
Pad-mounted: A general term describing switchgear equipment positioned on a surface-mounted pad located
outdoors. The equipment is usually enclosed with all exposed surfaces at ground potential.
Pressure relief device: A device which opens on overpressure, releasing gases from a compartment into the
ambient atmosphere. A pressure relief device can be a loose flap, a bursting disk, or even a constant opening to
the outside world.
Pressure withstand: Maximum pressure which can be withstood by an enclosure.
Relief opening area: Area provided by a pressure relief device to expel hot gases.
Response (bursting) pressure: Pressure at which a pressure relief device is ruptured or opens.
Rupture (bursting) disc: A non-reclosing pressure relief device that, in most uses, protects a pressure vessel,
equipment or system from overpressurisation. A rupture disc is a type of sacrificial part because it has a one-time-
use membrane that fails at a predetermined differential pressure, either positive or negative.
Short circuit: An abnormal connection (including an arc) of relatively low impedance, whether made accidentally or
intentionally, between two points of different potential.
Page 12
Static pressure: The pressure exerted by a liquid or gas when the bodies on which the pressure exerted are not in
motion.
Switchgear: A general term covering switching and interrupting devices and their combination with associated
control, metering, protective, and regulating devices; also assemblies of these devices with associated
interconnections, accessories, enclosures, and supporting structures, used primarily in connection with the
generation, transmission, distribution and conversion of electric power.
Von-Mises stress: The stress associated with the deformation of material such that the actual distortion energy is
equivalent to uniaxial simple tension case.
Yield point (yield strength): is defined in engineering and materials science as the stress at which a material
begins to deform plastically. Prior to the yield point the material will deform elastically and will return to its original
shape when the applied stress is removed. Once the yield point is passed, some fraction of the deformation will be
permanent and non-reversible.
Young’s modulus: It is defined as the ratio of the uniaxial stress over the uniaxial strain in the range of stress in
which Hooke's Law holds. It can be experimentally determined from the slope of a stress-strain curve created
during tensile tests conducted on a sample of the material.
1.4 Referred standards

EEMAC G14-1, Procedure For Testing The Resistance Of Metal Clad Switchgear Under Conditions Of Arcing Due
To An Internal Fault, Electrical Equipment Manufacturers Association of Canada (EEMAC), G14-1, 1987.
IEC 61482-1-2 Ed.1.0: 2007, Live working – Protective clothing against the thermal hazards of an electric arc. Part
1-2: Test methods - Method 2: Determination of arc protection class of material and clothing by using a constrained
and directed arc (box test).
IEC 60076-5 Ed. 3.0: 2006, Power transformers – Part 5: Ability to withstand short circuit.
IEC 60298 Ed. 3.0: 1990, A.C. metal-enclosed switchgear and controlgear for rated voltages above 1 kV and up to
and including 52 kV.
IEC 62271-200 Ed. 2.0: 2011, High-voltage switchgear and controlgear – Part 200: AC metal-enclosed switchgear
and controlgear for rated voltages above 1 kV and up to and including 52 kV.
IEC 62271-201 Ed. 1.0: 2006, High-voltage switchgear and controlgear – Part 201: AC insulation-enclosed
switchgear and controlgear for rated voltages above 1 kV and up to and including 52 kV.
IEC 62271-203 Ed. 2.0: 2011, High-voltage switchgear and controlgear – Part 203: Gas-insulated metal-enclosed
switchgear for rated voltages above 52 kV.
IEEE Std 1584-2002, IEEE Guide for Performing Arc-Flash Hazard Calculations.
IEEE Std C37.20.7-2007, IEEE Guide for Testing Medium Voltage Metal-Enclosed Switchgear for Internal Arcing
Faults.
NFPA 70E-2012, Standard for Electrical Safety in the Workplace, National Fire Protection Association (NFPA), 1
Batterymarch Park, Quincy, MA 02169-7471.
Page 13
2 CALCULATION OF PRESSURE USING A BASIC MODEL

2.1 In troduction
The first part of this chapter (Sections 2.2-2.4) focuses on describing the basic equations, assumptions and
limitations of the basic model. The derivation of the equations is given in ANNEX A.2. In Section 2.5 the pressure
curves measured in several selected test cases are compared with the results of calculations taking the actual test
arrangements and measured arc current and voltage into account. In Section 2.6 and ANNEX A.4, modifications of
the basic equations leading to enhanced models, which improve the agreement between simulation and test results
for certain aspects are presented.
2.2 Equations of the basic model

2.2.1 Basic arrangement and quantities
Figure 2-1 shows schematically an installation consisting of arc compartment, exhaust compartment, and
installation room/environment. The arc represented by the temporal development of energy input is ignited in the
arc compartment with volume . A pressure relief opening with cross-section connects the arc to the exhaust
compartment with volume . When the pressure in the arc compartment reaches the response pressure, the
relief device opens and gas flows into the exhaust compartment . From there, gas flows through the opening with
cross-section A23 into the installation room or environment with volume .
Arc Compartment Exhaust Compartment Installation Room
V1 V2 V3
p1 A12 p2 A23 p3
m 12 m 23
T1 T2 T3
Q1
Figure 2-1: Principal arrangement and quantities used for pressure calculation.
The type of insulating gas in each volume is characterized by the corresponding heat capacity ratio (adiabatic
index) and the specific gas constant . In the basic model, these quantities are assumed to be constant. The initial
state of the gas is defined by pressure and temperature . While volume may be filled with air or SF6, volumes
and are typically always filled with air.
Mass, density and specific heat constants of the gas in volumes and are (see ANNEX A.2.1):
Page 14
(2-1)
= =
= = (2-2)
= = (2-3)
−1 −1
= = (2-4)
The thermal energy as a part of the electrical energy heats up the gas.
= (2-5)
The thermal transfer coefficient , which is described in Section 2.3.6, describes the relationship between the
thermal and electrical energy. In the basic model, is taken as constant. The evaluation of the electrical energy
from arc voltage and current is explained in Section 2.3.5.
2.2.2 Mass flow
In the following, all time-dependent quantities are considered before and after a time step Δ . The time step Δ is
chosen so that the change of temperature and pressure in consecutive time steps is small. In particular after
opening of the relief device, the gas temperature might increase rapidly requiring a time step below 10 µs to
prevent numerical instability. The mass flow from the arc compartment into the exhaust compartment is given by:
Δ = Δ (2-6)
is the discharge coefficient, which considers the contraction of gas flow through an opening. Obstacles in the
gas flow such as a metallic mesh or lamellas may be included in this coefficient ( is the effective opening).
Reaching the response pressure in , the pressure relief device opens. , and are the gas pressure, gas
density and gas velocity within the opening , which are different from the values in and [Schmidt1958].
= (2-7)
2
= 1− (2-8)
−1
If the ratio of pressure in and i.e. / exceeds a value of 1.89 for air and 1.70 for SF 6, respectively, is
determined by the critical pressure ∗
; for smaller ratios is equal to the pressure in .
( , ∗)
= (2-9)
Page 15
∗
2 (2-10)
=
+1
The mass in volume is reduced after the time step Δ by the mass Δ
Δ = −Δ (2-11)
The mass from the exhaust compartment flowing into the installation room ( ) is given by:
Δ = Δ (2-12)
Density and flow velocity are calculated using the equations provided above with all indices incremented by one.
The change of mass in volume within Δ is the difference between the incoming mass Δm and the outgoing
mass Δm during Δ .
Δ =Δ −Δ (2-13)
2.2.3 Gas temperature
The temperature change in the arc compartment with volume after the time step Δ is determined by the
difference between the thermal energy input by the arc (Δ ) and the energy loss due to gas flow out of the
compartment (see ANNEX A.2.4 for details):
Δ −Δ −
Δ = (2-14)
The corresponding temperature change in is:
Δ − −Δ −
Δ = (2-15)
The summation of all temperature changes Δ provides the temperature at time .
2.2.4 Gas pressure
With given gas mass and temperature, the pressure in and at time is given by the ideal gas law:
( − 1) ( − 1)
= = (2-16)
Temperature and pressure in the installation room are calculated correspondingly.
2.3 Input para meters

2.3.1 Gas data
The model requires the input of two basic gas quantities, i.e. the heat capacity ratio and the specific gas constant
. They depend on the specific heats, and , (see Section 2.2.1). These quantities are assumed to be constant
and are given in Table 2-1 taken from publications. The value of is calculated directly from and using
Page 16
equations (2-3) and (2-4). The assumption of being independent of temperature is acceptable up to
temperatures where gases start to dissociate remarkably (approximately 2000 K for SF6 and 6000 K for air). For
higher temperatures, distinct maxima in the specific heat capacity curves occur which change the values of and
significantly (see Section A.4.8). This limits the applicability of the model. Typically such high gas temperatures
are reached in the arc compartment after opening of the relief device.
Parameter Air SF6

[Mende1975] [Solvay]
-1
specific heat capacity at constant volume 716 608 J kg /K
-1
specific heat capacity at constant pressure 1005 665 J kg /K
gas density 1.205 6.07 kg/m3
heat capacity ratio (kappa) 1.403 1.0936
molar mass 29 146 kg/kmole

-1 -1
universal (molar) gas constant 8314 8314 J K kmole
-1 -1
specific gas constant 287 56.9 J K kg
Table 2-1: Basic gas quantities at normal conditions (20 °C and 101.3 kPa).
For SF6 insulated switchgear, the flow of SF6 out of the arc compartment leads to an SF6/air mixture in the exhaust
compartment. Gas mixing is not considered in this model, i.e. the gas properties in the exhaust compartment are
taken to be those of pure air. However the mass and energy exchange is considered correctly. This simplification is
reasonable as long as the SF6 concentration is low (e.g. in large exhaust compartments). For high concentrations,
gas data from mixtures have to be considered (see Section A.4.4).
2.3.2 Volume
All volumes in the model are net volumes i.e. volume of compartment minus volume of built-in components. The
shape of the compartment is not considered. The volume of the built-in components might reach 10 to 20 % of that
of the arc compartment. The energy supplied by the arc is taken as homogeneously distributed inside the arc
compartment which is reasonable as long as the arc compartment is small, i.e. below 1m3. That is why the model
does not cover pressure waves, which might play a role in long, narrow compartments (channels).
2.3.3 Pressure relief opening
Relief openings are represented by effective areas, i.e. the geometric cross-section of the opening diminished by
the area of frames, slats, grills etc. (see Section 2.2.2). The discharge coefficient is assumed to be 0.7 for air
[Dubbel1997] and 0.8 for SF6 [Anantavanich2008]. The opening of the relief device occurs instantaneously at the
response pressure. For heavy relief flaps, the opening process possibly needs some time and might increase the
over-pressure in the compartment.
Page 17
If openings are large compared to the volume of the compartment, then the equations of the model are no longer
applicable. A critical dimension could be in the order of 10 % of the area of a side surface assuming the net volume
is a cube.
The response pressure of the relief device is most often given as a static value. If the value is determined from
arcing tests, it is a “dynamic” value. Experience shows that the difference between both is of the same order as the
scatter from sample to sample.
2.3.4 Arc current
For pressure calculation the temporal development of the single or three-phase fault current must be known. This
current development can be taken from test or from simulation. The asymmetry of the short circuit current might
influence the initial pressure build-up, however, has little influence on the later pressure curve. As long as the
supply voltage is much higher than the arc voltage, the arc resistance does not influence the current asymmetry
(see Section 5.2.4). Knowing the d.c. time constant of the circuit . ., the temporal development of the current ( )
in a three-phase system can be calculated using e.g. the formula:

= √2 sin + − − sin − . . (2-17)
inserting the angular frequency , the angle of fault initiation , and the shift between the phases, .
2.3.5 Arc voltage
The arc voltage is one of the most important parameters determining arc energy and hence the pressure rise. In
tests the arc energy is determined from measured line currents and phase-to-ground voltages:
Δ =( + + )Δ (2-18)
The term in brackets is the arc power. The arc voltage is the voltage drop of an arc between its roots. For pressure
calculations (and the comparison with measured values) averaged arc voltages are used. In a three-phase
system, the phase-to-ground voltage may not be identical with the arc voltage. If the arc appears between the
phases only two arcs appear simultaneously commutating between the phases. In this case the arc energy during
Δ is given by [Welich1984]:
Δ = 0.5 (| | + | | + | |)Δ (2-19)
If an arc burns between phase and ground (enclosure wall), the arc voltage is identical with the phase-to-ground
voltage and the arc energy during Δ is given by:
Δ = (| | + | | + | |)Δ (2-20)
Because of the ambiguity of equations (2-19) or (2-20), it is recommended to always use equation (2-18) and the
three measured phase-to-ground voltages for the determination of the arc energy. In principle, the mean arc
voltage, , over a certain period can be determined by equating the measured arc energy with equation (2-19).
This gives the correct mean arc voltage for a phase-to-phase arc. For simplicity, sometimes equation (2-20) is used
Page 18
as an input for calculation; however, it will give only correct arc energies when the lower (approximately factor of 2)
arc voltage for a phase-to-ground arc is used. This has to be kept in mind.
Arc voltage fluctuates, e.g. due to arc looping, and varies in time, e.g. caused by strong evaporation of electrode
material. The arc voltage differs between an arc ignited in an empty or in a fully equipped compartment. Some
experiments show that the arc voltage increases with rising pressure in the compartment and decreases later after
pressure reduction [Dullni1994].
If arc voltage data is missing, basic formulas, which have been extracted from three phase internal arc tests with
MV metal enclosed switchgear separately for air and SF6 and copper electrodes might be used [AiF2011]. These
voltages have to be applied together with the energy equation (2-19) for arcs between phases.
V 1 V V
= 30 + ≤ 40 (air) (2-21)
cm 2 cm kA cm
V 1 V V
= 40 + ≤ 50 (SF ) (2-22)
cm 2 cm kA cm
Here is the arc voltage between phases, is the distance between pole centres, and is the effective short
circuit current.
For single phase HV aluminum enclosures filled with 1 to 4 bar of SF6 the following formula has been derived from
a survey of available data for certain specific conditions (details see [König1984]).
= 250 + − 50 + 4 (2-23)
mm kA
Here is the arc voltage including the -factor, is the clearance between conductor and wall, and is the
effective short-circuit current.
Internal arc faults in electrical switchgear in 3-phase arrangements typically start as a phase to ground or as a
phase to phase fault. If this arc fault cannot be eliminated automatically in a short time (in the millisecond range),
and no single pole solid insulation is present, it will most likely develop into a three-phase arc fault. A three phase
arc fault consists of two parallel arcs with arc roots changing among the three phases. In the case of close
proximity of the enclosure, the fault may rapidly develop into three parallel phase-to-ground arcs.
Depending on the network voltage, the time constant of the asymmetrical arcing current will decrease to about half
of the value for a solid short circuit (12 kV, 20 kA will result in 22 ms time-constant out of the standard 45 ms). This
means that a fully asymmetrical current will sooner become symmetrical due to the arc resistance than the current
from a solid short circuit would do.
Page 19
Figure 2-2: Momentary arcing powers. Asymmetrical (τ = 45 ms) first 5 periods (100 ms) and symmetrical
one period.
2.3.6 - factor
The heat transfer coefficient determines the fraction of the electrical arc energy, which directly results in a
pressure rise in the arc compartment. Theoretical approaches to calculate based on a detailed energy balance
e.g. [Zhang2002] exist, however, this approach is not really feasible. In practice, is determined by fitting the
calculated pressure rise to the measured one before the operation of the pressure relief device. If exothermic
reactions occur, e.g. between aluminum and SF6, additional energy might heat up the gas resulting in a -factor
possibly larger than 1.
It is highly recommended that -factors are determined from tests under similar conditions. If no experiments are
available, -factors taken from publications should be used with caution. It has been found that in general in air
is lower than in SF6 [Dullni1994, Friberg1995]. Based on these experimental investigations, for air at standard
conditions with copper electrodes ranges from 0.40 to 0.65 and in case of SF6 from 0.50 to 0.70. For aluminum
electrodes might be higher due to exothermic reactions. The authors have calculated values of up to 1.6 with
the basic model. It has also been found by experiment that depends on gas density. This is important for
compartments with relief openings, where gas density is considerably reduced during the exhaust of gas (see
Section A.4.1). In general, decreases with gas density [Dullni1994, Zhang2002].
2.4 Application limits o f the basic model

Due to the assumptions implemented in the basic model, the user should be aware of how the application range is
limited. Typical applications with reliable results are:
 Pressure rise in a closed arc compartment in air and SF6

 Pressure rise in an arc compartment with relief opening in air and SF6
 Energy transfer from the arc to the exhaust compartment
Page 20
The limitations to be considered are as follows:
 The simulation of pressure development is feasible until the dissociation temperature is reached in the arc
compartment. The calculation should be stopped at temperatures of about 6000 K in air and about 2000 K
in SF6. In fact, here the calculations have been continued up to 20000 K for both gases. The temperature
limit is reached faster the smaller the volume of the arc compartment, the higher the arc energy and the
larger the relief opening (due to the reduced gas density).
 The model does not consider the evaporation of metal or insulation material, which influences gas
composition, density and temperature. This influence is negligible as long as the density of the insulating
gas (air or SF6) is larger than the density of the vapor. Due to the exhaust of gas from the arc
compartment, the density of the insulating gas drops within this compartment, and the proportion of gas
resulting from evaporation becomes more and more important. For long arcing times the gas in the arc
compartment may practically consist only of evaporated material. Because evaporation is not considered in
the model, the simulation results become uncertain when the insulating gas density in the arc compartment
drops significantly.
 If considerable gas flow occurs in any compartment (e.g. in elongated rooms or channels) the approach
with spatially averaged quantities is not applicable. As a consequence, reliable results are only achieved
for pressure relief opening areas limited to not more than 10 % of the side surface of the arc compartment
(see Section 2.3.3). Also, the compartments should have simple connectivity. Thus, for exsample, single
phase compartments interconnected by gas tubes are not considered. They would require the extension of
the basic model to more volumes, which is out of the scope of the basic model.
 The energy transfer out of the arc compartment determines the pressure rise in the exhaust compartment.
The model assumes a constant gas type in the exhaust compartment. Typically the exhaust compartment
is filled with air. In case of SF6-insulated switchgear the gas in the exhaust compartment will be a mixture
of SF6 and air. Therefore the assumption of only air in the exhaust compartment is violated when the SF 6
portion becomes remarkable. This might be the case for small exhaust volumes. These considerations do
not affect air-insulated switchgear.
Some of these limitations can be overcome to some extent by additional approaches (e.g. enhanced models, see
Section 2.6). These include for example the application of real gas data depending on temperature and pressure,
the evaporation of metal and insulation material, gas mixtures, the density dependence of the -factor and
exothermic reactions. Such modifications do not really increase the accuracy of the pressure calculation in the arc
compartment, but allow the extension of the calculation for longer arcing times and a calculation of the pressure
rise in the exhaust compartment or installation room. Spatial resolution, if necessary, can be achieved using CFD
tools. Anyhow these are in principle subject to the same limitations as discussed above.
2.5 Application of the basic model to selected test cases

2.5.1 General
Within the CIGRE A3.24 working group 70 data sets of internal arc tests covering MV and HV tests of air and SF 6
insulated switchgear have been collected and recalculated with the basic model. All of them have been evaluated
concerning typical pressure-related parameters. The test arrangements of the selected cases are shown in Figure
2-4 to Figure 2-9. These cases can be used by readers to benchmark their own calculation programs.
Page 21
The following procedure has always been applied:
 insulation gas and filling pressure are taken from the test conditions;
 geometrical input values are derived from available drawings;
 discharge coefficient is chosen between 0.7 to 1 (adapted to the measured pressure decay);
 energy input is based on the measured currents and averaged phase-to-ground voltages using equation
(2-18) (see Section 2.3.5);
 phase angles and dc component are adapted to fit measured currents; most often a peak factor of between
2.4 and 2.6 and a dc time constant of 22 ms is applied;
 -factor results from adapting the calculated to the measured pressure rise (slope ∆ /∆ up to the
response pressure of the relief device);
 response pressure pburst of the relief device is taken from measured pressure curves;
 the limitations described in Section 2.4 are respected.
The input parameters and initial values of the selected cases are listed in Table 2-3, Table 2-4 and Table 2-5
respectively. Some details of the test arrangements or switchgear configurations are given in Section 2.5.2.
Calculated and measured pressure curves are shown in Figure 2-10 to Figure 2-17. The red curves show the
calculated pressure in the arc compartment, the blue curves the pressure in the exhaust compartment, if available.
Grey curves present measured data. The simulation stops when the gas temperature in the arc compartment
reaches 20000 K. The changes in the gas pressure in the arc and exhaust compartments are characterized by
values determined from measured pressure curves. These values are depicted in Figure 2-3 and listed in Table
2-2.
Peak pressure Maximum pressure measured after opening of the relief device
Time to peak Time between start of pressure rise and maximum pressure
Response pressure Pressure value, at which the relief device opens
Response time Time, at which the relief device opens
Decay time Duration of pressure drop starting from peak pressure
Pressure slope Initial rise of pressure until
Table 2-2: Characteristic values for the pressure curve.
The change of gas pressure in the exhaust volume can be characterized by the two values and . The
onset of pressure rise in this volume is identical to determined from the pressure rise in the first volume.
When the arc compartment is already filled with gas at e.g. rated pressure, the pressure curve starts above the
zero line as shown in Figure 2-3.
Page 22
Figure 2-3: Characteristic values determined from calculated or measured pressure curve.
2.5.2 Test arrangements
 Welded cube-shaped steel container

 Single phase Cu terminals
 Linear electrode arrangement
 Arc ignited between two electrodes
 Circular relief device with bursting disc
 Exhaust into open air

Side view of container
Figure 2-4: Test cubicle used for cases A and E.
Page 23
Use of graphical symbols:
 Red arrow: Current in-feed

 Brown cones: current bushings
 Yellow symbol: arc ignition point
 Blue area: relief device
 Light green: SF6 or air-filled volume
 Welded switchgear housing (steel)

 Three phase Cu terminals
 Side-by-side electrode arrangement
 Arc ignited between Cu electrodes
 Rectangular relief device with bursting
disc
Top view of switchgear housing.
Figure 2-5: Switchgear used for case F.

 Three phase Cu bars
 Side-by-side electrode arrangement
 Arc ignited in fully equipped bus bar
compartment
 Exhaust into Channel
Front view of bus bar compartment
Figure 2-6: Switchgear used for cases D and G.
Page 24

 Double phase terminals (cable plugs)
 Side-by side electrode arrangement´
 Arc ignited in cable compartment
 Rectangular relief device with flap
 Combination of all volumes into one
Front view of switchgear (RMU)
Figure 2-7: Switchgear used for case C.
 Two welded steel housings

 Linear electrode arrangement
 Arc ignited in volume
 Exhaust into volume
Top view arc and exhaust volumes
Figure 2-8: Experimental arrangement used for case B.
Page 25
 Cast aluminum housings

 Single phase circular Al bar
 Coaxial electrode arrangement
 Arc ignited at spacer
Cross-section of encapsulation
Figure 2-9: Arrangement of single-phase HV components used for case H.
2.5.3 MV switchgear with air insulation
Case No. A B C D
Volume of arc comp. ( ) 0.509 0.509 0.648 0.27 m³

Volume of exhaust comp. ( ) >1000 1.275 >1000 0.58 m³
Volume of installation room ( ) n/a >1000 n/a >1000 m³
Initial pressure in 150 160 100 120 kPa abs air
Initial pressure in 100 100 100 100 kPa abs air
Area of the relief opening 0.00456 0.00456 0.0763 0.049 m²
Discharge coefficient of 0.7 1.0 0.7 1.0
Response pressure of relief device 276 285 35,3 220 kPa rel
Area of the opening 0 0.010 0 0.195 m²
Short-circuit current 14.5 14.5 14.5 38.8 kA rms
Number of phases 1 1 2 3
Averaged phase-to-ground voltage 314 424 400 250 V
-factor 0.4 0.55 0.7 0.6
Table 2-3: Input parameters and initial values for MV switchgear cases with air insulation.
Page 26
Case A 22
0.5
0.45
P1 calculated
0.4
P1 measured
0.35
pressure [MPa]
0.3
0.25
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
time [s]
Figure 2-10: Case A – Measured and calculated pressure development in V1 in air.
Case B 27
0.5
0.45 P1 calculated
0.4 P2 calculated
P1 measured
0.35
P2 measured
pressure [MPa]
0.3
0.25
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
time [s]
Figure 2-11: Case B – Calculated pressure developments in V1 and V2 in air and comparison with test.
Page 27
Case C 70
0.15
0.14 P1 calculated
P1 measured
0.13
pressure [MPa]
0.12
0.11
0.1
0.09
0.08
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
time [s]
Figure 2-12: Case C – Calculated pressure development in V1 in air and comparison with test.
Case D 14
0.4
0.35 P1 calculated
P2 calculated
0.3 P1 measured
P2 measured
0.25
pressure [MPa]
0.2
0.15
0.1
0.05
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
time [s]
Figure 2-13: Case D – Calculated pressure developments in V1 and V2 with air as filling gas and
comparison with test.
Page 28
For air as the filling gas in the compartment, calculation with the basic model and measured results show good
agreement (Figure 2-10 to Figure 2-13). Peak pressure and drop of pressure in the arc compartment show good
coincidence. The -factor is taken between 0.4 and 0.7, which is in accordance with published data. The
calculation of the pressure development in the exhaust compartment shows less satisfying agreement with the test
results (Figure 2-11 to Figure 2-13). The reason might be and gas data not depending on temperature and
pressure. Also, this could be a matter of the position of the pressure sensor during the particular test when the
exhaust compartment is much longer than wide (like a channel). In this case, the pressure measured at the end of
the channel will show a delay depending on the length of the channel and the gas speed. This effect could explain
the discrepancy in case D (Figure 2-13). For case C (Figure 2-12), all 14 volumes between the arc ignition point at
the cable terminals and the pressure relief device in the vertical exhaust channel at the side of the switchgear
arrangement were combined into one volume in order to achieve the best agreement. This measure is
recommended if there are small (intermediate) volumes with large openings between volumes.
The agreement between experiment and calculation is better than 10 % due to the adaption of calculation
parameters ( , response pressure, arc voltage, and discharge factor). Overall, the agreement between experiment
and calculation is better than 10 % considering that the arc voltage has been provided from the tests and the -
factor as well as the discharge factor have been adapted to give the best results. This result is supported by the
range of deviations of the peak pressure shown in Section 2.5.6 for all test cases.
2.5.4 MV switchgear filled with SF 6 gas
Case No. E F G
Volume of arc comp. ( ) 0.509 1.217 0.27 m³
Volume of exhaust comp. ( ) >1000 >1000 0.58 m³
Volume of installation room ( ) NA NA >1000 m³
Initial pressure in 150 166 120 kPa abs SF6
Initial pressure in 100 100 100 kPa abs air
Area of the relief opening 0.00456 0.062 0.049 m²
Discharge coefficient of 1.0 1.0 1.0
Response pressure of relief device 310 1400 220 kPa rel
Area of the opening NA NA 0.195 m²
Short-circuit current 14.2 25 38 kA rms
Number of phases 1 3 3
Averaged phase-to-ground voltage 350 1700 400 V
-factor 0.75 0.7 0.76
Table 2-4: Input parameters and initial values for MV switchgear cases with SF6 insulation.
Page 29
Case E 24
0.5
0.45
P1 calculated
0.4
P1 measured
0.35
pressure [MPa]
0.3
0.25
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
time [s]
Figure 2-14: Case E – Measured and calculated pressure developments in V1 with SF6 as filling gas.
Case F 03
1.8
1.6
P1 calculated
1.4
P1 measured
1.2
pressure [MPa]
0.8
0.6
0.4
0.2
0
0 0.1 0.2 0.3 0.4 0.5 0.6
time [s]
Figure 2-15: Case F – Measured and calculated pressure developments in V1 with SF6 as filling gas.
Page 30
Case G 13
0.4
0.35 P1 calculated
P2 calculated
0.3 P1 measured
P2 measured
0.25
pressure [MPa]
0.2
0.15
0.1
0.05
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
time [s]
Figure 2-16: Case G – Measured and calculated pressure developments in V1 with SF6 as filling gas and
comparison with test.
For the filling gas SF6, the reproduction of measurement and simulation with the basic model is quite reasonable for
cases E (Figure 2-14) and F (Figure 2-15), but unsatisfying for case G (Figure 2-16). The latter shows the correct
peak pressure, but also a much longer residence time of the gas in the arc compartment. The -factor is taken 0.7
to 0.76 consistently for all cases, which is within the range of published data. The best fit of the decaying part of the
pressure curves was obtained with discharge coefficients of the exhaust openings of 1.0, whereas for air as filling
gas the anticipated factor of 0.7 was prevailing. For SF6, even better agreement would be obtained for discharge
coefficients larger than 1. In particular in case G, the calculated residence time of SF6 in the arc compartment is
much longer than measured. Better agreement with test results would be achieved by increasing the gas outflow by
a factor of 2, which is equivalent to a discharge coefficient of 2. Since for air (see case D), the outflow is simulated
correctly, this cannot be attributed to a falsely determined outflow area, but rather seems to be a peculiarity of SF6.
Since the gas temperature in this case exceeds 2000 K, SF 6 is already dissociated to a large amount suggesting
that decomposition products are responsible for the higher flow velocity. For air, this effect might not be visible,
since decomposition starts at higher temperatures. Other effects like the observed reduction of the phase-to-
ground voltage after opening of the relief device or the reduction of the -factor as discussed in Section A.4.1
might also result in a shorter residence time, however, should be effective for both air and SF6.
Summarizing, the fit between experiment and calculation for SF6 is not as good as for air in particular with respect
to the decay of pressure after opening of the relief device.
Page 31
2.5.5 HV switchgear filled with SF 6 gas
Case No. H
Volume of arc comp. ( ) 0.39 m³
Volume of exhaust comp. ( ) >1000 m³
Volume of installation room ( ) NA m³
Initial pressure in 490 kPa abs SF6
Initial pressure in 100 kPa abs air
Area of the relief opening 0.006 m²
Discharge coefficient of 1.0
Response pressure of relief device 1000 kPa rel
Area of the opening NA m²
Short-circuit current 65 kA rms
Number of phases 1
Averaged phase-to-ground voltage 450 V
-factor 1.6
Table 2-5: Input parameters and initial values for HV switchgear cases with SF6 insulation.
Case 01
H
3
P1 calculated
2.5
P1 measured
2
pressure [MPa]
1.5
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6
time [s]
Figure 2-17: Case H - Measured and calculated pressure development in V1 with SF6.
For the HV single phase equipment (Figure 2-17), a good agreement between simulation and measurement can be
achieved adapting to a high value of 1.6. This is most probably due to excessive evaporation of electrode and
housing material in particular aluminum. The associated reaction is exothermic and increases the amount of energy
available for gas heating. This case is better reproduced by an enhanced model, which takes electrode evaporation
Page 32
into account. The discharge coefficient of the exhaust opening is again taken as 1.0 and the exhaust of gas is
simulated correctly though the gas temperature also exceeds 2000 K.
2.5.6 Evaluation of key parameters for all test cases
For all test cases, the peak pressure in volume was calculated and compared with the result from tests.
The values are shown in Figure 2-18 irrespective of any differences of the tests with respect to geometry, electrical
parameters and kind of filling gas. The deviation with respect to the peak pressure in the arc compartment is
around +/-10% with some rare extensions to 15 or 20%. However, one has to keep in mind that this good
agreement has been obtained with the use of the measured arc power i.e. arc current and voltage and a careful
adaptation of the -factor. In general the deviation between predicted and actual test results will be higher, since
an adaptation of the arc energy input can only be done after the tests.
Figure 2-18: Deviation between measured and calculated peak pressures in V1 irrespective of geometrical and
electrical parameters and kind of filling gas.
2.6 Enhancements of the basic model

Enhanced models are based on the same gas equations as are used for the basic model. However, some
simplifications introduced in the basic model are avoided and some parameters may obtain dependencies deduced
from theory or experiments. The modifications allow a better adaption of measured data to the calculated data, but
also introduce a higher degree of freedom with the uncertainty that the included effect might be more or less
relevant in the general case. These modifications in general do not increase the accuracy in predicting the peak
pressure in the arc compartment, but do increase the accuracy in simulating the time dependence of the gas
outflow and pressure development in the exhaust compartment. Details of possible modifications are provided in
ANNEX A.4.
Page 33
2.7 Summary
The attempt to re-calculate pressure curves measured inside the arc compartment during an internal arc fault gives
good agreement between test and simulation as long as the input arc energy is known and the -factor and the
discharge coefficient are suitably chosen. Pressure peaks and decay can be simulated within a deviation of 10 %.
The comparison also indicates that for most arrangements common input parameters can be used such as a -
factor of 0.5 for air and 0.7 for SF6, a relief area discharge coefficient of 0.7 and an approximated arc voltage
determined from phase distance. Clearly one has to expect that the deviation of predictive calculations is larger
than 10%. Enhanced models with adapted input parameters, temperature-dependent gas properties or the
consideration of gas mixtures – also including evaporation of electrode material – are appropriate to improve the
agreement between simulation and test results.
REFERENCES:
[AiF2011]: Final Report of a German research project, AiF-Projekt 15657N, “Reduzierung der Druckbeanspruchung
elektrischer Anlagen im Störlichtbogenfall” (“Reduction of the pressure stress in electrical installations due to
internal arcs“), Project leader: G. Pietsch, 2011.
[Anantavanich2008]: K. Anantavanich, G. J. Pietsch, “Inclusion of SF6-air mixtures in the calculation of pressure

rise due to internal arcing in switchgear installations”, Proc. XVIIth Int. Conf. on Gas Discharges and their Appl.,
Cardiff, UK, pp. 165-168, 2008.
[Dubbel1997]: Editors W.Beitz, K.H. Grote, “Taschenbuch für den Maschinenbau“ (“Pocket-book for mechanical
engineering”), Springer Verlag, 19th edition, 1997, section B (Mechanik) 6.2.
[Dullni1994]: E. Dullni, M. Schumacher, G. Pietsch, “Pressure rise in a switchroom due to internal arc in a
switchboard”, Proc. 6th Int. Symp. on Short-Circuit Currents in Power Systems, pp. 4.5.1 - 4.5.7, Liège, Belgium,
1994.
[Friberg1995]: G. Friberg, G. Pietsch u. M. Schumacher, “On the description of pressure rise in the surroundings of
high current arcs in metal enclosed compartments with pressure relief“, Proc. 11th Int. Conf. on Gas Discharges
and their Applications, Tokyo, Japan, Vol. I pp. 18-21, 1995.
[König1984]: D. König, Th. Facklam, „Pressure rise in metal enclosed SF6-insulated HV switchgear of single-phase
enclosure type due to internal arc”, Electra no. 93, pp. 25-52, 1984.
[Mende1975]: D. Mende, G. Simon, „Physik Gleichungen und Tabellen“ (“Physics equations and tables“), VEB
Fachbuchverlag Leipzig, 1975.
[Schmidt1958]: E. Schmidt, „Technische Thermodynamik” (“Technical Thermodynamics”), 7th edition in German,

published by Springer, 1958.
[Solvay]: „Schwefelhexafluorid“ (“Sulphur hexafluoride”), Company Brochure SOLVAY GmbH.
[Welich1984]: D. Welich, „Zur Ausbildung mehrpoliger Störlichtbögen und zur Innendruckentwicklung von
luftisolierten, metallgekapselten Mittelspannungsschaltzellen“ (”On the formation of multi-phase internal arcs and
Page 34
pressure rise in air-insulated, metal enclosed MV switchgear“), PhD thesis in German, Technical University
Dresden, 1984.
[Zhang2002]: X. Zhang, G. Pietsch, E. Gockenbach, „Investigation of the thermal transfer coefficient by the energy
balance of fault arcs in electrical installations”, IEEE trans. On power delivery, 2006.
Page 35
3 SENSITIVITY ANALYSIS OF THE BASIC MODEL

3.1 In troduction
Chapter 2 and ANNEX A describe the equations of the model for the pressure rise calculations and the algorithm
for numerical integration. The sensitivity studies use this algorithm for numerical evaluation of test cases collected
by WG A3.24. Each test case provides a set of parameters called reference values. The parameters are varied
continuously around a reference value and a series of pressure curves are provided as a result (ANNEX F Section
F.2).
In order to allow for a better understanding of the relations between the parameters and their effects on the results,
some simplified equations are provided assuming constant arc power (Sections 3.2 and 3.3). The impact of the
selected parameters and the effects of replacing SF6 with air are described in Section 3.4.
The sensitivity analysis is limited to the basic model (Section 2.2).
3.2 Model para meters

The input parameters are defined in Chapter 2. This chapter uses the same definitions and notations, with a few
exceptions which are explicitly pointed out in the text. The subscripts 1, 2 and 3 below refer to arc compartment,
exhaust compartment and installation room / environment respectively.
For the sensitivity analysis the parameters are regrouped in the following way:
 energy input: effective short circuit current - , angle of fault initiation – ; d.c. time constant – ; mean
arc voltage - ; fraction of arc energy producing the pressure rise - , and , ; number of phases -
(1 or 3); duration of short circuit current - ; the two last symbols are not introduced explicitly in
Chapter 2
 switchgear design and initial conditions: arc and exhaust compartment volumes - and ; arc and
exhaust compartment relief opening areas - and , discharge coefficient for arc and exhaust
compartment relief openings - and ; arc compartment relative response pressures - , ; arc and
exhaust compartment initial pressures and initial temperatures - , , , , , , , ; the symbols for
initial values are not explicitly introduced in Chapter 2, in ANNEX A they are noted as ( = 0)
 gas type: heat capacity ratios - and ; specific gas constants - , and ,
The subscripts related to gas type (SF6 and Air) may be avoided in some equations.
3.2.1 Energy input
The energy ( ) results from the integration of equation (2-20) in Section 2.3.5. For symmetrical current ( =90°)
and using (2-5), the part of energy absorbed by the gas ( ) within a number of entire periods of power frequency
can be expressed by:
2√2
( )= (3-1)
Page 36
where = , is the number of cycles, is the frequency (50 or 60 Hz). Thus the parameters , and
√
have exactly the same effect on the provided energy. Let ̇ = . We can consider this
expression as the average input power (heat source). The unit of ̇ is W.
Equation (3-1) does not apply for asymmetrical current. However, the asymmetry parameters and have limited
influence on the pressure curve (Section 3.4.1).
3.2.2 Switchgear design and initial conditions
The parameters , , , , , , , , , , , and , , , define the main mechanical characteristics of

the switchgear and the conditions at the beginning of the simulation.
The absolute (measured) response pressure of the arc compartment relief device is the sum of the relative bursting
pressure and the absolute initial pressure in the exhaust compartment: = , + , . In the simulation,
the time when the absolute pressure in the arc compartment reaches this value is considered as the instant of the
relief device opening.
The initial gas density in the arc compartment equals = , / , . The density depends on gas type: it
is more than five times higher for SF6 than for air if the pressure and the temperature are the same.
3.2.3 Gas type
There are only two independent parameters in the model strictly related to the gas type: the heat capacity ratio κ
and the specific gas constant . The two remaining parameters related to gas type i.e. and are also related to
the switchgear. The factor is discussed in Section 3.2.1 as part of the energy input equation. The sensitivity
study is not relevant for the coefficient because its sole effect is to reduce the value of the opening area
(equation (A-22), ANNEX A).
Changing the gas type requires a change of both and . In a proper simulation of pressure rise, one should also
adjust and (Section 2.5). The arc voltage would change too. However, there is no common rule for
modifications of these parameters. For this reason, in the sensitivity studies in this chapter the change of gas
involves only and ; all remaining parameters are retained unchanged.
3.3 Model results

The most important output of the model is the pressure curve in the arc compartment and the exhaust
compartment during the process. The simulation finishes in two ways: it either stops normally when the simulation
time elapses or else it is aborted when the gas temperature reaches 20000 K.
A typical arc compartment pressure curve has the following intervals (Figure 2-3):
 Rapid pressure rise ( ) until the opening of the pressure relief device ( ).
 Further pressure increase until the peak pressure is reached ( , , , ), called the overshoot period.
For some conditions, it is possible that the pressure decreases immediately after the opening; in this case
there is no overshoot. The difference between maximum pressure and absolute response pressure is
called the overshoot level (see Section 3.3.2).
Page 37
 Pressure decrease usually with two identifiable periods: before and after the end of arc.
The exhaust compartment pressure curve has two intervals: pressure increase and pressure decrease after the
maximum pressure is reached ( , , , ).
Essentially four consecutive periods in the pressure development can be distinguished, which differ in the energy
input and/or flow conditions:
 before the bursting of the arc compartment relief device: arc energy is provided to the arc compartment, no
mass flows
 after the bursting of the relief device: energy is still provided and a mass flows from the arc compartment to
the exhaust compartment
 after the bursting of the exhaust pressure relief device: mass flows between all compartments
 after arc extinction ( ) – there is no more energy provided to the arc compartment but mass flow
continues
3.3.1 Before bursting of the pressure relief device
Before the bursting of the relief device and assuming constant gas parameters as well as constant energy input
rate ( ) ≈ ̇ ∙ , the pressure curve in the arc compartment is a linear function of time:
−1 −1
( )= , + () = , + ̇∙ (3-2)
where is the gas mass in and is the heat capacity at constant volume (ANNEX A).
Figure 3-1 compares the pressure curve calculated with equation (3-2) to the basic model. The slight difference at
the beginning of pressure rise is due to the current asymmetry neglected by equation (3-2).
The approximated values of and can be calculated with the following equations:
̇
≈ ( − 1) (3-3)
− ,
≈ (3-4)
̇ ( − 1)
By “fast process” we describe the process with , calculated with equation (3-4), to occur within one period of
power frequency. Usually the fast processes happen in small volumes. Equation (3-2) does not apply for fast
processes. In this case the pressure increase rate from basic model simulation is generally higher than given by
equation (3-3).
For slow processes, the pressure is built up over several periods and then the power input rate can be assumed
constant. Therefore equation (3-2) is a helpful approximation of the pressure rise for slow processes.
Page 38
0.49
0.44
0.39 3
2.8
0.34
[MPa]
2.6
0.29 2.4
2.2
0.24
2
0.19 1.8
1.6
0.14
1.4
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
0.07 [s] 0.12 0.17 0.22
Figure 3-1: Comparison of the pressure rise before bursting of the pressure relief device evaluated with
equation (3-2) (pink line) and with the basic model (blue line).
3.3.2 After pressure relief bursting with a large or infinite exhaust compartment
If the exhaust compartment is infinite compared to the arc compartment, it will not affect the processes in the arc
compartment. In this case for slow processes, the pressure curve in the arc compartment after the burst can have
one of two forms: with overshoot or without overshoot. Without the overshoot, the pressure in the arc compartment
drops immediately after the opening of the relief device and , = .
It is shown in ANNEX F.3 that for slow processes the overshoot condition can be directly derived from the input
parameters:
̇
> , (3-5)
,
/ /
where = 40.42 and = 55.27 . The constants depend only on and (see ANNEX
F.3 for more details). If the condition (3-5) above holds the overshoot occurs, otherwise the pressure in the arc
compartment drops immediately after the opening of the relief device. Moreover, the overshoot level is greater
when the ratio ( ) between the left and right sides of (3-5) is higher. The expression (3-5) can be simplified
considering that the initial temperature equals the ambient temperature ( , = 293 K).
The influence of on the overshoot condition of the pressure curve is shown in Figure 3-2.
Page 39
0.5
0.45
289V, r=1
0.4 157V, r=0.543
251V, r=0.868
0.35
392V, r=1.356
[MPa]
487V, r=1.685
0.3
0.25
0.2
0.15
0.1
0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95
[s]
Figure 3-2: Typical air pressure curves for different arc voltages [V] and corresponding values of the ratio
( ) between the left and right side of (3-5): with overshoot (pink & blue, > ), intermediate (dark-green,
= ) and without overshoot (orange & red, < ).
3.3.3 After bursting of the relief device with a limited size exhaust compartment
If the pressure in the exhaust compartment is greater than approximately half of the pressure in the arc
compartment, the flow of gas out of the arc compartment will be slowed down and the pressure curve will be
altered. More exactly, in Section 2.2.2 a gas dependent “critical ratio” is deduced from equation (2-10), relating the
gas pressure in the exhaust compartment ( ) to that in the arc compartment ( ) for which the latter is not
influenced by the former. The reciprocal value of the critical ratio is denoted here by .
≤ (3-6)
where the constant depends only on : = 0.586 and = 0.528.
At the instant of the pressure relief device bursting = , + , and = , , so the following equation
results
1
, ≥ −1 , (3-7)
In Figure 3-3 the pressure in both compartments is shown. The blue line is the pressure curve calculated with an
infinite volume exhaust compartment. The moment where the increasing pressure in the exhaust compartment
starts influencing the pressure in the arc compartment can be clearly observed (a zoom shows that the separation
starts exactly at the marked point).
Page 40
0.5 simulation for air

0.45
0.42 MPa
0.4
0.35
0.3
[MPa]
0.25
0.22 MPa
0.2
0.15 arc compartement with infinit exhaust volume

arc compartement with exhaust volume of 1.275m³
0.1 exhaust compartement
critical ratio of 0.528
0.05
0
0 0.09 0.18 0.27 0.36 0.45 0.54 0.63 0.72 0.81 0.9 0.99
[s]
Figure 3-3: Pressure curve for arc and exhaust compartments: Effect of the exhaust compartment
increasing pressure on the arc compartment pressure decay.
A simple rule of thumb results from equation (3-7) to determine when an effect of the exhaust compartment on
pressure in the arc compartment can be neglected:
A. the relative bursting pressure of the relief device of the arc compartment is greater than the following
fraction of the initial pressure in the exhaust compartment:
, > 0.706 , for SF6 and
, > 0.893 , for air

B. the pressure in the arc compartment is sufficiently high to ensure that the condition (3-6) holds
The second condition should be verified by simulation.
Figure 3-4 shows the case where the condition A above holds ( , = 0.27 MPa vs , = 0.1 MPa) but the
condition B does not. The pressure in the exhaust compartment (green line) increases relatively fast and affects the
pressure in the arc compartment (pink line) before the maximum is reached.
Page 41
0.45 simulation for SF6

0.41 MPa
0.4
0.35
0.3
[MPa]
0.25 0.24 MPa
0.2
0.15
arc compartement with infinit exhaust volume
arc compartement with exhaust volume of 1.275m³
exhaust compartement
0.1 critical ratio of 0.586
0.05
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
[s]
Figure 3-4: Pressure curves for arc and exhaust compartments: Effect of exhaust compartment increasing
pressure on arc compartment peak pressure.
3.4 Sensitivity analysis

The complete sensitivity analysis would consist in the evaluation of all results for each combination of relevant
parameters but this is not feasible because of the huge amount of output data. In this brochure the analysis is
limited to the evaluation of arc compartment results: ( , ) and ( , ).
The sensitivity is expressed by the variation related to the reference value. For ( , ) the sensitivity is also
expressed by equations (3-3) and (3-4).
3.4.1 Sensitivity to power input
The power provided to the system has a great influence on the pressure curve for both the arc and the exhaust
compartments. From equation (3-1), the parameters , , are equivalent i.e. varying any of them has the
same effect on the pressure curve, e.g. Figure F-1, Figure F-2, and Figure F-3 in ANNEX F.
General observations
Increasing power:
 accelerates the process before bursting of the pressure relief device; for slow processes the pressure slope
is proportional to the power input and the time to burst is inversely proportional to the power
input
 increases the maximum pressure in arc compartment if overshoot is present. The peak pressure growth is
generally higher for air than for SF6. In the absence of an overshoot condition the peak pressure remains
the same and equals the bursting pressure (Figure F-4 in ANNEX F).
 increases the maximum pressure in the exhaust compartment
 favors the overshoot condition (criterion (3-5))
Page 42
Asymmetry of current
The shape of the arc current has a limited influence on the pressure curve and has an effect only for fast
processes. The variation of peak pressure as a function of fault initiation angle is small (Figure F-5 in ANNEX F).
The influence of the d.c. time constant might be more substantial (Figure F-6 in ANNEX F).
Current duration
If the arc current ceases before the pressure relief device bursts ( < ) or if it continues after bursting but
ceases when the pressure is still increasing ( < < ) then the maximum pressure equals the
instantaneous pressure at this moment (Figure F-7 in ANNEX F). Shorter current durations may relieve the impact
of pressure rise for SF6 switchgear, whereas for air current duration is hardly relevant, since all effects happen
rapidly after arc initiation.
3.4.2 Sensitivity to volume
The peak pressure does not depend on the arc compartment volume as long as the process can be considered as
a slow process (see definition in Section 3.3.1) and the exhaust compartment is relatively large. Moreover, the time
to reach the peak pressure is proportional to the volume (Figure 3-5). This means that increasing the volume
will slow down the process but will not change the type of process: the overshoot level and the maximum pressure
will remain the same.
0.506 m³
0.253 m³
0.6 0.329 m³
0.405 m³
0.55
0.481 m³
0.5 0.557 m³
0.45 0.633 m³
0.708 m³
0.4 0.784 m³
[MPa]
0.860 m³
0.35
0.936 m³
0.3 1.012 m³
0.25 burst 0.42 MPa
0.2 0.5
0.4
0.15
0.3
[s]
0.1 0.2
0.08 0.18 0.28 0.38 0.48 0.58 0.68 0.78 0.88 time peak
0.98 pressure
0.1
[s] 0
0.2 0.4 0.6 0.8 1
[m³]
Figure 3-5: Pressure development over time with varying volume of arc compartment [m3] (simulation for
air).
On the other hand decreasing the volume will accelerate the process and for smaller volumes the peak pressure
may increase notably (the process can no longer be considered as slow), especially with respect to an
asymmetrical current (Figure F-8 in ANNEX F).
Page 43
3.4.3 Sensitivity to opening area and bursting pressure
Arc compartment opening area and bursting pressure have an important influence on the pressure curve.
It follows from criterion (3-5) that increasing or reduces the overshoot level.
The opening area of arc compartment is one of the principal parameters controlling the gas outflow and thus
the maximum pressure. Under overshoot conditions, the maximum pressure is higher for smaller opening areas. In
Figure F-9 in ANNEX F for example, 50 % of the reduction of opening area yields a 43 % increment of maximum
pressure for air and 32 % for SF6. The time to reach the peak pressure increases even more significantly: 121 %
for air and 143 % for SF6.
The response pressure of the relief device has an obvious impact on the pressure curve: the lower the bursting
pressure, the lower the maximum pressure. For the case shown in Figure F-10 in ANNEX F, the maximum
pressure reduction is 6 % for air and 15 % for SF6 for 50 % of bursting pressure reduction. It should be noted that
increasing the bursting pressure beyond a certain level, for which the overshoot is absent, leads to the trivial case
where the maximum pressure equals the absolute bursting pressure.
3.4.4 Sensitivity to exhaust volume
As explained in Section 3.3.3, the exhaust volume influences the pressure curve in the arc compartment. Figure
3-6 shows that the maximum pressure may also be affected. There is no overshoot in this case so the pressure
drops directly after the burst and increases again about a hundred of milliseconds afterwards. The arc must last a
sufficiently long time in order to build up the gas pressure in the exhaust compartment, so that it starts affecting the
pressure in the arc compartment.
It is difficult to formulate more precise rules to describe the mutual interaction between the pressures in arc and
exhaust compartments. It should be noted that the influence of the latter is more important for SF6 than for air.
1.275
0.4
0.638
0.35
0.828
0.3 1.02
[MPa]
1.211
0.25
1.403
0.2
1.594
0.15 1.785
1.976
0.1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
2.168
[s]
Figure 3-6: Pressure curves in arc compartment and in exhaust compartment (dashed lines) influenced by
changing exhaust volume [m3], (SF6, = 0.008 m2, = 2 s, = 0.509 m3).
Page 44
3.4.5 Sensitivity to initial pressure and temperature
The impact of the initial pressure on the pressure curve in an arc volume is depicted in Figure 3-7: with increasing
initial pressure the peak pressure increases except for those conditions which have no overshoot (orange and
yellow curves). The burst time decreases in accordance with equation (3-4) assuming that the response pressure
of the relief device is kept the same.
Figure 3-8 shows how the pressure curve in an arc volume depends on the initial gas temperature. There is no
impact before the pressure relief device bursts in accordance with equations (3-3) and (3-4). For higher
temperatures there is no overshoot. If the initial temperature decreases, the maximum pressure and the time to
peak increase slightly. Considering that the initial temperature varies only within the ambient temperatures
permitted by the standard, the initial temperature has a limited effect on the pressure curve.
It is appropriate to study the initial gas pressure and temperature together as two factors defining the initial gas
density in the arc volume. In Figure 3-9 the same simulation as above is presented but with the pressure adjusted
in order to maintain the same gas density. The pressure curves match except at the beginning. It can be concluded
that the initial pressure and temperature can be combined in one parameter namely initial gas density. In real
situations, the gas density is constant for sealed compartments and pressure varies as a function of temperature.
In simulations, the same principle should be applied: fix the gas density and adjust the pressure as function of
temperature.
0.45
0.4
0.35
1.5
0.3 9.75
1.2
[MPa]
0.25 1.425
1.65
1.870
0.2
2.1
2.325
0.15 2.55
2.775
0.1 3
0.05
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
[s]
Figure 3-7: Pressure development for different initial gas pressures (air).
Page 45
0.43
0.38
0.33
[MPa]
0.28
20 °C
-38 °C
5 °C
0.23
50 °C
0.18
0.13
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
[s]
Figure 3-8: Pressure development for different initial gas temperatures (air).
0.43
[-38 °C, 0.12 MPa]

0.38
[ -5 °C, 0.14 MPa]]
[ 20 °C, 0.15 MPa]]
0.33 [ 50 °C, 0.16 MPa]]
0.28
[MPa]
0.23
0.18
0.13
0.08
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
[s]
Figure 3-9: Pressure development for different initial pressures and temperatures combined as gas
densities (air).
3.4.6 SF 6 vs air
Replacing SF6 by air modifies the pressure curve of the arc compartment in many aspects.
a. The rate of pressure rise depends on the heat capacity ratio: for air is more than four times
higher than for SF6 (precisely 4.27); thus is shorter for air than for SF6.
b. After the bursting of the pressure relief device, the process is slower for SF6: time to peak for air is about
55 % of the time to peak for SF6 under overshoot conditions.
Page 46
c. The overshoot condition can change: it is possible that overshoot is present for air and is absent for SF6
(e.g. Figure F-11 and Figure F-12 in ANNEX F). In effect, ≈ 0.7346· (see criterion (3-5)). For any
SF6 case with the ratio greater than 0.7346 and smaller than 1 (0.7346 < < 1), replacing SF6 by air will
turn on the overshoot condition. If overshoot is present for SF6 ( > 1) then it will be higher for air. If <
0.7346 the overshoot is absent for both air and SF6.
d. The peak pressure is lower by about 25 % on average for SF6 than for air as long as the overshoot is
present. If there is no overshoot in the arc compartment, the maximum pressure remains unchanged and
equals the absolute response pressure ( ) for both gases.
e. It is important to consider the impact of the arc duration and the simulation time when comparing the
pressure curve for air and SF6.
3.4.7 Impact of input parameters (summary)
The sensitivity analysis also allows the evaluation of the impact on results caused by the lack of precise test data
and tolerances of input data, including the switchgear geometry. Table 3-1 illustrates the worst errors on peak
pressure for all test cases when the input parameters are varied by ± 10 % around the reference value giving a
bandwidth of 20 % variation. For the most sensitive parameters, the impact on maximum pressure is no higher than
the variation of input value. For slow processes this impact is lower.
Peak pressure variation

Parameter
Slow process Fast process
20 % of variation (± 10 %) Air SF6 Air SF6
Arc power ( ̇ ) 14 % 13 % 18 % 19 %
Arc compartment volume 7% 3% 19 % 19 %
Arc compartment relief opening 13 % 12 % 15 % 12 %
Arc compartment response
19 % 19 % 20 % 10 %
pressure
Arc compartment initial
6% 6% 8% 6%
temperature
Arc compartment initial pressure 6% 6% 8% 15 %
Exhaust compartment relief
0% 4% 10 % 11 %
opening
Exhaust compartment
0% 1% 2% 1%
temperature
Table 3-1: Variation of peak pressure when input parameters are varied (worst of test cases).
Page 47
3.5 Summary
The model has many parameters. For the sensitivity study some of them can be combined into composite
parameters: the energy input rate (or average power) includes , , and , the initial gas density
includes , and , and the absolute bursting pressure includes , and , .
The most sensitive parameters are:
 energy input rate (combining arc voltage, arc current and thermal transfer coefficient ),
 burst characteristics (including bursting pressure and opening area)
 gas type
The parameters with limited impact are:
 asymmetry of current (d.c time constant and angle)
 initial conditions (density of gas)
There are two types of processes: slow and fast. A process is considered fast if the pressure relief device bursts
within one period of the power frequency. The impact and importance of the input parameters differ for slow and
fast processes.
For slow processes, the arc compartment volume has practically no impact on the maximum pressure in the arc
compartment. However an increase in arc compartment volume slows down the process before and after the
opening of the relief device. An easy method is provided in 3.3.2 for the assessment of the overshoot condition for
slow processes. It should be evaluated first because of its impact on the pressure curve: without overshoot the
case is trivial because the maximum pressure equals the absolute bursting pressure.
Under certain conditions, the exhaust compartment may also have an impact on the maximum pressure in the arc
compartment, especially for SF6. This impact should be carefully evaluated while comparing the pressure curve
between SF6 and air. A helpful rule of thumb is provided in Section 3.3.3.
Page 48
4 CALCULATION OF PRESSURE WITH CFD SOFTWARE

4.1 In troduction
While the basic model described in Chapter 2 delivers spatially averaged pressure values (one pressure value per
compartment), Computational Fluid Dynamics (CFD) software provides spatially distributed pressure values – it
calculates pressure at each point of the meshed compartment volume. For this purpose, the basic equations of
fluid dynamics, the conservation laws of mass, momentum and energy are solved three-dimensionally together with
an enthalpy and a state equation.
CFD has been used for a long time in arc modelling, especially in the development of high and low voltage
interrupter chambers. It is nowadays also used for the calculation of the flow distribution of hot gases expelled from
switchgear experiencing an internal arc, especially for MV switchgear.
4.2 Motivations for CFD

CFD allows the following to be considered:
 Space-dependency of an arc model (energy input, vaporization..)

 Dynamic processes like reflection, diffraction and interference of gas flows (wave theory)
 Compressible gas flow
CFD is typically used in:
 Assessing the actual geometry of the switchgear and installation room (simulating actual electrical
installations when they differ from the manufacturer’s requirements or from the test conditions).
 Analysing the influence of the location of pressure relief openings in rooms.
 Analysing the influence of specific flap design, the influence of grids and absorbers.
In all these cases the expected short-circuit conditions of the actual power system are usually considered instead
of the rated short-circuit values.
For example: A switchgear tested with a straight pressure relief duct on the top of the installation must be used with
a rectangular duct (Figure 4-1). It is expected that the duct angle creates a jump in the pressure, therefore creating
pressure increases upstream in the switchgear. CFD is used to evaluate the extent to which this happens.
Note: A methodology for comparison between an actual and a reference switchgear is defined in Section 7.3.
CFD requires a dedicated workflow, which is described in Section 4.3.
Page 49
(a) (b)
Figure 4-1: Example for two different arrangements of a gas exhaust duct. (a) Test condition. (b) Actual
installation.
4.3 Workflow in CFD

Overview of the workflow is shown in Figure 4-2. Programming interfaces provide the possibility to introduce gas
and arc models. A simplified approach can be chosen, or enhanced ones, according to the objectives of the
simulation.
PRE-PROCESSOR
Geometric description and meshing of
the fluid domain
USER-INTERFACE
Specification of boundary conditions
PROGRAMMING
and calculation settings PROGRAMMING
INTERFACE
INTERFACE
Gas Data
Arc model
EQUATION SOLVER
Transient solution of equation system, which is
solved by iteration.
PROGRAMMING POST-PROCESSOR PROGRAMMING

INTERFACE Presentation of results e.g. spatial
INTERFACE
e.g. Absorber resolved pressure developments Further effects
Figure 4-2: Workflow for CFD.
Page 50
4.3.1 Mesh the fluid domain: pre-processor activity
The first step in the process is to define the geometry of the fluid domain. The gas (air or SF 6 or mixture) volume is
called “the fluid domain”. It is usually modeled in a CAD package. Boundaries are the metal sheets of the
switchgear enclosure or equipment, and the building walls, roof, ceiling. In the end the model looks like an
installation room drawing, however it is actually the gas volume that is the fluid domain.
(a) (b)
(c) (d)
Figure 4-3: An example of an air-insulated switchgear consisting of two functional units with stacked gas
duct on the top;(a) photograph, (b) sectional view of a functional unit, (c) Three dimensional drawing of
the volumes, which have to be meshed and (d) discrete meshes of the flow domain.
The second step is to mesh this fluid (gas) domain, which consists of splitting it into a number of small volumes
(also called finite elements). Each (stationary) volume will have its own values of gas velocity (vector in 3 space
Page 51
directions x, y, z), gas temperature, gas pressure and density. The mesh is created automatically by appropriate
meshing algorithms.
It must be pointed out that small geometric details of the switchgear enclosure or of the room construction are
usually not needed to derive reasonable pressure results. A simplified geometry is normally defined, which focuses
on main shapes and dimensions only. If, however, obstacles and switchgear elements are anticipated or even
intended to influence the gas flow, they have to be modeled. This will help in achieving a mesh of high quality and
numerical results of sufficient accuracy. If needed, the mesh can be refined in regions of particular interest. It has to
be taken into account that the higher the number of finite volume elements, the longer the computing time.
The mesh size has to correlate with rates of pressure rise: a large mesh size cannot correctly simulate the time and
space distribution of a short pressure peak. This has to be considered for wave reflection, diffraction and
interference. For instance - after the rupture of a disk in a gas insulated switchgear (GIS) or the flow of a pressure
wave in an air insulated switchgear (AIS) room.
There is no need to define a discharge coefficient for pressure relief device opening, as long as it is a free opening
(e.g.: hole in the wall).
When pressure relief is made using a grid, a labyrinth or all types of porous media [Besnard2008], the pressure
drop across this device must be modeled. Formula (4-1) can be implemented in CFD:
1
Δ =− + Δ (4-1)
Κ 2
2 2
Where Δ [Pa = N/m = kg/(ms )] is the pressure jump, Δ [m] is the device thickness, [m/s] the gas velocity,
3 2
[kg/(ms)] the fluid viscosity, [kg/m ] the fluid density, Κ [m ] and [1/m] are parameters modelling the grid or
porous medium.
For high velocity values experienced in case of an internal arc, the first term, proportional to , is low compared to
the second term. Therefore, this first term is commonly neglected in internal arc modelling, and the coefficient is
sufficient to describe the porous media. This coefficient is called the “head loss coefficient”.
Figure 4-4 shows an example of two-dimensional CFD simulation results around pressure relief openings with a
labyrinth. It shows the pressure relief opening device and the simulated pressure drop across it in 2D. The air flows
from right to left; the red color indicates that the pressure is increased upstream. The use of the head loss
coefficient removes the need for a full geometric description of the device, calculating the same pressure drop
using the head loss coefficient as a boundary condition.
Page 52
(a) (b)
Figure 4-4: An Example of a pressure relief opening with a labyrinth. (a) Outline of pressure relief
opening. (b) Pressure distribution.
4.3.2 Arc and gas models in CFD, and other enhancements
The programming interface allows the implementation of user-defined algorithms. In the simplest approach, the arc
is modeled by an energy input, which is homogeneously distributed in the arc compartment volume, using the
thermal transfer coefficient described in Chapter 2.
In a more detailed approach [Besnard2009], the arc heating power is confined to a small number of finite volumes
in the vicinity of the arc initiation point. In order to balance the temperature rise in these finite volumes, a model of
the radiation process is needed. As a result, the temperatures in the vicinity of the arc initiation point reach high
values (11000 K in air typically), whereas the arc compartment still includes cold gas regions, as do the other
compartments. This accounts for the high temperature gradients existing during the internal arc event, leading to
high density gradients.
The most complete approaches, where the arc would be modeled using physical equations describing the arc
roots, the arc plasma column, the effect of electro-magnetic fields on the motion of the arc, the transfer of energy
from the arc plasma to the surrounding gas etc. have never been applied to internal arc to our knowledge.
All enhancements described in ANNEX A.3 and A.4 for the basic model can also be considered in CFD, especially:
 Gas model : temperature dependance, or temperature and pressure dependence; real gas law
 Material evaporation
 Thermal transfer in absorbers
In addition, turbulence and heat transfer can also be considered. Compared to the basic method, CFD allows a
space-dependency to be chosen for all enhancements.
Page 53
4.3.3 Solving the case with a CFD solver
In a CFD solver, the partial differential equation derived from basic equations of fluid dynamics is converted into an
algebraic one, which then can be solved iteratively. There exist several numerical methods for solving the partial
differential equation e.g. the Finite Element Method (FEM), Finite Difference Method (FDM), and Finite Volume
Method (FVM).
There exist several solver types, coupled or non-coupled ones, considering compressible gas flows or not and
implicit or explicit solvers. Each solver has its own settings.
All small elements of the mesh are connected together, and basic equations of fluid dynamics are solved by the
program in every finite element by correctly calculating mass, impulse and energy conservation from one volume to
another. In a three-dimensional transient analysis, results are time- and space-dependent. All conservation
equations are solved in time at discrete time intervals, the results at a given time step being dependent on the
solution at previous ones. The choice of the time step is of major importance in the calculation as it may lead to an
unstable or unreliable solution if it is not chosen appropriately. Some programs rely on a time step stability criterion
which may be set by the user while the program adjusts the time step according to it.
Initial conditions (gas temperature, gas pressure, gas velocities) shall be defined by the user, as well as boundary
conditions, which are appropriate to the solver used.
4.3.4 Examples of CFD results: post-processing activities
Finally the post-processor is used to present simulation results. Pressure, temperature and flow velocity are
available at each point of the flow domain over time. The solver can calculate average gas pressure on surfaces or
in volume automatically, if the respective surface or volume is defined.
Post-processing tools allow animations (videos) of the flow, which provide a visual support of the phenomena to be
created. Example of the distribution of pressure, temperature and gas velocity at 55 ms after the initiation of
internal arc is shown in Figure 4-5, Figure 4-6 and Figure 4-7 respectively.
Page 54
Duct exhaust
Arc location
Figure 4-5: Isovalue of pressure (relative) (Pa) – Case of an AIS switchgear, 55 ms after arc ignition (50
kA).
Figure 4-6: Isovalues of temperature (K) – Case of an AIS switchgear, 55 ms after arc ignition (50 kA).
Page 55
Figure 4-7: Isovalues of velocity (m/s) – Case of an AIS switchgear, 55 ms after arc ignition (50 kA).
4.4 Summary
With CFD software, pressure, temperature and flow velocity distributions over time are available in a three-
dimensional resolution. To use it a certain workflow has to be followed.
The main advantages, compared with the basic model described in Chapter 2 are:
 spatial resolution of the results

 pressure waves are included
 the actual geometry of the switchgear and installation room is considered
 the influence of the location of pressure relief openings can be analyzed.
The CFD method also has the advantage that it can be extended to study in space and time the simultaneous
deformation of the surrounding enclosure and structures, using fluid-structure interaction methods. Such methods
are however reserved to highly skilled personnel.
The accuracy depends on the quality of the models used to describe the internal arc physics, as well as on an
appropriate meshing of the flow domain and a proper discretization of the time domain.
In the end, it is pointed out that the use of CFD for internal arc simulation is complex. It is recommended that a CFD
methodology be calibrated with actual test results, on a sufficient sample of cases. Furthermore, accuracy of results
depends on the appropriate choice of many modelling parameters, and therefore, on the knowledge and
engineering skills of the user.
REFERENCES:
Page 56
[Besnard2008]: C. Besnard, 2008, “Numerical simulation for design of safe electrical equipment in case of internal
arcing fault”, REE. 11, 35-38.
[Besnard2009]: C. Besnard, 2009, “Internal arc simulation in MV/LV substations”, CIRED 2009, paper 0536.
Page 57
5 INTERNAL ARC TESTING

5.1 In troduction
This chapter deals with the verification of personal safety in case of an internal fault arc in switchgear through
testing, as laid down in the IEC standards.
Frequently, internal arc testing is referred to as "arc resistance" or "internal arc withstand testing" and also the
focus in the earlier chapters of this technical brochure is on pressure rise calculation. The aim of internal arc testing
is to verify that the occurrence of an internal arc in a switchgear panel does not lead to direct injury to personnel in
pre-defined accessible areas. The design of the switchgear panel must be such that the hot exhaust gasses and
debris of the internal arc are guided to safe zones. A large section of this chapter is devoted to the different effects
of arcs burning in air and in SF6.
Internal faults inside metal-enclosed switchgear can occur in a number of locations and can cause various physical
phenomena. The arc energy resulting from an arc in any insulating medium within the enclosure will cause an
internal overpressure and local overheating which will result in mechanical and thermal stressing of the equipment.
Moreover, the materials involved may produce hot decomposition products, in the form of gases or vapors, which
may be discharged to the outside of the enclosure, and endanger personnel or the general public.
Relevant tests are defined in the standards IEC 62271-203 (Ed. 2.0, 2011) for GIS, IEC 62271-200 (Ed. 2.0, 2011),
IEC 62271-201 (Ed. 1.0, 2006), and IEEE Std C37.20.7 (2007) for metal/insulation enclosed-switchgear.
room simulation
indicator
Figure 5-1: Indicator racks located at the front and side of an MV panel for internal arc testing.
5.1.1 GIS equipment > 52 kV
Evidence of the internal arc withstand ability of the enclosure against bursting and burn-through shall be
demonstrated by the manufacturer when required by the user. The IEC 62271-203 standard allows this ability to be
demonstrated by test or by calculations based on test results performed on a similar arrangement or by a
combination of both. Procedures and applications are described in [Trinh1992; CIGRE WG 23.03]. Tests must be
Page 58
carried out with the normal insulating gas, usually SF6, at rated filling density. The switchgear is considered
adequate if no external effects other than the operation of pressure relief devices occurs within the specified time
and if escaping gases are directed so as to minimize the hazard to personnel.
In test practice, because equipment of this voltage class is normally SF6-filled, and release of (contaminated) SF6
into the environment may not be acceptable, it is common practice that such tests are performed on GIS (sections)
that are contained in a pressure-resistant container of adequate size.
5.1.2 Metal/insulation enclosed switchgear ≤ 52 kV
Internal arc testing of metal-enclosed switchgear is intended to offer a tested level of protection to persons in the
immediate vicinity of the switchgear in the event of an internal arc. Effects from an internal fault arc, such as
overpressure acting on covers, doors, inspection windows etc., as well as the thermal effects of arc(s), arc roots,
ejected gas(es) and glowing particles are included. In contrast to internal arcing in GIS > 52 kV, the standards IEC
62271-200, IEC 62271-201, and IEEE Std C37.20.7, allow no possibility of verifying internal arc withstand ability by
calculation, even when based on the testing of equivalent designs. For this reason, and because of the generally
closer proximity of medium voltage installations to the public compared to high-voltage installations, internal arc
testing of metal-enclosed medium voltage switchgear is common.
5.2 Standardization and Test Experience

5.2.1 Standardization status
With the release of IEC 62271-200 in 2003 an Internal Arc Classification (IAC) was defined, taking into account
various levels of accessibility of the switchgear:
 Type A: Accessible by authorized personnel only;

 Type B: Accessible by general public;
 Type C: Not accessible, i.e. out of reach (pole-mounted switchgear);
For assessment of the thermal effects of the hot gases expelled from the installation due to pressure rise caused
by the fault arc, special black cotton cloth indicators of size 15x15 cm are used in a steel frame to avoid mutual
ignition (see Figure 5-1). The indicators are mounted on a rack (vertically and horizontally) arranged in a
checkerboard pattern, covering 40-50 % of the area of the accessible switchgear side, or of the 3x3 m2 area below
the switchgear in case of pole-mounted apparatus.
An important (by test-experience the most critical) criterion to pass internal arc tests besides the mechanical
withstand of the enclosure is the absence of ignition of any indicators by hot gases. Ignition by glowing particles,
however, is allowed. High-speed video is normally used in order to make a distinction between the causes of
ignition. However, in many cases, the real reason of ignition (hot gases or particles) cannot be identified, for
example if the flight of a particle is not in the area of sight of the camera.
Depending on the accessibility type, two degrees of flammability of the cotton indicator cloth are required,
2 2
expressed in their specific weight (150 g/m for type A, 40 g/m for type B, C). The variation fabric imitates the
clothing of people, with the heavier cotton for the authorized personnel and the lighter for the general public's
garments.
Page 59
Indicators have to be located at all accessible vertical sides of the switchgear, for type A at 30 cm distance and for
type B at 10 cm distance. In addition, horizontal indicators have to be installed in a prescribed way.
In order to represent the flow of expelled hot gases, the room in which the switchgear is to be installed is simulated
with a floor, ceiling and two walls perpendicular to each other (see Figure 5-1). The simulated room does not allow
the impact of pressure rise on the building structure to be assessed, but is only intended to represent a realistic
environment for the flow of exhaust gases around the switchgear.
Acceptance criteria to qualify the switchgear for an IAC classification are the following:
 Criterion 1: Doors and covers must not open. Deformations must not touch the indicator racks or walls;
 Criterion 2: No parts above 60 g must be projected; enclosure must remain intact during arcing;
 Criterion 3: Arc must not burn-through an accessible side lower than 2 m high;
 Criterion 4: Indicators must not ignite due to the effect of hot gases;
 Criterion 5: The connection of the enclosure with the grounding point remains intact.
Although it is easy to measure and it’s the most important factor, IEC / IEEE standards do not request a pressure
measurement in the arc compartment during testing. Another point to note is that, due to the different and
sometimes conflicting conditions in testing (e.g. in temperature rise tests ventilation openings are welcome and in
internal arc testing ventilation openings are not welcome), a proper identification of the equipment and relief areas
of the equipment which was tested should be included in the test reports, in the absence of guidelines from the
main technical standards..
5.2.2 Test result statistics
Results of internal arc tests have been analyzed by a major test laboratory. Most recent statistics are based on 91
tests carried out in 2005 and 2006. It is the test laboratory's experience that in approximately 80 % of the tests all
criteria have been fulfilled, see Figure 5-2. The most prominent failure mode is related to criterion 4: the absence of
ignition of indicators. In 15 % of all tests, indicators ignited.
By comparison, earlier data (2001-2002 when IEC 60298 (Ed. 3.0, 1990) was in use) are also evaluated (from 137
tests): 32 % did not fulfil all criteria; also in that period, in 23 % of the tests vertical indicators ignited, and in 11 %
horizontal ones (IEC 60298 made a distinction between ignition of vertical and horizontal indicators).
Page 60
100%
not fulfilled fulfilled
80%
60%
40%
20%
0%
1 2 3 4 5 total result
criterium as specified in IEC 62271-200 A6
Figure 5-2: A major test laboratory's experience with achievement of criteria of IEC 62271-200
(population: 91 tests in 2005 and 2006).
5.2.3 Nameplate designation and certification
As a result of an internal arc test, IEC classification based on tests is denoted in the nameplate as follows:
 Classification IAC BFLR ("B" means accessibility type B, "FLR" means the Front, Lateral and Rear sides
have been confronted with indicator racks and passed criterion 4 and the other criteria);
 Numerical values, for example 12.5 kA 0.5 s (means test have been performed with 12.5 kARMS and 0.5 s
duration).
Many test laboratories (including some belonging to the Short-circuit Testing Liaison) do not issue certificates on
internal arc tests only. The reasons for this are as follows:
 Identification of the relevant parameters which determine test results in the design verification is not clear.
A well-defined system of documentation does not exist yet. Too little is known about what design
parameters determine a positive test result.
 Relation of test voltage to rated voltage.
 There are questions regarding the reproducibility of test results.
 The variety of rooms in which switchgear is located in service can never be simulated in test with a single
representative test situation.
Page 61
5.2.4 Conditions of current and voltage
Figure 5-3: The effect of arc voltage on test current. The vertical scale is per unit (pu) of the rms current
value.
The applied voltage should be equal to the rated voltage of the switchgear. In the case of test laboratory
limitations, internal arc tests can be performed with lower than rated voltage. This is subject however to the
following conditions:
1. IEC 62271-200 (Ed. 2.0, 2011) stipulates that the asymmetrical peak value of the short-circuit current should
not be lower than 90 % of the peak under rated voltage conditions. The internal arc has a reducing effect on
the asymmetrical peak at reduced (low) voltages. The arc voltage counteracts the driving voltage, thereby
reducing the arc current (asymmetrical peak value as well as a.c. component). An example of this is shown in
Figure 5-3, showing current wave traces from a 36 kV (rated voltage) switchgear, tested with a source voltage
of 7.7 kV. The measured prospective and arc current at the applied voltage of 7.7 kV are indicated together
with the calculated current at a supply voltage of 36 kV. For the calculated current at 36 kV the arc voltage
from the measured at 7.7 kV is used. Figure 5-3 demonstrates the following asymmetrical peak values:
prospective peak is 2.55; the measured peak at 7.7 kV is 2.05 while the expected value at 36 kV supply
voltage is 2.44.
This reduction of peak factor at lower supply voltages is illustrated in Figure 5-4 to some more extent.
Guidelines can be given on the ratio of supply voltage / rated voltage ( = / ) that does not lead to
unacceptable asymmetrical peak reduction.
In Figure 5-4, the asymmetrical peak factor (as a fraction of the prospective peak) is plotted against for a
number of rated voltages. Following the IEC 62271-200, supply voltage reduction is allowed, as long as the
fraction of prospective peak current is not lower than 0.9. The dot shows as an example the case of 17.5 kV
switchgear, tested with a supply voltage of 0.8x17.5 = 14 kV and leading to a peak reduction of 0.92, within
the limit set by the standard. It can be concluded that supply at 7.7 kV resulted in an asymmetrical peak value
more than 10 % below the prospected value, making the test invalid according to the standard IEC 62271-
200.
Page 62
A method of meeting the asymmetrical peak requirement also at low supply voltage is to increase the rms
current. Though asymmetrical current can then be brought within the standard, the overall stress to the
switchgear panel may be higher than that occurring in service.
The difficulty of this approach is that the arc voltage must be entered as a parameter, which requires a
database with practical values.
2. In addition, in circuits with lower voltage there is the risk of premature arc extinction, which makes the test
invalid. This occurs due to the lower available voltage that should re-ignite the arc after every current zero.
3. a.c. current must be kept at a constant level during the test duration of up to 1 s, and if this is not possible, the
duration of the test must be extended until the value of ∫ (taken as being proportional to the arc energy
assuming a constant arc voltage) equals the specified value (within 0 to ±10 % tolerance). This under the
provision that the first three a.c. half-cycles are as specified and that the current shall not be reduced by more
than 50 % of the specified value at the end of the test.
Such calculations, however, can only be performed in hindsight, since arc behavior cannot be predicted. This
leads to imprecise representation of stresses.
fraction of prospective peak factor
0.95 36 kV
24 kV
0.9
17.5 kV
2.5 prospective peak

0.85 actual peak
2
1.5
12 kV
1
0.8
0.5
-0.5
testvoltage -1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
0.75
0.4 0.5 0.6 0.7 0.8 0.9 1
Figure 5-4: Reduction of prospective asymmetrical peak factor (vertical axis) by arc voltage (700 V) vs test
voltage (fraction of rated voltage, horizontal axis) for 4 rated voltages. Inset: prospective- and reduced
current peak (current in per unit of rms current).
Page 63
5.2.5 Observed effects of insulation gas on the effects of internal arc
In this section, the special status of SF6 gas in internal arc testing is reviewed [Smeets2008]. The background of
this is the present discussion on banning SF6 as a filling gas during internal arc testing of medium voltage SF6
insulated switchgear. In the present standard IEC 62271-200 (Ed. 2.0, 2011) it is stated (clause 6.106.3): "For
environmental reasons, it is recommended to replace SF6 with air at the rated filling pressure ( 10 %)" including
the note: “Test results with air instead of SF6 are considered to be representative”.
From environmental reasons there is a clear motivation for this, since solid (metal-sulfides and -fluorides) as well as
gaseous SF6 decomposition products (SF4, H2S, SO2, HF, S2F10, S2OF10) can be poisonous, especially in the
presence of humidity. In addition, test laboratories wish to minimize their emission of clean SF6, a greenhouse gas,
and certainly of polluted SF6. Technically, however, it is not clear yet, that testing in air presents similar conditions
as testing in SF6.
A short overview of some past published work by manufacturers is presented in Section 5.2.5.1, based on three
series of tests, followed in Sections 5.2.5.2 and 5.2.5.3 by the results of two series of tests to compare internal
arcing phenomena in SF6 and air (and other gases).
5.2.5.1 RESULTS FROM THE LITERATURE

3
The first series of tests was performed with a transformer substation model, with a 0.3 m SF6 compartment
[Daalder1989]. Three-phase current (15 - 20 kA) was supplied from a 7.2 kV circuit to arcs between Cu electrodes
with 100 mm distance. The main difference between SF 6 and air was found to be:
 A significantly higher arc voltage in SF6 (720 V) than in air (480 V);
 The period during which pressure is above 50 % of its maximum value in case of SF6 is 2.4 times longer
than in air (in this test).
3
In a second research program a series of tests were performed with an arcing volume of 0.2 m , exhausting in a 8
m3 air-filled volume [Dullni1994]. Test current was single phase (10 - 20 kA) and arcing was in a 100 mm gap. The
importance of these tests is the measurement of pressure-rise in the exhaust volume, as a simulation of the room
in which the switchgear is located. It was concluded that:
 Rise and fall (after pressure relief) of pressure in the arcing volume are faster in air than in SF6;
 The pressure rise in the exhaust volume is twice as high when the arc was burning in SF 6. This is attributed
to the higher heat capacity of SF6 and the longer duration of stay in the arcing volume;
 A wide range of arc voltages has been encountered, in SF6 (500 - 1900 V) notably higher than in air (400 -
1000 V).
A third series of tests were carried out with "typical gas-insulated metal-enclosed switchgears", with three-phase
current of 16 kA [Bjørtuft2005]. The authors conclude on the difference between internal arcing in SF6 and in air:
 A significant faster pressure increase in air than in SF6. This has led to higher internal pressures before
pressure relief in air;
 differences exist in exhaust characteristics in SF6 (slower cooling).
Page 64
5.2.5.2 THREE-PHASE TESTS IN A CLOSED TANK
In the period 1987-1988 a large number of internal arc tests were carried out on a full-scale model of a 10 kV
3 3
switchgear (0.12 m volume) and 20 kV switchgear (0.24 m volume) [Pettinga1989]. Test current was in the range
4.5 - 18.5 kA. There was no pressure relief device on the tanks (8 mm steel), the fault arc was bypassed with a fast
making switch as soon as pressure came close to the withstand capability of the tank (approx. 9 bar). Three-phase
arcs were initiated with a 0.5 mm copper wire. The gap length was 50 mm. Power was supplied from a three-phase
10 kV generator source.
In Figure 5-5, pressure-rise data are presented for different gases: nitrogen, argon, air, SF 6 for the highest and the
lowest current values applied in the larger tank volume (0.24 m3). It can be observed that the pressure-rise at both
current values is slower in SF 6 than in the other gases, and thus confirms the other sources [Daalder1989,
Dullni1994, Bjørtuft2005].
18.3 kA
pressure rise (b)
4.6 kA
Ar 1.2bar
2 Air 1 bar
N2 1 bar
SF6 1.2 bar
1
0 100 200 300 400 500 600 700
arc duration (ms)
Figure 5-5: Pressure rise from a three-phase fault internal arc in various gases (argon, air, nitrogen and SF6
with initial pressure 1 or 1.2 bar, see legend ) in a cubicle model of 0.24 m3 ( 18.3 and 4.6 kA).
5.2.5.3 SINGLE-PHASE TESTS IN A TANK INCLUDING PRESSURE RELIEF
A series of new tests were performed, purely aimed at comparing the effects of arcing in SF6 with air under
identical conditions of current, driving voltage, arc duration, geometry, contact material etc. Test parameters were:
3
arc current 14.2  0.3 kA, driving voltage 15.5 kV, frequency 60 Hz, tank volume 0.53 m , electrode material Cu,
gap 7.5 cm. Filling (over)pressure prior to arcing is 0.7  0.1 bar. In each test, the tank is equipped with a rupture
disk set for pressure relief at approximately 2.8 bar.
Measured quantities were arc current and voltage, tank pressure. The exhaust gas was monitored by infrared
thermography; the arc was monitored with high-speed video.
An impression of the effect of the exhaust gas can be obtained from Figure 5-6, where tests with arcing in SF6 (left)
and air (right) are compared (arc duration 1 s with symmetrical current).
A dummy was placed 5 m from the exhaust opening (see Figure 5-6). The different effect of SF6 and air exhaust
gases on the T-shirt, worn by the dummy, is illustrated in Figure 5-7. This observation, however, does not allow a
scientific evaluation, though the clothing which interacted with SF6 is most affected.
The different effects of arcs burning in air and SF6 are:
Page 65
1) Arc energy. A relevant quantity in internal arcing is the electrical energy supplied to the arc inside the
tank. The arc energy is calculated as iauadt, with ia, ua the momentary arc current and voltage. Typical
results for arc durations of 0.5 and 1 s show clear differences between SF6 and air (see Figure 5-8):
o the energy of SF6 arcs is (initially) smaller than of arcs in air. This is a direct consequence of the
initially lower arc voltage of the SF6 arc;
o the energy in air arcs rises linearly with arcing time, whereas the energy of the SF6 arc shows a
steeper rise than proportional with time. As a result, at longer arc duration, the energy supplied to
SF6 is higher than that supplied to air.
dummy
Figure 5-6: Photographic impressions of the release of hot gases as a result of arcing in SF6 (left column)
and air (right column). Arc duration was 1 s, pictures are taken with an interval of 0.2 s.
SF6 air
Figure 5-7: T-shirt worn by dummy and affected by SF6 gas (left) and air (right).
Page 66
5
air 1 s
4 SF6 1 s
arc energy (MJ)

pressure relief action
3 air 0.5 s
2 SF6 0.5 s
0
0 200 400 600 800 1000 1200
time (ms)
Figure 5-8: Arc energy vs time for air- (blue) and SF6-filled volume (red) for 0.5 s and 1 s arc durations.
The difference in arc energy is clearly caused by arc voltage. During arcing, the true-RMS voltage in air
decreases over time while it increases in SF6 (up to 550 V after 1 s). In a number of tests, the increase in
SF6 arc voltage occurs around the time when the rupture disc bursts. Fluctuations and sudden jumps in arc
voltage in SF6 (Figure 5-8) suggest erratic motion of the arc. The increase of the arc voltage may, however,
also be related to high gas pressure inside the compartment. The investigations in reference [Dullni1994]
revealed an increase of the arc voltage with higher pressure in the volume. Considering the different time
intervals, when air and SF6 are at high pressure in the compartment (see Figure 5-9), it can be concluded
that the arc in air develops high arc voltage in the beginning of the test with a drop in arc voltage after 0.4 s,
whereas the arc in SF6 shows its highest arc voltage after 0.5 s when the pressure is still rising.
In the literature there are quite different statements about the value of SF6 arc voltage compared to that in
air at the same filling pressure: In some experiments a lower voltage was obtained, in others a higher one.
If a pressure-dependent arc voltage is assumed, these differences may easily come from the different
development of pressure during arcing. The different behavior of air or SF6 arcs on electrodes and their
different instabilities may also play a part in this.
Under identical conditions, the arc energy (and pressure rise) may vary, depending on the dynamics of the
arc. Only a few tests are repeated under identical conditions in the database of Working Group A3.24. Two
of these are shown in Figure 5-8. As can be seen, the arc energy (during the first 0.5 s) in SF6 and in air
differ at the most 5 %. However, these single-phase results may not be generalized to conclusions in three-
phase tests.
2) Cooling of exhaust gas after arcing. By monitoring the exhaust jet with time-resolved infra-red
thermography, an impression of the cooling process of the hot gases could be obtained. This is illustrated
in Figure 5-10, in which the maximum temperature in the exhaust gas volume is shown over time. In Figure
5-11, infrared pictures of the exhaust gas are shown for SF 6 and air, recorded at comparable time after arc
initiation.
From this, it becomes clear that:
Page 67
 the exhausted air is hotter than the exhausted SF 6 (actually the maximum temperature in the air is
o
above 2000 C, since the IR camera saturated above this level).
 the cooling down after arcing in SF 6 is much slower than in air, both in the case of the 0.5 s and the
1 s arc durations.
 the exhausted air has a wider jet stream than the exhausted SF6.
4
arcing volume
3
pressure rise (bar)
exhaust volume
0
0 500 1000 1500
time (ms)
Figure 5-9: Pressure rise in arcing- and exhaust volume for air-filled (blue) and SF6-filled (red) arcing
volume (1 s arc duration). Vertical markers: pressure relief action (diaphragm burst).
max temperature
SF6 1s arc
air 0.5 s arc

SF6 0.5 s arc
air 1 s arc
0 300 600 900 1200 1500 1800 2100

time after arc initiation (ms)
Figure 5-10: Estimated maximum temperature of exhaust gas (air, SF6) vs time after arc initiation.
Page 68
SF6, after 1s SF6, after 1.8s
air, after 1s air, after 1.8s
Figure 5-11: Infra-red thermal images from the expelled gases resulting from an internal arc test with 1 s
duration. Upper pair: tank filled with SF6; lower pair: tank filled with air; left pair: 1 s after arc initiation;
right pair: 1.8 s after arc initiation.
3) Exhaust in a defined volume. In a number of tests, the expelled gas was not exhausted to the environ-
3
ment, but to a collector tank of 1.2 m . Pressure was measured in this tank as well. In Figure 5-9, pressure
measurements of SF6 and air are combined for the two-tank situation. It can be concluded that:
a. The pressure rise in the arcing volume in case of air is much faster and reaches higher values than in
SF6. This is in accordance with all earlier investigations [Daalder1989, Dullni1994, Pettinga1989];
b. The pressure relief disc operates at a (somewhat) higher pressure in air than in SF6. This is (probably)
because the inertia of the disc bursting process causes the disc to operate at a somewhat higher
pressure when the rate of pressure rise is higher. This phenomenon is frequently observed in tests.
c. The pressure in the exhaust volume reaches a higher value in the SF 6 case than in the case with air.
From these tests it follows that the arc energy of arcs in air in a worst-case situation could be regarded
as being lower than in SF6.
d. Pressure inside the arc compartment. As shown by several authors the maximum pressure in a closed
arc compartment is higher if it is filled with air instead of SF6. This can be true even when the arc
energy in SF6 is higher than in air. The reason is the larger heat capacity of SF6, which compensates
for the higher arc energy. The smaller heat capacity of air leads to a faster pressure rise and an earlier
bursting of the rupture disc. When the pressure relief disk opens it limits the pressure rise in the arcing
compartment. There is a tendency for the opening to occur at a higher pressure in air due to inertia
effects. From this it follows that the arc compartment will be stressed in a comparable or even more
severe way by an arc in air.
e. Exhaust of gases via an intermediate compartment. If the overpressure of the arc compartment is
directly discharged into the environment (room), the hot gas stream will affect the indicators
immediately. However, in general, metal-enclosed switchgear consists of several compartments with
only the "arcing" compartment filled with SF 6. In this case, hot SF6 first of all will exhaust to a
neighboring air-filled compartment (intermediate room, e.g. cable compartment, pad mount enclosure)
within the switchgear before leaving it e.g. through fissures.
Page 69
4) Ignition of indicators. Ignition of any material means exothermic reactions start to run. This happens
when a certain activation energy level is exceeded. Cotton consists of carbon-hydrogen-oxygen
compounds. In air, ignition is an oxidation process, mainly the reaction of carbon with oxygen to produce
CO and CO2. A similar process (without oxidation) happens with SF6. To reach the activation energy, heat
must be transferred from the gas stream to the cotton. Given the complexity of the energy transfer
(turbulence, chemical reactions, strongly inhomogeneous time-dependent flow and temperature patterns),
modelling of this process is beyond the reach of the Working Group A3.24. The influential parameters are
gas temperature, heat transfer duration, thermal conductivity and gas velocity. They act as follows:
a. Gas temperature: The higher the temperature difference between gas stream and indicator (as with
air), the higher is the energy flux applied to the indicator.
b. Heat transfer duration: The longer the duration of the gas flow (as in SF 6), the more energy will be
transferred to the indicators.
c. Thermal conductivity: In the temperature range from 1500 to 4000 K the thermal conductivity of SF6
is larger than that of air, enlarging the energy flux applied to the indicator.
d. Gas velocity: The higher the gas velocity (as with air), the higher is the turbulence in enlarging the
temperature at the boundary of the gas stream (increasing the temperature gradient to the indicator).
Moreover, with higher gas velocity, the boundary layer between hot gas and the surface of the indicator
is thinner. Higher turbulence and a thinner boundary layer improve heat conduction to the indicator and
can lead to easier ignition of indicators.
Given that these various contributing factors have both ignition-enhancing and -impeding effects in air and
SF6 further research is necessary to determine the overall effect.
5.3 Summary
Based on the results, the authors conclude that relevant differences exist in the behavior of fault arcs in SF6 and in
air, and in their effects on switchgear and the environment.
Replacing SF6 by air (all other parameters being equal) in internal arc testing leads to comparable or higher
pressure rise in the arcing compartment. Pressure rise in an adjacent compartment or a switchgear room into which
the exhaust gas is released may be lower in tests with an air-filled arc compartment than in tests with an SF6-filled
arcing compartment. No conclusions exist on other criteria required to pass an internal arc test such as the ignition
of the cotton indicators and enclosure burn-through. This needs further detailed investigation.
The following conclusions are drawn:
a. Arc compartment: The mechanical stress of the arcing compartment filled with air is higher than
with SF6, i.e. if criterion 1, 2 of IEC 62271-200 (Ed. 2.0, 2011) are fulfilled with an air test, this will
be true for SF6 as well.
b. Intermediate compartment: When exhaust gas from the arcing compartment is released into
adjacent compartment(s) the mechanical stress of it is larger in tests with SF6 than with air.
c. Indicators: The ignition probability of cotton indicators depends in a complicated manner on the
energy content and temperature of the exhausted gas, as well as the energy transfer to the cotton
Page 70
material. Therefore, it cannot be concluded that SF6 or air is more effective. The verification of
personnel safety by criterion no. 4 of IEC 62271-200 should be given with SF6 as well as with air
assuming that the energy of the exhausted gas is sufficient for indicator ignition and knowing that
the exhaust of both SF6 and air occurs as a stream with a reach farther than the distance from the
panel to the indicators also after reflecting at walls and ceiling. More tests would be needed for
final confirmation.
Also, the authors noted that IEC / IEEE standards currently do not request a pressure measurement in the arc
compartment during testing. A lot of useful information which could be used in the design of the switchgear can be
obtained from the measurement of the overpressure curve.
REFERENCES:
[Bjørtuft2005]: T. Bjørtuft, O. Granhaug, S.T. Hagen, J.H. Kuhlefelt, G. Salge, P.K. Skryten, S. Stangeherlin,
"Internal Arc Fault Testing of Gas Insulated Metal Enclosed MV Switchgear", 18th Int. Conf. on El. Distr. (CIRED),
2005.
[CIGRE WG 23.03]: CIGRE WG 23.03, “Pressure Rise in Metal-Enclosed Switchgear of Single Phase Enclosure
Type due to Internal Arc. Evaluation of various International Test Results and Study of Calculation Procedure”,
Electra 93, pp. 25-52, 1984.
[Daalder1989]: J.E. Daalder, O. Lillevik, A. Rein, W. Rondeel, "Arcing in SF 6-MV-Switchgear. Pressure Rise in
Equipment Room", Int. Conf. on El. Distr (CIRED), 1989.
[Dullni1994]: E. Dullni, M. Schumacher, G. Pietsch, "Pressure Rise in a Switchroom Due to an Internal Arc in a
Switchboard", Proc. 6th Int. Symp. on Short-Circuit Currents in Power Systems, Sept. 6-8, 1994.
[Pettinga1989]: J.A.J. Pettinga, "Pressure-rise tests due to a High-Current Internal Arc in a MV Cubicle Model", Int.
Conf. on El. Distr. (CIRED), 1989 and KEMA internal report 00880-DZO 88-2046 (88-10), 1988.
[Smeets2008]: R. Smeets, J. Hooijmans, H. Bannink, H. Barts, P. Leufkens, N. Uzelac, P. Milovac, D. Kennedy, G.

J. Pietsch, K. Anantavanich, “Internal Arcing: Issues Related to Testing and Standardization”, CIGRE Conference
2008, Paper A3-207, Paris, France, 24.-29.08.2008.
[Trinh1992]: N. Giao Trinh, “Risk of Burn-Through – a Quantitative Assessment of the Capability of Gas Insulated
Equipment to Withstand Internal Arcs”, IEEE Trans. On Pow. Del., vol. 7 no. 1, 1992.
Page 71
6 EFFECT OF INTERNAL ARC ON STRUCTURES

6.1 In troduction
The internal arc is initiated at a particular place in the switchgear during a type-test, see IEC62271-203 (Ed. 2.0,
2011) for high voltage GIS switchgears for example. There are three major effects which affect the switchgear and
adjacent personnel.
1. Mechanical stress on the switchgear due to the overpressure

2. Mechanical stress on the building walls due to the overpressure
3. Burn-through
Sections 6.2, 6.3 and 6.4 describe and evaluate these effects in detail.
 The first effect (see Section 6.2) is the overpressure generated by the arc that creates mechanical stress
on the enclosure structure causing deformation of walls, blown out doors, etc. Generic constructions are
shown in Figure 6-1.
a. In Figure 6-1 (a), a typical construction of an air insulated switchgear (AIS) with grounded
enclosure used in medium voltage is shown. The walls of the enclosure are made of a sheet metal
with a certain thickness connected by bolts spaced by a certain distance. For simplicity, the doors
are not shown in the figure. The pressure relief devices are flaps which are structurally weaker
than the main enclosure, so that in case of high pressure inside the panel they will be first to
rupture and release the excess pressure. Usually the flaps start to open at an internal pressure
around 0 to 30 % higher than the external pressure.
b. Figure 6-1 (b) shows a typical construction of a gas insulated enclosure used in medium voltage
switchgear (GIS). The walls of the enclosure are made of a sheet metal with a certain thickness
and the walls of the enclosure could be welded or bolted together. The pressure relief device is
usually a rupture disk which opens at a specific overpressure, for example 200-300 kPa depending
on the application and the protection philosophy.
c. Figure 6-1 (c) shows a typical construction of a high voltage GIS busbar arrangement. The external
envelope is composed of a metallic cylinder with a certain thickness which contains a concentric
conductor. Insulating spacers support the conductors and divide the GIS into separate gas
compartments. The pressure relief device is usually a rupture disk which opens at a specific
overpressure, for example 1000-1500 kPa depending on the application and the protection
philosophy.
 The second effect (see Section 6.3) has an impact on the switchgear room and building. Hot gases will be
ejected through the pressure relief device of the switchgear enclosure and cause pressure buildup in the
rooms and buildings. The pressure relief devices of the installation room start to open when the
overpressure reaches a certain value.
 The third effect (see Section 6.4) is the “burn-through” effect. This effect is caused by the arc which can
burn on a surface of the metallic enclosure (like a switchgear wall or panel, or GIS bus duct). This melts
and then punctures the wall of the enclosure.
Page 72
(a) Air insulated switchgear (AIS) medium voltage (b) Gas insulated switchgear (GIS) medium voltage
HV conductor
Gas at higher pressure than the Pressure relief device

external pressure before arc
Enclosure
Insulating spacer
(c) Gas insulated switchgear (GIS) High Voltage
Figure 6-1: Typical construction of enclosures.
6.2 Mechanical stress on switchgear due to the o verpressure

The overpressure caused by internal arc may exert severe stress on the enclosure of switchgear. If the structure is
not strong enough it may fail releasing high velocity metal debris and hot gases to the surroundings (see Figure
6-2). This is a major safety issue for personnel inside the installation room. In this section, calculation methods to
estimate the mechanical withstand capability of the enclosure structure against internal arc events are provided.
The present section starts with a calculation method for static loads and then moves into complex structures and
dynamic loads.
Previous sections of this brochure have shown how to calculate the overpressure values generated during an
internal arc. Consider that the overpressure curve is known for a non-deformed geometry and an assessment of
the withstand strength of the enclosure during an internal arc is to be made. In the case where the deformation of
structure is important, the overpressure curve considered from calculation with the basic model may differ from the
actual pressure rise. The enclosure fails when the stress exceeds the mechanical strength of the material.
Page 73
Figure 6-2: Test object after arc resistance test.
A switchgear is composed of metal plates or cylinders welded or bolted together. During the design phase the
designer should know, through simple calculations, whether the metal plates or flanges being used are strong
enough to deal with overpressure events. The idea is to estimate the load bearing capacity of the metal plates and
check this against the expected overpressure. In the following section some guidelines are provided for the
designer to gauge the strength of metal plates with different mounting conditions namely bolting, welding etc.
6.2.1 Calculated results for the failure of metal plates due to static overpressure
In this section typical failure details of standardized metal plates due to static overpressure are presented. The
plate will deform elastically up until the yield point and then deform plastically until failure. The material properties
governing these deformations as well as the typical load and deformation values which a plate can sustain are
provided for principal alloys.
All the plates are square with dimensions 1 m x 1 m, with thicknesses of 2 mm, 5 mm and 6 mm. Three types of
boundary conditions, with the plate edges fixed in different ways, are analyzed:
1
1 The plate edges are simply-supported. This is a simplified analysis for welded edges . See Type 1 in
Figure 6-3.
2
2 The plate edges are fastened in place by distant equally-spaced bolts. 8 bolts of diameter 50 mm.
See Type 2 in Figure 6-3.
1
Simply supported means fixed edges with pivoting. In reality the deformation seen in switchgear is such that the edges of sheet (wall) are
not fixed during pressure build-up (See an example in Figure 6-6). The edges move in 3-d. For accurate mechanical stress simulation, the
switchgear tank deformation has to be calculated by 3-d simulation.
2
The choice of 50 mm bolts simulates a rigid bolted connection.
Page 74
3 The plate edges are fastened in place by closely spaced bolts. 72 bolts of diameter 10 mm located
every 25 mm. See Type 3 in Figure 6-3.
All three types are calculated for yield and failure using materials AISI 304 and AISI 10103.
Since the plates have non-standard supports and a large deflection for which an analytical solution doesn’t exist, a
finite-element method (FE method) is used to calculate the structural strength. The material properties used are
given in Table 6-1.
Young’s Yield Tangent Ultimate

# Material
Modulus Strength Modulus Strength
(Pa) (Pa) (Pa) (Pa)

1 AISI 304 1.93E11 220E6 1.8E9 520E6
2 AISI 1010 2E11 305E6 1.45E9 360E6
Table 6-1: Material properties of two AISI alloys. The ultimate strength (or ultimate tensile strength) is the
maximum stress that the material can withstand while being stretched before failing.
Typical constructions of plates used in the examples are shown in Figure 6-3. In Figure 6-4 the typical deformation
of those 3 plates is shown. Figure 6-5 shows the typical mechanical stress profile of the plate for those 3 boundary
conditions. The summary of the results is shown in Table 6-2 and Table 6-3.
Type 1: Continuously (simply) Type 2: Bolted (spaced) Type 3: Bolted (dense)

supported
Figure 6-3: Boundary conditions of metal plates for failure analysis.
FEA results are for illustration purpose. It is important to note that such studies are highly dependent (among other
parameters) on the mesh refinement, as well as on the criterion used to define rupture.
3
AISI 304 is a corrosion-resistant stainless steel and AISI 1040 is a wrought carbon steel.
Page 75
Type 1: Continuously (simply) supported
Type 2: Bolted (spaced)
Type 3: Bolted (dense)
Figure 6-4: Typical deformation profiles of the plate Figure 6-5: Typical von Mises stress profiles of the
(5 mm thick AISI 1010) under different boundary plate (5 mm thick AISI 1010) under different
conditions. boundary conditions.
2 mm Plate 5 mm Plate 6 mm Plate

Support Pressure Maximum Pressure Maximum Pressure Maximum
# Material
Type value at deflection at value at deflection at value at deflection at
yield yield yield yield yield yield
Page 76
(Bar) (mm) (Bar) (mm) (Bar) (mm)

1 AISI 304 1 1.2 18 1.5 15.5 2.1 14.8
2 AISI 1010 1 1.3 19.8 2.6 18.4 3.6 17.3
3 AISI 304 2 0.03 6 0.07 3.4 0.1 3.2
4 AISI 1010 2 0.07 7.6 0.13 5.0 0.2 3.7
5 AISI 304 3 0.15 8.2 0.45 7.6 0.5 6.8
6 AISI 1010 3 0.3 9.9 0.6 8.5 0.9 9
Table 6-2: Typical values at yield for plates.
2 mm Plate 5 mm Plate 6 mm Plate

Support Maximum Maximum Maximum
# Material Pressure Pressure Pressure
Type deflection deflection deflection
value at value at value at
before before before
failure failure failure
rupture rupture rupture
(Bar) (mm) (Bar) (mm) (Bar) (mm)
1 AISI 304 1 18.4 195 48 192 56 188
2 AISI 1010 1 8 87 18.9 83 21 76
3 AISI 304 2 1.9 52.7 4.8 50.6 6.3 50
4 AISI 1010 2 0.6 19 1.4 18 1.9 17
5 AISI 304 3 5 61.5 13.5 63 15.2 59
6 AISI 1010 3 2.4 24 4.7 21 5.5 20.7
Table 6-3: Typical values at failure for plates.
The following general observations can be made from Table 6-2 and Table 6-3:
 A simply-supported plate can withstand higher pressure than a bolted plate because of a rotational degree
of freedom at boundaries, which transfers the critical failure stress to the geometric center of plate and
hence reduces high localised stress on the support.
 A plate with a higher number of bolts can withstand higher pressure than a plate with a lower number of
bolts. A plate with a smaller number of bolts fails due to high localised stress around these bolts.
 A thicker plate will deform less and will withstand higher pressure.
 The ductility of material AISI 304 helps it in withstanding higher overpressures even though it has low yield
strength.
This section provided the basic understanding of the effects of different types of material and support on the
mechanical strength of metal plates. A basic assumption is made that the load acting on the plates is static, which
means it is applied slowly so that dynamic effects can be neglected, but it has provided a guideline for the strength
of plates.
In the next section, the rate of rise of overpressure will be taken into account when estimating the mechanical
stresses during an internal arc.
Page 77
6.2.2 Calculation results for the failure of metal plates due to dynamic
overpressure
Since the internal arc overpressure is a short duration time-varying load, the inertial effect has to be considered
when calculating deformation and stresses. The structure behaves differently if the force is applied slowly
compared to when it is applied in an instant.
Mathematical models are created for dynamic loading using a simplified mass-spring representation for the
structure. This can be used to estimate the deformation for simplified geometry in the elastic and plastic regimes.
6.2.2.1 ELASTIC DEFORMATION REGIME
The method is based on converting a structure into a spring mass system and solving it for arbitrary excitation.
Equation (6-1) is the governing equation.
( )
+2 + = (6-1)
Where is deformation, is time, is mass, ( ) is force, is the natural frequency, and is the damping ratio.
The equation can be solved analytically or numerically for an arbitrarily-varying load for small time steps to yield
displacement at time
( )= { ( ) sin − ( ) cos } (6-2)
where
( )= ( )+ ( ) cos (6-3)
( )= ( )+ ( ) sin ω (6-4)
where is the damped natural frequency.
So, for a given time-varying overpressure the deformation of a simplified plate can be calculated in the elastic
regime with equation (6-2). The stresses are calculated from the resulting displacement and compared against the
mechanical failure properties of the material.
It is important to remember that equation (6-2) is valid for displacement in the elastic regime only when there is no
permanent deformation.
6.2.2.2 ELASTO-PLASTIC REGIME
Page 78
When the load is so high that the structure undergoes plastic deformation, the above mentioned equation (6-2) is
not valid, as it doesn’t account for changes in the stiffness of the structure in the plastic region.
So equation (6-2) is rewritten for a given instant of time for the next incremental change in loading. This is then
solved for a different stiffness in the plastic region. For simple cases and low plastic deformation, the restoring force
is considered constant after the yield point. However, a separate non-linear stiffness function can be used for
elastic and plastic parts if needed.
The dynamics equation for a nonlinear single degree of freedom system becomes incremental at a given time
̈ + ̇ + = (6-5)
where is displacement , ̇ is velocity and ̈ is acceleration for , which is the applied force, Δ represents the
change in value during an infinitesimal time increment at the th time step, is the mass , is the damping, and
is stiffness.
The solution of equation (6-5) yields
Δ =Δ (6-6)
where , are effective stiffness and force:
6 3
= + + (6-7)
Δ Δ
6 Δ
Δ =Δ + ̇ +3 ̈ + 3 ̇ + ̈ (6-8)
Δ 2
The stiffness for the th time step is taken from the actual force versus displacement graph of the material or it can
be assumed constant for the elastic region and ‘0’ for plastic region. The above equations can be solved for each
time step starting from the given initial condition. The stresses are calculated from the deformation plot of the
above equations.
As mentioned earlier the applicability of above equation becomes difficult for complex structures and high plastic
deformations. For example, take the case of a plate which is deformed by a given time-varying load and undergoes
plastic deformation. However this plastic deformation is not uniform throughout the plate. A part of the plate
undergoes plastic deformation whereas the other areas remain elastically deformed. This presents a problem for
stiffness being used in the above-mentioned formula at a given time step.
So the next logical step is to divide the plate into small discrete subsections and define the stiffness of these
subsections based on an elastic or plastic regime. The division of the structure into discrete elements can be
handled efficiently with Finite Element Analysis (FEA). The finite element method is a numerical method for solving
partial differential equations in physical systems. The solution of standard differential equations exists for simplified
cases. When complicated geometry or non-linearity are involved, an approximate numerical solution is the right
way to proceed. The finite element formulation results in simultaneous algebraic equations for the degrees of
freedom. These algebraic equations provide a solution for the primary variable at a discrete number of points rather
than during the continuous solution. The mechanical body is modeled as the union of smaller sub-bodies called
finite elements. The response of these finite elements is formulated by the governing differential equation. The
Page 79
assembly of these elements provides the solution for the complete structure. So the finite element method
formulates the equations for finite elements and combines them to get the solution for the complete body.
6.2.3 Application
As stated earlier certain areas suffer plastic deformation and others elastic deformation. In order to account for this
non-uniform stiffness, finite difference/element methods are used. One case from the previous internal arc tests is
taken to check the validity of the method (Case A in Table 2-3). The before and after test pictures of the switch tank
are shown in Figure 6-6. FEA analysis was used to calculate the von Mises stress and the static mechanical
deformation of this tank under two different conditions; in the first simulation the tank was filled with SF6 gas. In the
second simulation the tank was filled with air (see Table 6-4).
Simulation no. 1 Simulation no. 2
Type of Gas SF6 Air

Construction: Welded Welded
Steel: AISI, 1010, 6 mm thick AISI 1010, 6 mm thick
13.8 bars in 70 ms (from 13.8 bars in 12 ms (from simulations,
Maximum Pressure:
measurement) using basic model)
Table 6-4: Two cases that are used for the FEA analysis of the mechanical deformation and von Mises
stress.
Figure 6-6: Deformation of structure after an internal arc test in the SF6-filled switch.
6.2.3.1 SIMULATION WITH AN SF 6 -FILLED TANK
In the first case, a non-linear transient FEA calculation is performed on the SF 6-filled switch tank. The simplified
representation of the tank is divided into discrete finite elements which are connected together, as seen in Figure
Page 80
6-7. The edges of the bottom plate of the switch tank are simply fixed to the floor. A bilinear representation of
stress-strain graph for AISI 1010 is taken as an input. Geometric non-linearities due to excess deformation of
structure are also considered. A time-varying pressure as specified above is applied to the inside surfaces of the
model. After the calculation, a plot of deformation and stress over time is generated and maximum deformation
(Figure 6-7) is compared against the deformation measured during the type test. The stresses (Figure 6-8) are also
compared against the ultimate strength of the material to check for failure.
Figure 6-7: Total deformation of structure Figure 6-8: von Mises stress for corresponding
(simulation) after an internal arc test in an SF6-filled deformation.
tank.
The order of magnitude of the deformation is valid and the shape of the test object after the test is well predicted by
the finite element computation. As can be seen from the above calculation the total residual deformation is 81.5
mm, which matches closely with the measured deformation in the picture of 81 mm.
6.2.3.2 SIMULATION WITH AN AIR-FILLED TANK
In the second case, the FEA analysis was carried out on the air-filled switch tank. Although the calculated
maximum pressure in this case is identical to the measured maximum pressure in SF 6-filled tank, the time to peak
pressure is much less in the air-filled than in the SF6-filled tank. This excites the eigen modes (fundamental
frequencies) of the tank differently than the pressure rise in SF6.
The residual deformation of the tank is shown in Figure 6-9 and the corresponding von Mises stress in Figure 6-10.
The maximum deflection is 85.3 mm, which is higher than the deflection in SF6 (81.6 mm). So for this particular
tank, testing in air is more severe than testing in SF6.
Page 81
Figure 6-9: Total deformation of structure Figure 6-10: von Mises stress for corresponding
(simulation) after internal arc test in air. deformation.
The comparison between the mechanical deformation in the air-filled and SF6 filled tanks is shown in Figure 6-11.
Figure 6-11: Deformation time-history (simulation) Dynamic effects for test in SF6 and air.
The response of an enclosure to the loading in SF6 or air depends on the mechanical construction. If the frequency
of an eigenmode of the structure is close to the frequency calculated from the time to reach the peak pressure for
internal arc, the mechanical stress gets worse. Since the rate of rise is higher for air than for SF6, air is in principle
able to excite more eigenmodes. The representative case presented in this subsection indicates that the testing in
air is more severe than testing in SF6 (with respect to the mechanical stress on the switch tank) and replacement of
SF6 gas with air during internal arc testing is justified.
The bottom line is that by use of theoretical simplification the failure of a switchgear structure can be predicted.
Simplified calculation without dynamic effects guides the designer during the design stage. Inclusion of dynamic
Page 82
effects and elastic deformation on a simple structure can be estimated with the help of equations using spring-
mass approximation as provided in the present subsection. The model can be extended to include small plastic
effects with an incremental form of the structural dynamic equation. The finite element method could be used to
fully include the effects of large deformation and non-linear material properties. Application of the finite element
method provides good match with testing done on the GIS tank. While the FEA method can be used with relative
confidence to obtain a fair representation of the deformed shape and thee qualitative effects of modifications on the
enclosure, its use to predict rupture is far from being straightforward and requires considerable judgment and
engineering skills from the user, as well as appropriate and precise representation of the material behavior until
rupture.
6.3 Mechanical stress on building walls due to overp re ssure

In the previous section the overpressure caused by an internal arc and the associated stress on the enclosure of
switchgear was analyzed. Using FEA software, the mechanical deformation and von Mises stresses of the
switchgear tank / enclosure can be simulated once the pressure rise inside the tank is known, thus providing
engineers with an important input for the switchgear design.
In this section the effect of the internal arc fault on the installation room is analyzed i.e. the mechanical stress on
building walls, as a result of the pressure rise caused by ejection of hot gases through the pressure relief devices of
the switchgear.
Section 6.3.1 shows how to estimate the maximum pressure in the installation room. Section 6.3.2 lists the
allowable maximum pressures for different wall structures. Sections 6.3.3 through 6.3.6 provide examples of
pressure rise calculations of installation rooms with the basic calculation method and CFD simulations. Finally,
Section 6.3.7 discusses the sizing of the pressure relief openings in installation rooms in general.
6.3.1 Estimation of the maximum pressure on the room walls
In the presence of an internal arc in switchgear, after opening of the pressure relief devices, the pressure field
inside the installation room is not homogeneous in space:
 In the vicinity of the pressure relief opening of the switchgear, the pressure is the highest. The closer the
gas exhaust to the wall or ceiling, the higher the local pressure on this surface.
 Wave reflections especially in room corners generate local pressure peaks.
Such a local pressure peak exerts a force on walls or ceiling. However, since the involved area is small, the force
resulting from the integration of pressure over the surface might not be significant.
It is possible to differentiate two conditions with respect to the power surface density P/S defined as follows
[Douchin2013]:
 P is the mean electrical arc power for three-phase faults (MW) (see also section 2.3.5)
 S is the total area of room pressure relief openings (m2)
The power surface density P/S is a parameter representing the inflow of arc power into the installation room - after
full opening of the pressure relief device of the switchgear - in relation to the amount of energy being lost through
the room pressure relief openings.
Page 83
 Condition 1: If P/S is large, the mean room pressure rises and reaches a maximum with local pressure
variations staying within ±20% of the uniform overpressure. In this case, the maximum overpressure
applied on a given wall may be estimated to be 1.2 times the peak of the average overpressure within the
room. The time after which uniform overpressure is reached depends on the volume of the room.
 Condition 2: If P/S is small, the pressure inside the room is determined by propagating pressure waves and
the local overpressure can vary considerably in space, from zero to twice the average overpressure on a
given wall.
For the first condition, calculations using the basic model may provide reasonable estimates for the mean pressure
in the installation room, whereas for the second condition, using CFD is recommended. For the particular cases
treated in [Douchin2013], the transition between both conditions occurred between 40 and 50 MW/m².
Irrespective of the power surface density, there are always high pressure peaks - locally concentrated and of short
duration (typically less than 50ms), which may be several times the room average overpressure.
The force resulting from the internal pressure can be calculated and considered in the sizing of walls and ceilings.
As these results are time-dependent, dynamic effects must be considered in accordance to building codes.
In the subsequent example, the pressure relief device of AIS switchgear is located on top of a vertical chimney
inside the switchgear. The gases are therefore directed to the ceiling. Right above the relief opening, on a small
surface area, the pressure reaches a high value for a short time 11 ms after initiation of the arc. On Figure 6-12, it
can be recognized that the maximum local peak is 15000 Pa, about 4 times the peak of the space-averaged
overpressure. Values always show the overpressure in the room.
Figure 6-13 displays the pressure inside the room 260 ms after fault initiation. At this time, the overpressure
decreases from the left side (3900 Pa near the faulty switchgear) to the right side (3200 Pa near the room
opening). The pressure is rather uniformly distributed around 3550 Pa with a variation of ± 10 %.
Page 84
Local peak pressure. Note

that at this instant the
overpressure is zero at the
other side of the ceiling.
Room pressure relief
Switchgear
Figure 6-12: Example of pressure field within a switchgear room at the beginning of the event (11 ms after
fault initiation); units are in Pa overpressure.
Switchgear pressure relief
Arc compartment
Figure 6-13: Example of pressure field within a switchgear room. Homogenous pressure stage (260 ms
after fault initiation); units are in Pa overpressure.
Page 85
6.3.2 Pressure withstand for different wall types
The pressure withstand capability of different types of wall

constructions has been analyzed by a number of authors, and
permitted overpressure values were proposed. This section
provides an overview of 5 papers that have covered this
subject:
 In [Pigler1976], a table of permitted pressure values

in installation housings dependent on the wall
material is given. This table is based on 24 cm thick
walls with carefully designed joints between adjoining Source:
TÜV Rheinland /
walls and to the ceiling on top. This is only general
Berlin-Brandenburg
information. To be more precise, wall dimensions and Figure 6-14: View of a building destroyed
the type of joints have to be specified in more detail. after internal arc fault in the switchgear room.
For example, a wall with a large area is less stiff than
a wall with small area from constructional point of view. The larger wall experiences the higher force at the
same room pressure due to the larger surface area. That means, walls of different surface areas - even
when they are built from the same material - react in different ways. With this in mind, Table 6-5 can be
only used as a first indication for the expected wall withstand.
Wall type Permitted overpressure in the room

bricks (solid bricks, perforated bricks, gas formed concrete) 3-10 mbar
reinforced brick wall 30 mbar
precast concrete members 50 mbar
cast in place concrete > 70 mbar
concrete space cell 130 mbar
Table 6-5: Pressure strength of walls [Pigler1976].
 In [Primus1999], values for permitted overpressures in switchgear rooms depend on the construction
(Table 6-6). They vary from 3 to 10 mbar for a brick wall without further specifications of the brick wall.
Wall type Permitted overpressure in the room
Brick wall (solid brick, perforated brick, gas concrete) 3 up to 10 mbar

Armored (reinforced) brick wall 25 mbar
Ready-mix concrete components 50 mbar
Job-mixed-concrete ≥ 70 mbar
Concrete room cell construction 160 mbar
Table 6-6: Pressure strength of walls [Primus1999].
Page 86
 In contrast to that, Table 6-7 from [Hollmann1999] provides overpressure values for brickwork (masonry)
with specification of the wall thickness. A permitted overpressure of 10 mbar is given here for non-
reinforced brick walls with 24 cm thickness.
Wall material / Thickness Allowed overpressure
Brick wall 24 cm (solid brick, perforated brick, gas concrete) approx. 10 mbar
Armored (reinforced) brick wall 24 cm, Lime sand stone wall
approx. 25 mbar
(Solid brick)
Table 6-7: Pressure strength of walls [Hollmann1999].
 The withstand values of flat walls from [Graf1987] are of interest. The paper presents values for sand-lime
brick with different grouts and distinguishes between explode (burst) and cracking values (Table 6-8). The
bursting pressure for sand-lime bricks is about 26 mbar with impulse stress, cracking occurs at approx.
25 mbar. The permitted stress values can be assumed to be between 3 and 25 mbar. For a binding
determination of permitted stress values, it is recommended to consult a structural engineer.
Wall Material / Thickness Pressure stress

Gas concrete 12 cm thickness exploded at approx. 15 mbar
1/2 perforated brick, cement big clefts at approx. 25 mbar
Sand-lime brick
grout exploded at approx. 60 mbar
Exploded by a pressure pulse of 65 ms
Sand-lime brick 1/2 solid brick, lime grout
at 26 mbar
Table 6-8: Pressure strength of walls [Graf1987].
 In [Pusch2007] the acceptable overpressure values for walls are provided (Table 6-9) similar to the values
given above.
Wall material/thickness Permitted overpressure values

Brick wall, 24 cm
up to 10 mbar
(solid brick, perforated brick, gas concrete)
Job-mixed-concrete, 24 cm up to 70 mbar
Ready-mixed concrete, concrete quality B 50, 14 cm ... 20 cm up to 160 mbar
Table 6-9: Pressure strength of walls [Pusch2007].
6.3.3 Pressure results derived using the basic calculation method
The basic calculation model can be considered when the pressure is almost homogenous within the room, i.e. for
2
high power surface density P/S (typically above 50 MW/m ). Applying the basic method from chapter 2, one has to
cope with the following potential difficulties:
 Complex Geometry: Between the arc compartment and the installation room there can be several
intermediate compartments and often one of them has more than one input or exhaust area (a busbar
Page 87
compartment for instance). In this case, it is not easy to choose the correct volumes and opening areas for
the basic model.
 Unreasonably high gas temperatures: When the pressure in the room is calculated for a long time
(sometimes up to one second), the gas temperature in the arc compartment might rise to unrealistically
high values.
 Numerical instabilities: When the pressure in the arc compartment is close to the external pressure, the
outflow function becomes sensitive to small variations of the gas density and might fluctuate with ever
increasing amplitudes.
In order to apply the basic method to a room pressure calculation, while being conservative in overestimating the
pressure peak, the following approaches can be followed:
1. The volume of the switchgear is ignored and the arc is taken as burning in the installation room i.e. only
one single volume is considered [Douchin2013]). However, the -factor determined for the enclosed
switchgear is applied by considering air as filling medium. This approach is not applicable for gas filled
switchgear.
2. The -factor is assumed decreasing with lower gas density in the arc compartment due to the ejection of
air or gas into the installation room (for more details see Annex A.4 on enhanced models). This considers
the enclosure effect of the switchgear as published in [Schumacher et al.].
With respect to approach 1, the following rules need to be considered. It has to be remarked that these rules have
been obtained from tests performed with air as filling gas and might be different for SF6.
 If and are known, these values should be applied. However, experience with different switchgear
arrangements [Douchin2013] has shown that -factors below 0.3 could lead to an under-estimate of the
room pressure. If the manufacturer specifies < 0.3, then 0.3 shall be used instead. Indeed, the here is
linked to the pressure rise in the room, not the one in the switchgear compartment. It is not necessarily the
same value. Furthermore, material evaporation is not modeled in the basic method. The use of a lower
than 0.3 could then lead to under-estimation of the room pressure.
 If and are not known, typical values may be considered. For , the typical range for air is [0,4; 0,6].
For , typical range for air is [300 V; 600 V] (refer to Section 2.3.5 and 2.3.6 for further details).
 The room volume is calculated substracting the volume of all installed switchgear.
 The discharge coefficient of the room relief area should be in the range of 0.6 to 1. The relief device is
modeled as a fully open area. If a grid or mesh exists, the relief area may be multiplied by a factor less than
1 in order to get the effective opening.
6.3.4 Pressure results derived from CFD simulation
The basic calculation method is not applicable when the power density P/S is too small i.e. below 40 MW/m².
Pressure then might not become uniform within the room. Local pressure peaks cannot be evaluated with the basic
model or the effect of multiple pressure relief openings and their locations.
Page 88
CFD is the right tool to cope with these issues. The switchgear room is modelled in this case with all relevant flow
resistances and volumes in a three dimensional model and introduced into a CFD tool.
 Flow blockades in the room can now be considered.

 The position of pressure relief openings and the geometry of the room can be correctly introduced.
 Dynamic processes like reflections, diffractions and interferences (wave theory) are automatically
computed.
In highly dynamic processes, the maximum pressure builds up in the corners of the room, because pressure waves
are reflected and generate local maxima similar to standing waves in a tube.
This is illustrated in the following example. In Figure 6-15 the pressure measurement locations are marked with red
dots and the pressure relief openings are marked in green. The arc fault is assumed to take place in the orange
panel with the yellow arrow. The installation room is a transformer room with big openings in the doors behind the
switchgear. The gas flows out of the switchgear, passes the first small openings (colored in green) and leaves the
building through the transformer doors.
By integrating the velocity over the opening area at the time of its maximum it is possible to get the maximum
volume flow over a pressure relief opening surface. This approach provides the necessary information to determine
the dimensions of the opening. Producers of pressure relief openings can use this value of maximum volume flow
to choose the right dimensions of pressure relief devices available from their range of products.
Figure 6-16 shows the overpressure over time in the point MP3 of the right hand ceiling corner in the room. As
mentioned, the CFD simulation takes care of the reflection, diffraction and interference of pressure waves. After
400 ms the pressure will be released through the openings shown in green color in Figure 6-15.
To get the value of the total force on the wall at a specific time it is necessary to integrate the local pressure values
obtained by CFD over the wall or ceiling. The maximum overpressure at the ceiling, as visible in Figure 6-17, is e.g.
16 mbar at time 20 ms (left figure as contour plot, right figure displays pressure distribution in red).
The average overpressure is given by the equation:
∫ ( ) 15300 N
= = = 1.5 mbar (6-9)
104 m2
This is only 10 % of the peak overpressure. This approach is helpful for engineers engaged in calculations to
evaluate the necessary stiffness of the switchgear room walls. It highlights the fact that most often, a local high
overpressure (limited in space) might not affect the sizing of the entire wall.
Page 89
Figure 6-15: Model of a switchgear room with Figure 6-16: Overpressure evolution over time on
pressure measurement points (MP). measuring point “MP3”.
Figure 6-17: Contour plot (left) and spatial pressure distribution (right) exerted on the wall; pressure scale
ranges from -740 Pa to 1580 Pa.
6.3.5 Pressure on building walls: comparison with test results (case A)
6.3.5.1 TESTED INSTALLATION AND RECORDED ARC POWER
Page 90
Room
Arcing compartment
Room
Opening
C Arc location
A
B
Pressure relief device Cylinder (modeled with a

square section in CFD)
Figure 6-18: Test arrangement.
In a simplified test arrangement, a three-phase arc was initiated in a small volume adapted from a bus-bar
compartment of switchgear. The hot gas was ejected through a cylinder, at the end of which a pressure relief
device (disk) was mounted. The gas (air) was then captured in an 8 m³ volume with an own relief opening.
Pressure rise in the arc compartment and adjacent room were detected by several pressure sensors denoted as A,
B and C in Figure 6-18:
 Sensor A: in the cylinder, near the relief device (arc compartment)

 Sensor B: in line with cylinder axis, on the room front wall (the one with the relief opening)
 Sensor C: on the room side wall, 35 cm from the rear wall
Dimensions:
Arcing compartment 0.3 m3
Cylinder volume and length 0.039 m3, 0.54 m
Room volume 8 m3 (2 m x 2 m x 2 m)
Room opening 0.3 m2
Pressure relief area 0.049 m2
The arc current was 20 kA. The arc power was recorded, as shown in Figure 6-19. It can be seen that, with an arc
voltage of 800 V, the power curve fits the simulated one for the first two cycles. Afterwards, the arc voltage
decreases to a lower value; however, for the pressure peaks recorded during the first two cycles, the value of 800
V is relevant. From the pressure curve of sensor A, a response pressure of 203 kPa of the relief device is deduced.
Page 91
In the following, the measured pressure curves were tried to simulate by a CFD tool as well as by the basic model
presented in this technical brochure.
eq(2-20) with Uarc = 800V

Arc power
measurement (eq (2-18))
1,0E+08
8,0E+07
6,0E+07
(W)
4,0E+07
2,0E+07
0,0E+00
0 0,02 0,04 0,06 0,08 0,1
Time (s)
Figure 6-19: Arc power, measurement versus calculation according to (2-18).
6.3.5.2 CFD RESULTS
Measurements from sensors A, B, C are shown in Figure 6-20, with the corresponding results derived from CFD
fitting the measured traces quite well. Though being in the same volume, the sensors at location B and C exhibit
different pressures. CFD indeed shows that pressure, temperature and velocity fields are not uniform within the
8 m³ room until 45 ms, which is the instant where the average pressure in the room achieves its maximum. At this
time, local pressures in the room are within the average value ± 20 %, and stay in this range afterwards (see Figure
6-23). A uniform pressure is also anticipated from the power density parameter P/S introduced in the section before,
which reaches a mean value of 120 MW/m².
It has to be remarked that the mismatch between measured and simulated pressure development in the arc
compartment might originate from an adverse position of sensor A. Since it is close to the relief valve, it rather
monitors the low pressure of the exhaust room after opening of the relief device instead of the high pressure of the
arc compartment.
Sensor A Sensor B Measurement

250 60
CFD
50
200 Measurement
Pressure (kPa)
Pressure (kPa)
Measurement 40
150 CFD
CFD 30
100 20
50 10
0 0
0 0,02 0,04 0,06 0,08 -10 0
0,1 0,02 0,04 0,06 0,08 0,1
-50 Time (s)
Time (s) -20
Page 92
Sensor C Measurement
40 Temperature field (K) – 21 ms
CFD
35
Measurement
30
CFD
Pressure (kPa)
25
20
15
10
5
0
-5 0 0,02 0,04 0,06 0,08 0,1
Time (s)
-10
Pressure field (Pa) – 25 ms Velocity field (m/s) – 25 ms
Figure 6-20: Temperature, pressure and velocity fields and derived temporal development of overpressure
at different sensor positions calculated with CFD.
6.3.5.3 BASIC METHOD RESULTS
 Welded cubic steel container

V2  Linear electrode arrangement

 Arc ignited between three electrodes
V1  Circular relief device
3
 Exhaust into a 8 m container
Side view of container.
Figure 6-21: Simplified design of test arrangement.
Page 93
For the basic model, the geometry of the test arrangement in Figure 6-18 is simplified and replaced by the volumes
sketched in Figure 6-21.
Figure 6-22 shows that for the arc compartment (sensor A), a reasonable fit of the rising slope of the pressure
development is achieved with = 0.22. Though the discharge coefficient is taken as ( ) = 1 allowing for the largest
opening area, the pressure drop exhibits a similar mismatch the measurement as recognized from the CFD
calculation.
Fit to pressure rise and peak value can also be achieved when inserting = 0.37 at a lower response pressure of
160 kPa (Figure 6-24). Since the moment of relief disc opening was not detected in the test, such a variation is
possible.
Sensor A
2,50
2,00 Simplified method with kp=0,22

Pressure (bar relative)
measurement
1,50
1,00
0,50
0,00
0 0,02 0,04 0,06 0,08 0,1
Time (s)
Figure 6-22: Measured pressure in the arc compartment and calculation with the basic method, best fit
with = 0.22.
As discussed in 6.3.5.2, the two measurements at points B and C are different showing that the pressure is not
uniform in space within the room. However, they become similar after 45 ms. The average pressure in the 8 m³
room calculated with CFD is used as reference for the overpressure calculated with the basic model. The discharge
coefficient of the relief opening of the exhaust room is assumed to be 0.7.
The basic method underestimates the pressure peak, when fitting with = 0.22: it is 10 kPa and therefore 45 %
below the reference value (18.3 kPa). If the -factor is increased to 0.37, the overpressure in the exhaust room
matches the value measured by sensor C. The result is shown in Figure 6-23. It is noticed that after 45 ms the
basic method delivers an almost constant overpressure in the room, whereas measurement and CFD simulation
give a declining pressure.
Page 94
50
Sensor B (CFD result)
40 Sensor C (CFD result)
room average (CFD result)
Simplified method with alfa =0,8
30
Pressure (kPa)
20
10
0
0 0,02 0,04 0,06 0,08 0,1
-10 Time(s)
Figure 6-23: Comparison of overpressures in the 8 m3 room calculated with CFD and with the basic
method.
6.3.5.4 ENHANCED METHOD WITH PRESSURE-DEPENDENT -FACTOR
A declining pressure in the exhaust room as measured can be simulated with an enhancement of the basic model
with respect to the -factor. In dependence of the decreasing gas density in the arc compartment after opening of
the pressure relief device, the -factor is reduced in time decreasing the exhaust of hot air into the 8 m³ room. This
measure leads to a faster reduction of the pressure in the exhaust room [Dullni2013], whereas the maximum of the
average pressure is hardly affected (sensor C in Figure 6-25.). However, the enhanced method still cannot reflect
the local pressure peak recorded by sensor B (Figure 6-24).
Page 95
R
el.
pr
es
s
ur
Figure 6-24: Pressure measured by sensors A (upper traces) and B (lower traces) and calculated with the
enhanced method; fit with = 0.37.
R
el.
pr
es
s
ur
Figure 6-25: Pressure measured by sensor C and calculated with the enhanced method.
6.3.5.5 CONCLUSIONS FOR CASE A
For the first 40 ms after pressure relief opening, the distribution of pressure is not uniform, and pressure peaks
occur locally (see sensor B for example in Figure 6-20). Afterwards, the pressure becomes homogenous with a
spatial variation within ± 20 % of the average room pressure. 45 ms is the instant when the room average pressure
reaches its maximum value.
For the different calculation methods, one can conclude that:
Page 96
 CFD: the pressure development within the switchgear room is properly modelled
 Basic and enhanced methods: before 45 ms, the pressure is not uniform, and local pressure peaks (sensor
B) cannot be reproduced. After 45 ms, basic method and enhanced methods may provide the average
overpressure with the possibility to underestimate the pressure considerably.
 For all methods: results are sensitive to the response pressure of the relief device, the right choice of
and discharge coefficients.
6.3.6 Pressure on building walls: comparison with test results (case B)
An installation room with a volume of 16 m³, which could be extended to a volume of 40 m3 have been tested with
the same air insulated switchgear. The arc current was always 20 kA. In this section, pressure measurements are
shown and compared with results from CFD calculation.
The switchgear was equipped with an arc energy absorber installed in one of the pressure relief openings of the
switchgear (the one with an area of 0.1 m² in Figure 6-27). Such a device can for example be modeled with CFD.
6.3.6.1 16 m 3 ROOM
Room opening
Figure 6-26: 16 m3 AIS room installation and pressure sensor locations.
Page 97
Figure 6-27: 16 m3 AIS room: main data of relief openings and volumes.
Room pressure measurement signals are filtered with a low-pass 200 Hz filter (Yi corresponding to Pi). The three
filtered signals within the room are shown in Figure 6-28. The pressure in volume V4 was detected by sensor P1.
R
el.
pr
es
s
ur
Time [s]
Figure 6-28: Pressure measurement within the 16 m3 AIS room.

The measurements are within an average ± 20 % at the instant of the maximum pressure. In general, P4 is lower
than the two others. CFD results are presented in Figure 6-29.
Page 98
Figure 6-29: Overpressure for the 16 m3 room measured by different sensors and calculated by CFD; eta is
the heat absorption efficiency of the absorber.
Within the room (P2 to P4), CFD overestimates the pressure, though the fit to the pressure inside the switchgear
(P1) is perfect for the first 50 ms. It is remarkable that the measured relative pressure in the room exhibits a
negative excursion. This might be explained by an elastic oscillation of a wall of the room excited by the first
pressure pulse though no proof can be provided afterwards. An increase of the room volume by only 1 %, for
example, is sufficient to reduce the relative pressure in the room by 10 mbar. From Figure 6-29, an oscillation of the
measured pressure around the calculated pressure curve is recognized.
6.3.6.2 40 m 3 ROOM
By shifting an internal wall, the installation room could be enlarged to a volume of 40 m³. The three filtered
measurement signals of the pressure within the 40 m³ room are shown in Figure 6-30. The overpressure values are
within an average ± 20 % after 110 ms. Until then, the pressure distribution within the room is less uniform (average
± 50 %). CFD results are given in Figure 6-31. Within the room (sensors P2 to P4), the pressure calculated by CFD
fits better to the measurements than for the 16 m³ room. It seems that in this case no pressure oscillation was
excited. CFD overestimates the pressure by 10 to 20 %.
Page 99
R
el.
pr
es
s
ur
Time [s]
Figure 6-30: Pressure measurements within the 40 m3 AIS room.
Figure 6-31: Overpressure for the 40 m3 room measured by different sensors and calculated by CFD.
6.3.6.3 CONCLUSIONS FOR CASE B
This case shows again that the pressure distribution within the installation room can be far from uniform until a
given instant. CFD results are close to measurements, and conservative.
Page 100
6.3.7 Important parameter for the mechanical stress on walls: The size of
pressure relief openings
According to IEC Technical Report 61936-1, the user is obliged to take the pressure rise due to an arc fault into
account in the construction of buildings.
Relief openings of the room most often have no influence on local pressure peaks no matter how large the opening
area is. Fortunately, these peaks are limited in space and time, and probably may not have much influence on the
civil design of the room. On the other hand, when pressure becomes uniform in space after some delay time, the
peak value of this uniform pressure is controlled by the size of the room openings. The relevant delay time may be
estimated from the time a pressure wave crosses the length of the room assuming a reasonable speed of sound.
Basically: the larger the opening area, the smaller the pressure inside the room averaged over the volume. By
appropriate dimensioning of such relief areas, the stress on the building generated by an internal arc can then be
minimized.
When increasing the size of the room openings, the uniform pressure in the room decreases as illustrated in Figure
6-32. For the area of the room pressure relief openings, a value should be chosen lying in the knee of the curve,
which is most effective.
1,00
per unit of the closed room pressure peak
0,90
0,80
0,70
0,60
0,50
0,40
0,30
0,20
0,10
0,00
Room opening area [m2]
Figure 6-32: Room pressure function of room opening area.
If the room is closed (without any opening), the pressure inside the room will increase continuously during the arc
fault duration reaching a high value with possible rupture of the room. Therefore, such installation rooms should
only be used when they are so large that within the rated arc duration the pressure rise even for a freely burning
arc inside the room does not reach critical stress values.
Page 101
6.3.8 Summary
Pressure on building walls as a result of an internal arc in switchgear is a critical parameter for building
construction. The force resulting from the internal pressure can be calculated and considered in the sizing of walls
and ceilings. Tables in this section provide guidance for the wall design, but local building codes must be
considered for detailed structural design.
There are two conditions, based on which the applicable calculation method is determined.
The first is a condition, where the ratio of mean arc power, P, and total area of room openings, S, is large (see
6.3.1). In this case, the room pressure becomes uniform sometime after the occurrence of the internal arc. Still, the
force resulting from the pressure is a dynamic load on walls and ceilings and requires consideration of relevant
building codes. Though also under these conditions the CFD method is recommended, the basic calculation model
may be used to calculate the room average pressure ignoring local pressure peaks at the beginning of the arc fault.
The enhancement of the basic model with respect to a density dependent -factor helps simulating the
anticipated drop in overpressure and prevents numerical instabilities.
In the second condition, where P/S is small, the pressure inside the room is determined by propagating pressure
waves and the local pressure can vary considerably in space. CFD tools are required here to calculate the whole
temporal and spatial development of pressure.
It should be noted that the Working Group had only limited access to test results of pressure rise in installation
rooms, because such pressure measurements are normally not possible during internal arc tests. Simulated
installation rooms in test labs are typically not enclosed. Therefore, the accuracy of the results could not be
determined here due to the limited number of cases.
6.4 Burn-th rough

In the previous sections the effect of the pressure on the switch enclosure and building walls has been seen. In this
section the third effect of the internal arc is looked at - the “burn-through” effect. This effect is caused by the arc
which can burn on a surface of the metallic enclosure (like a switchgear wall or panel, or GIS bus duct). This melts
and then punctures the wall.
First the causes and consequences of burn-through are examined. Then the burn-through time for both aluminum
and steel enclosures is evaluated. Lastly, burn-through differences between SF6- and air-filled switchgear are
looked at.
6.4.1 Causes and consequences of burn-through
The burn-through effect is caused by an arc which can burn on a surface of a metallic enclosure (like a switchgear
wall or panel or GIS bus duct). In the case of three-phase enclosures (all three phases in the same volume such as
MV GIS), the arc can be initiated by a phase to ground fault but it will generally rapidly evolve to a phase to phase
fault due to the higher electrical stress between phases compared to the phase to earth distance . In the case of
single-phase enclosed devices (each phase in a separate volume such as HV GIS), the arc will always be between
phase and ground. This grounded surface might be exposed to the burn-through. This section focuses only on the
burn-through in the single-phase enclosed constructions.
Page 102
The arc will produce a pressure build-up within the faulty volume and an erosion process will take place at the arc
root location. In relation to the enclosure, a portion of the total arc power transfers to the enclosure wall and heats
the wall material so that it changes from solid to liquid and finally to gas (Figure 6-33) which could finally produce
the burn-through if the arc stays at the same place for a sufficient time (Figure 6-34).
The effects cannot be fully predicted by simple thermal conduction models since too many parameters play an
important role. It is also difficult to predict the movement of the arc and to determine the energy input at the arc
root.
The time to burn-through depends mainly on the current density, the thickness of the enclosure wall, the type of
material and the duration of the fault. The erosion combined with pressure stress can finally cause the burn-through
of the enclosure wall.
In the case of burn-through the safety of operators or persons near the switchgear cannot be ensured since hot
gases and metal droplets are blown out of the hole in an uncontrolled manner. This is an unlikely event due to the
combination of the low rate of burn-through occurrence and the minimum presence of personnel near the GIS.
Statistical analysis done by both a utility and a manufacturer reported in [Chu1982] gave an estimation of the
probability of personnel being injured from direct impact by a burn-through of between 1.5 E-5 and 3.3 E-5 per
substation and year.
Evaporation heat
Melting heat
Q/m
Figure 6-33: Temperature behavior [T] of electrode material versus heat energy [Q] per mass [m]
(schematically).
Figure 6-34 presents two typical enclosure burn-through scenarios. In the left one, it is likely that the erosion
process due to the arc root was predominant. The right photograph indicates that the pressure seems to have torn
the wall of the enclosure in the region of the arc possibly associated with a reduction in the mechanical strength of
the material due to the high temperature of the material as a result of arc heating.
Page 103
Figure 6-34: Typical views of a HV GIS enclosure in the case of burn-through.
6.4.2 Evaluation of the burn-through time
Available studies to evaluate the burn-through time are listed in Table 6-10.
Rod to plane Real GIS arrangement

Arrangement
Analytical [Kuwahara1982] [Babusci1998]

[Kolbe1975] [Kuwahara1982]
Experimental [Kolbe1975] [Lutz1983]

[Petterson1977] [Bernard1982]
[Trinh1992]
Table 6-10: Some literature references dealing with the burn-through phenomena.
Figure 6-35: Movement of an arc in axial and azimuth direction [Babusci 1998].
4
Once the arc is established, unbalanced magnetic forces drive the arc in both the axial and azimuthal directions ,
see Figure 6-35. The axial motion of the arc along the conductor stops once the arc reaches a barrier (such as an
4
Chu [1982] mentions that a stationary arc can burn in HV GIS if the fault is supplied from two sides (two sources).
Page 104
insulating spacer). At that point the only motion driven by the magnetic forces is an erratic azimuthal motion. Both
motions are explained by Boeck and Krüger [Boeck1992].
The results of many experiments show that the erosion of the enclosure shell is minimal while the arc is moving
axially and the burn-through occurs only during the time the arc stays axially stagnant. This is due to the fact that
the axially stagnant arc is concentrated on one area of the enclosure for the whole duration.
The total time from the instant of the arc ignition to the enclosure burn-through could be divided into two parts:
= + (6-10)
where is the total time from the instant of the arc ignition to the enclosure burn-through, is the time during
which the arc is traveling axially along the conductor, and is the time during which the arc stays axially stationary
and may cause a burn-through.
could be analytically determined from a model of the axial arc velocity assuming columnar arc geometry.
According to Lutz and Chu in [Lutz1983] this leads to erroneous results. can be better evaluated by experimental
formulas given as following:
0.28
= [Chu1985] (6-11)
where is the axial arc velocity [m/s], the arc current [kA rms], the electrode spacing [m] and the ratio of the
gas density to air density at one atmospheric pressure. can be drawn by dividing the distance from arc initiation
to its final axial stationary position by the axial velocity.
Table 6-11 shows the comparison of arc velocities between calculations with equation (6-11) above and two test
cases of real GIS arrangement carried out by a manufacturer. SF6 gas pressure, current, and electrode spacing
were 6 bar, 40 kArms, and 122.5 mm, respectively. As seen in the table, measured arc velocities are faster than
those calculated from equation (6-11), showing 20 % variation.
Velocity of arcs [m/s]
Test shot
Calculated with
Test result
equation (6-11)
#1 18.8 (118 %)
15.9 (100 %)
#2 19.8 (124 %)
Table 6-11: Comparison between calculated arc velocity and test results in a real GIS arrangement.
To determine one can use either an analytical or an experimental approach.
In the analytical approach, the models consider a small area of material heated up by the arc energy and the
propagation of this molten zone by conduction and radiation. Material liquefaction and vaporization should be
considered in the model but it was proven that the main part of the erosion occurs in the liquid phase [Kolbe1975].
Page 105
The models consider that the power is transmitted to the electrode or material through a circular spot
[Babusci1998], [Chabrerie1993].
This analytical approach does not reflect the real situation because it ignores azimuthal motion of the arc. The
azimuthal motion significantly increases the time to burn-through. However, the models are an instructive method
of studying the influence of parameters such as current, enclosure thickness, and material.
The experimental approach to determining is made using rod-to-plane tests and coaxial arrangement tests.
6.4.2.1 ROD TO PLANE TESTS
Lutz and Chu [Lutz1983] proposed the use of the following formula based on the Kolbe proposal [Kolbe1975] for
aluminum plate based on their rod to plane tests:
282 39.1
( , ℎ) = .
∙ ℎ− .
(6-12)
where is the current in kA rms, ℎ the plate thickness in mm and the time in ms. Authors found good agreement
between measured times and calculated times for their modification of Kolbe’s formula.
Pettersson proposed another formula for determining the burn-through time for an aluminum plate:
.
540 ∙ ℎ
( , ℎ) = (6-13)
Figure 6-36 compares the results obtained with the two equations, for two different thicknesses. One can see that
for lower currents the differences between the two equations are larger but that they become smaller at higher
currents.
In the case of a GIS compartment with a coaxial arrangement, one can observe greater burn-through times than
those for rod to plane geometry because the azimuthal motion of the arc produces a less intensive vaporization of
the enclosure wall. For aluminum, the burn-through time could be between 1 and 2.5 times the burn-through time
evaluated for rod to plane [Lutz1983]. Therefore, the burn-through time cannot be accurately predicted from rod-to-
plane burn-through equations and coaxial tests in a real GIS configuration are needed.
Page 106
Figure 6-36: Burn-through estimation for a rod to plane arrangement for aluminum (comparison between
Kolbe and Pettersson results).
6.4. 2.2 COAXIAL TESTS
Standards require that the arc be initiated in a location where it is likely to be axially stationary from the beginning
of the arc ignition. Therefore it is only important to consider for apparatus design but tests show a large
scattering of the results due to the unpredictable azimuthal motion of the arc.
Based on an experimental approach, G. Bernard proposed the following equation for a coaxial assembly made of
aluminum enclosures [Bernard1982]:
.
87.4 ∙ ℎ
( , ℎ) = .
(6-14)
Where is the r.m.s current in kA, ℎ the plate thickness in mm and the time in ms.
Results from this formula can be compared with manufacturer data for single-phase aluminum enclosed GIS from
245 kV up to 550 kV. This comparison is shown in Table 6-12. The times are not total burn-through times ( ) but
only the times where the arc is axially stationary ( ). These times were extracted from the raw test results.
h[mm] 5 6 7 8 9 10 12 14
I [kA]
15 20 x x x 16 x x x
32 34 9 x x -6 x x x
40 22 x x x x x x x
50 -3 x x x x x 16 x
63 49 x 40 x x -3 6 -18
Table 6-12: Differences between burn-through times calculated with Bernard’s formula and results from
manufacturers experiments [%].
Page 107
A cross means that there is no relevant test data available for this combination of current and enclosure thickness
at the time. When several values were available for the same combination of current and thickness, only the
shortest time value is considered in the table.
In most cases, the Bernard formula shows good agreement with test results as can be seen in Table 6-12 and
Figure 6-37.
For 63 kA tests and a 14 mm enclosure, one can see an important difference between the test results and the
Bernard formula where the Bernard formula predicts a much longer time (-18 %= -104 ms). This could lead to an
overestimate of the actual performance of the enclosure. The reasons for this discrepancy are not clear. One
possible explanation could be that the pressure inside the compartment was so high as to have torn a hole in the
enclosure, already weakened by the arc erosion (see the right hand picture of Figure 6-34). Bernard’s formula for
the experimental approach and the analytical model presented in publication [Babusci1998] are only valid if the
predominant factor is the erosion of the arc and are not valid if the remaining thickness cannot withstand the
pressure inside the enclosure.
Figure 6-37: Comparison between test results and time calculated with Bernard's formula for burn-through
time estimation at 50 kA and 63 kA for aluminum enclosures.
6.4.3 Aluminum versus steel enclosures
Steel enclosures experience longer times than aluminum enclosures. This may be explained by steel having a
higher heat capacity (product of density and specific heat) than aluminum, thus requiring more energy to melt the
material, as illustrated in Table 6-13.
Material Density Melting Temperature Heat capacity per unit
Page 108
volume (density x
specific heat)
3 3
Steel 7.85 g/cm 1400 ºC 3.53 J/(cm K)
Aluminum 2.70 g/cm3 660 ºC 2.43 J/(cm3 K)
Table 6-13: Properties of steel and aluminum.
Using the equation given by Babusci et al [Babusci1998]
ℎ
= ∙ (6-15)
where is the time to burn-through, ℎ is the thickness of the enclosure and is the arc current, is a factor which
depends on the material, the time to burn-through for steel is about 4 times longer that for aluminum enclosures for
the same values of enclosure thickness and current.
= ≈ 4.3 (6-16)
To summarize:
The generic formula for burn-through can be written as:
ℎ
= (6-17)
where is time to enclosure burn-through, ℎ is enclosure thickness (mm), is arc current (kA), is a constant
characteristic of the enclosure, and is a constant characteristic of the arc current.
The time to burn-through increases with the increase of the enclosure thickness ℎ and the decrease of the arc
current . It will be considerably larger for steel than aluminum. Table 6-14 summarizes the different constants.
Although none of the formulae directly include any gas characteristics (SF6 or air) the constant
could include gas properties based on the theoretical derivations of Babusci [Babusci1998]. Another effect of the
gas on burn-through is the different pressure that SF6 or air can develop during arcing inside the compartment.
The principles for the GIS stationary arc examples can also apply to medium voltage switchgear but care should be
taken when using the specific formulae and parameters.
Authors/Reference Material Arc
Page 109
condition
Moving,
[Babusci1998] Aluminum coaxial 173 2 1
arrangement
Moving,
[Bernard1982] Aluminum coaxial 87.4 1.77 0.67
arrangement
Moving,
[Chu1982] Aluminum coaxial 150 2 1
arrangement
Moving,
[Trinh1989] Aluminum coaxial 179 1 0.73
arrangement
Stagnant,
[Petterson1977] Aluminum 540 1 1
Rod to plane
Moving,
[Babusci1998] Steel coaxial 750 2 1
arrangement
Table 6-14: Comparing properties of steel vs. aluminum.
6.4.4 Replacing SF 6 by air in type tests for the burn-through
No reference has been found for the comparison of burn-through for SF6 and for air. Nevertheless some
differences in the fundamental characteristics of arcs in SF6 and air could be extrapolated into impacts on burn-
through behavior. These are presented in Table 6-15 which compares qualitatively the expected burn-through
times for SF6 and air and are illustrated by Figure 6-38.
From this qualitative approach, it can be seen that a large number of parameters indicate that shorter burn-through
times take place in SF6 rather than in air. It should be noted that this is a theoretical hypothesis (for a more detailed
understanding of these influencing factors, refer to ANNEX G); there is not enough test data or experiment
available today to validate this claim.
Figure 6-38: Phenomena taking place at the arc foot root. (The numbers correspond to those of the
influencing factors in Table 6-15.)
Page 110
Quantitative comparison of the Qualitative comparison of resulting burn-

Influencing factors influencing
factors for SF6 and air through times, , for SF6 and air
(1) Electrode fall voltage SF6  Air (SF6) ≈ (Air)
Input (2) Heat of chemical reaction

SF6 > Air (SF6) < (Air)
power (exothermic energy)
(3) Radius of arc root
(concentration of input power) SF6  Air (SF6) ≈ (Air)
(4) Heat transfer from arc to gas

SF6 > Air (SF6) > (Air)**
(conduction)**
(5) Heat transfer from arc to gas
(convection)**
(6) Heat transfer within metal
SF6 = Air (SF6) = (Air)
Power wall(s)
loss (7) Heat transfer from enclosure
to gas (conduction)**
(8) Heat transfer from enclosure
SF6 > Air tb (SF6) > tb (Air)**
to gas (convection)**
(9) Heat consumption by metal
SF6 = Air tb (SF6) = tb (Air)
evaporation
(10) Axial velocity - Contribution
SF6 < Air tb (SF6) > tb (Air) *
Arc “ta” *
motion
(11) Azimuthal velocity SF6 < Air tb (SF6) < tb (Air)
(12) Drag forces at arc foot SF6  Air tb (SF6)  tb (Air)
(13) Pressure exerted on the

Other SF6 > Air tb (SF6) < tb (Air)
tank wall after disk opening
(14) Pressure exerted on the
tank wall *** SF6 < Air tb (SF6) > tb (Air) ***
(peak pressure)
*: IEC standard 62271-203 (Ed. 2.0, 2011) requires the initiation of the arc close to its final position for the test so “ta”
is close to 0. Therefore consideration of the axial velocity is of minor importance for test conditions.
**: These influences are negligible.
***: Burn-through is likely to happen some tenth of milliseconds after the burst disk opening, when the higher peak
pressure in air has already vanished. For that reason this effect is of minor importance compared to (13).
Table 6-15: Factors influencing the burn-trough comparatively for SF6 and air.
6.5 Summary
In this section three effects of the internal arc on structures have been analyzed:
Page 111
 Mechanical Stress on switchgear due to overpressure

 Mechanical stress on building walls due to overpressure
 Burn-through
The effects of an internal arc on mechanical structures can be predicted using analytical formulas or complex
mathematical models. Mechanical deformation (and von Mises stresses) of the switchgear enclosure can be
reasonably accurately calculated with off-the-shelf FEA software. With the basic model, the average pressure in the
room can be determined (with some limitations that need to be taken into account), and that value can be used to
dimension the room walls in Air-Insulated Substations. But for Gas Insulated Substations, the simplified method
may lead to underestimation of the pressure. With CFD calculations the localized time-dependent pressure is
obtained, which can be used to determine the size and placement of pressure relief openings in the building.
In addition it has been shown how the burn-through time can be evaluated, using different empirical formulae.
Furthermore, test results and calculations are compared. It should be noted that a large scatter in the test results is
possible since the energy released in one test may be different in another test depending on the erratic arc
movement for example. This scatter should be considered in order to allow a sufficient safety margin in relation to
the calculation results.
Lastly these tools can be used to assess the different pressure rise and mechanical stress on switch enclosures
when SF6 gas is replaced with air. With respect to the burn-through behavior, the equivalence when replacing SF6
with air is not that obvious. From the qualitative approach developed in 6.4.4, the burn-through time in SF6 is likely
to be shorter than in air. However, there is not enough test data and experiments available today to validate this
statement.
REFERENCES:
[Babusci1998]: Babusci, G. and Colombo, E. and Speziali, R. and Aldrovandi, G. and Bergmann, R. and
Lissandrin, M. and Cordioli, G. and Piazza, C.” Assessment of the Behavior of gas-insulated Electrical Components
in the Presence of Internal Arc”, 1998.
[Bernard1982]: Bernard,G , “Electrical faults mastery in high voltage SF6 insulated substations”, Revue générale de
l'électricité (RGE) EDF, Vol 4/82, 1982.
[Boeck1992]: Boeck, W. A. and Kruger, K.,” Arc motion and burn through in GIS”, IEEE Transactions on Power
Delivery,Vol.7, p 254—261,1992.
[Chabrerie1993]: Chabrerie J.P., Devautour J., Teste Ph., “A Numerical Model for Thermal Processes in an
Electrode Submitted to an Arc in Air and Its Experimental Verification”, IEEE Transactions on Components,
Hybrids, and Manufacturing Technology, Vol 16, No 4, 1993.
[Chu1980]: Chu, F.Y., Law, C.K., Boggs, S.A. "Dynamics of Power Arcs in CO-Axial Electrode Geometry",
GaseousDielectrics 11, March 1980.
[Chu1982]: Chu, F.; Ford, G. and Law, C. “Estimation of Burn-Through Probability in SF6 Insulated Substations”,
Power Apparatus and Systems, IEEE Transactions, PAS-101, 1391-1399, 1982.
[Chu1985]: Chu, F.Y and Lutz, F. and Braun, J.M and Stuckless, H.A, “Effects of power arc faults in gas-insulated
substations”, CIGRE–Symposium, Brussels, Contribution 340-04, 1985.
Page 112
[Douchin2013]: Douchin J., Gentils F., “Pressure rise in switchgear rooms in case of internal arc in AIS MV
switchboards; importance of room design and simplified calculation method”, CIRED, 22nd International Conference
on Electricity Distribution, Stockholm, 2013.
[Dullni1994]: E. Dullni, M. Schumacher, G. Pietsch, “Pressure rise in a switchroom due to an internal arc in a
switchboard”, 6th Int. Symposium on Short Circuit Currents in Power Systems, Liege, 1994.
[Dullni2013]: Dullni, E., “Feeling the pressure”, ABB Review 3/13, pp. 54-59, 2013.
[Graf1987]: D. Graf, Druckentwicklung beim Störlichtbogenfall in Schaltanlagen und in Schaltanlagengebäuden,

Referat auf Informationstagung der Betonbau GmbH, 1987.
[Hollmann1999]: Hollmann and Driescher (Störlichtbogenbeanspruchung und -begrenzung in Mittelspannungs-

Schaltanlagen, Handbuch zur VDE-Seminarveranstaltung, VDE Bezirk Kurpfalz, 1999.
[Kuwahara1982]: Kuwahara, H and Yoshinaga, K. and Sakuma, S and Yamauchi, T. and Miyamoto, T.,”
Fundamental investigation on internal arcs in SF 6 gas filled enclosure”, Mitsubishi Electric Corporation, 1982
[Kolbe1975]: “Durchschmelzen von ebenen Metallplattenelektroden durch Hochstromlichtbögen in SF 6”, ETZ-A

Bd.96, 1975.
[Lutz1983]: Lutz, F. and Chu, F. Y., “Burn-through of GIS Enclosure Due To Power Fault Arcs”, EEE Transactions
on Power Apparatus and Systems, 1983.
[Petterson1977]: K.G. Pettersson, E.V. Granstroem, “Design of Gas-Insulated Substations with Respect to Internal
Arcing”, Report IEE, Publ. 157, 1977.
[Pigler1976]: F. Pigler, “Overpressure in Switchrooms or Substations resulting from an internal arc in electrical
equipement“, Energiewirtschaftliche Tagesfragen (EWT), journal 3, 1976.
[Primus1999]: I.-F. Primus, Störlichtbogenfeste Gebäudekonzepte – geprüfte Konstruktionsprinzipien – Nachweise

durch Störlichtbogenprügungen, Handbuch zur VDE-Seminarveranstaltung, VDE Bezirk Kurpfalz, März 1999.
[Pusch2007]: P. Pusch, Schaltberechtigung für Elektrofachkräfte und befähigte Personen, 5. Aufl. 2007.
[Trinh1992]: Trinh, N.G., “Risk of burn-through-a quantitative assessment of the capability of gas-insulated
equipment to withstand internal arcs”, IEEE Transactions on Power Delivery, Vol 7, p225-236, 1992.
[Trinh1989]: Trinh, N.G., “Evaluation of the Risk of Burn through due to Internal Arc of Gas-Insulated Equipment”,
CEA Trans., March 1989.
Page 113
7 INTERNAL ARC SIMULATION REVIEW

7.1 In troduction
Current standards such as the IEC 62271-203 (Ed. 2.0, 2011) for HV metal-enclosed switchgear allow the
replacement of internal arc tests by calculation results if there is mutual agreement between the manufacturer and
user. Evidence of withstand capability against internal arc faults shall be demonstrated by the manufacturer when
required by the user. Evidence can consist of a test or calculations based on test results performed on a similar
arrangement or on a combination of both.
Other applicable standards such as the IEC 62271-201 (Ed. 1.0, 2006) and IEC 62271-200 (Ed. 2.0, 2011) which
deal with switchgear below 52 kV insulating enclosed and metal-enclosed respectively do allow for the replacement
of SF6 by air in the internal arc withstand tests. However, they do warn that if tests are carried out with air instead of
SF6, the pressure rise will be different and care should be exercised in the interpretation of the test results. Some
utilities require internal arc withstand tests for all HV switchgear designs where SF6 is replaced by air considering
such tests to be more severe.
These international standards leave the subject open to the agreement between manufacturer and user. However
there is in many cases a lack of specific criteria from the user side and it can be a subjective issue for the user to
accept any argument from the supplier or none at all. Therefore there is a need to provide some guidance in how
to interpret and validate a justification of the internal arc withstand capability of switchgear based on calculations.
A similar problem occurred in the case of replacing short-circuit testing of Power Transformers by calculations. It is
expensive to perform such a type test and it is accepted that the test may be replaced by a theoretical evaluation of
the ability of the transformer to withstand the dynamic effects of a short circuit. Annex A of IEC 60076-5 (Ed. 3.0,
2006) provides guidelines to perform such an evaluation. Even if the dynamic effects of short-circuit currents in
transformers and internal arcs in switchgear are different physical phenomena, it is possible to simulate both
despite a statistical spread of test results. A similar structure in the theoretical evaluation can therefore be
established for the switchgear internal arc tests.
Similarly to the short-circuit testing of Power Transformers, the switchgear user might accept replacement of the
internal arc withstand test of the specific switchgear by a design analysis carried out by the manufacturer based on
tests of a similar design.
In this review, the calculated withstand capability of switchgear against the stresses caused by internal arc can be
evaluated by following two methods.
1. By comparison with reference switchgear which has already been tested (Section 7.3)
2. By check against the manufacturer design rules for internal arc withstand. These rules must be based on
previous test results. (Section 7.4)
Section 7.2 will first explain the information to be provided before the review takes place.
Page 114
7.2 In forma tion for the design review

Before performing the review, a complete list of the calculation assumptions, input data and results should be
provided. This information should be sufficient to be able to reproduce the calculation results, following similar
calculation methods described in this brochure, by the user or an independent body if required. In order to be able
to review these calculations, the manufacturer should provide a list of the input parameters and assumptions used
in the calculations as shown in Table 7-1.
Mechanical
Electric Geometric Media
characteristics
Response pressure of the

relief device
Fault current Volume of arcing and exhaust
Type of gas (Air, SF6, Mixture)
(rms,peak) compartments if applicable Bursting pressure of the
Filling pressure arcing enclosure
Rated voltage Relief opening area
, Material of enclosure
Circuit (3Ø/1Ø) Exhaust compartment openings
-factor Wall thickness of the
Arc voltage Phase to phase distance
Ambient temperature enclosure
Arc duration Phase to ground distance
Material of conductor Manufacturing type of the
Type of fault Inner diameter of GIS enclosure
enclosure (casting, plate
welding, etc.)
Table 7-1: Design input for internal arc withstand simulation review.
It is important to identify test objects by unique numbers, drawings, test reports, photographs, etc. Furthermore, the
objects or designs that are subjected to calculations and/or simulations of the effects of internal arcing have to be
identified as well in reproducible and transparent reports.
7.3 Comparison of actual and re ference switchgear using design

paramete rs
The results of a test made in similar switchgear can be used with certain margins to forecast the behavior of the
switchgear under evaluation. The validity of the results of a test carried out in a functional unit of a particular metal
enclosed design of switchgear might be extended to another one provided that the original test was carried out
under more onerous conditions and this other functional unit can be considered as similar to the tested one in the
following aspects:
 structure and strength of the enclosure.

 architecture of the partition (IEC 62271-200).
 performance of the pressure relief device, if any.
 insulation system.
 physical influences (pressure rise, gas flow and and/or burn-through).
Table 7-2 and Table 7-3 can be used to determine the “similar” condition between two switchgears. Due to the
differences in pressure rise between typical MV metal-enclosed switchgear and internal arc withstand of HV GIS,
Page 115
two different tables are shown. Table 7-2 provides a comparison criteria between internal arc withstand in two MV-
compartments within a family of MV (rated below 52 kV) switchgear. Table 7-3 shows the same criteria for a single
High Voltage compartment within a family of HV (Rated above 52 kV) gas insulated switchgear. In these tables the
validation criterion is based on the comparison of the design parameter with the reference switchgear. For
instances ≤ sign with phase to phase clearance means that the phase to phase clearance in the calculated
switchgear should be equal or less than that used in the reference case.
The switchgear under evaluation can be compared with the reference switchgear by contrasting the calculation
results:
1. Pressure rise in the compartments of the switchgear and exhaust volumes

2. Mechanical stresses in the enclosure
3. Burn-through time of the enclosure
There are cases where the switchgear has been tested, but the gas exhaust has a special design due to client
needs or request (gas ducts, pressure relief…). As an example, consider that a particular switchgear passed the
test with a given gas exhaust system. Due to client requirements, the special gas exhaust conditions will be more
onerous than the one used in the test. A relieving condition is that the short-circuit current is lower than the one
used in the test. In these cases the comparison criteria which could be used are as follows:
 First, a simulation of the tested arrangement shall be made. Results will be used as reference.
 Then, the actual arrangement is calculated with the same parameters, changing only the geometry and the
current value.
- The pressure curve of the actual arrangement shall always remain below that of the reference
arrangement.
- If structural stresses are available, then pressure results can be ignored. The criteria would be only
the peak value of the structural stress, for the relevant part of the switchgear being different from
the reference switchgear. It shall be:
 Lower for the actual arrangement than the tested one, or
 Lower than rupture stress. This can be considered only in case the pressure given in the
reference simulation matches the tests measurement, within 10 % for example.
Page 116
Validation
Item Design parameter Condition
criterion
(1) (2) (3) (4)
1 Phase to phase clearance ≤
This concerns the region where the arc is

2 Phase to earth distance same
initiated.
3 Net enclosure/compartment volume ≥
Rated pressure of insulating gas, if

4 ≤
applicable; see note 1
This concerns the region where the arc is

5 Cross-section of conductors ≥
initiated.
Raw material of conductors (Al or Cu or This concerns the region where the arc is
6 same
their alloys) initiated.
Applying the rules of IEC 62271-200 (Ed. 2.0,

7 Location of the point of arc initiation same
2011).
8 Insulating material exposed to the arc same
The position of the exhaust area in the

compartment has to be the same.
9 Exhaust area ≥
Larger areas are only acceptable if an exhaust
duct is used.
10 Exhaust opening pressure ≤ Applicable to fluid-tight compartments.
Strength of fixing elements of relief Applicable to non-tight compartments.

11 ≤
device (flap) The relief device has the same design.
This also includes the strength of partitions and

12 Strength of the enclosure/ compartment ≥
bushings. Note 2 and 3.
13 Thickness of the enclosure walls ≥
14 Strength of the doors and covers ≥ Note 2 and 3
15 IP degree of protection of enclosure ≥ If relevant for indicator ignition criterion.
16 Short-circuit current ≤
17 Arc duration ≤
Note 1: For SF6 insulated switchgear the test is performed with air (see IEC 62271-200 at clause 6.106.3) at the same rated filling pressure as for SF6.
Note 2: An assessment of the strength might require calculations or FEM stress analysis.
Note 3: The assessment of the strength in particular has to consider the distance between all fixing points (bolts, hinges and latches).
Table 7-2: Criteria for comparison of MV switchgear (below 52 kV) panels with tested samples with regard
to pressure rise withstand capability by internal arc fault. 2nd CD of IECTR 62271-307 (2012).
Page 117
item Design parameter Validation Condition

criterion
(1) (2) (3) (4)
1 Phase to phase clearance ≤
2 Phase to earth distance ≤
3 Enclosure/compartment volume ≥
4 Pressure of insulating gas ≤
5 Cross-section of conductors ≥
6 Raw material of conductors (Al or Cu) Same
Applying the rules of IEC 62271-203 (Ed.

7 Location of the point of arc initiation Same
2.0, 2011)
Distance from the arc initiation point to

8 ≥
the expected nearest obstacle
9 Insulating material exposed to the arc Same
The position of the relief area in the

10 Exhaust/ventilation opening area ≥
compartment has to be similar
11 Exhaust opening pressure ≤
12 Pressure relief design Same
This also includes the strength of partitions

Strength of the enclosure/ and bushings.
13 ≥
compartment
Note 1 and 2
14 Thickness of the enclosure walls ≥
15 Raw material of enclosure (Al or Steel) Same
Manufacturing type of enclosure

16 Same
(casting, plate welding, etc.)
17 Short-circuit current ≤
18 Arc duration ≤
Note 1: An assessment of the strength might require hand calculations or FEM stress analysis (see Section 6.2)
Note 2: The assessment of the strength in particular has to consider the distance between all fixing points (bolts, hinges and latches)
Table 7-3: Criteria for comparison of HV gas-insulated switchgear (above 52 kV) section with tested
samples with regard to internal arc withstand capability.
7.4 Design Evaluation

In cases where no valid reference switchgear has been tested, the user may accept results of calculations which
satisfy the manufacturer’s design rules for internal arc withstand capability. The rules for arc withstand capability on
which the manufacturer has based the design of the unit to be evaluated should possess a solid experimental
Page 118
basis. The rules should be derived from representative tests. The users should first validate the soundness of the
manufacturer’s design rules by asking for the following information:
 Experience of manufacturer in switchgear design

 List of internal arc tests performed and geometric/electrical details of the tests considered
 Result of tests and their impact on design rules
 Evidence of internal arc failures in service: reports of actual internal failures on similar equipment if
available
The results of the calculations such as pressure rise, mechanical stresses in the enclosure and time to burn-
through will be compared with the limits of the design. In order to allow for the absence of test evidence, a
minimum safety margin of 20 % should be taken into account.
7.5 Summary
The switchgear user might accept the replacement of the internal arc withstand test of the specific switchgear by a
design analysis carried out by the manufacturer based on tests of a similar design. In order to be able to perform
this review the user should receive the necessary data to be able to reproduce at least an estimate of the analysis
presented by the manufacturer.
If the analysis is based on test results of similar switchgear, it is important to define the boundaries of this similarity
by comparison of the design parameters affecting the internal arc withstand capability of the switchgear. Table 7-2
and Table 7-3 provide an example of such comparison.
For High Voltage Switchgear rated over 52 kV tests are much less frequent and it might be probable that no test
results of similar switchgear are available. In such cases the user can accept results of calculations which satisfy
the manufacturer’s design rules for internal arc withstand capability. These design rules have to be validated by the
previous design experience and test evidence provided by the manufacturer.
Page 119
8 CONCLUSION
The goal of A3.24 WG was to assess the calculation methods and software tools that can be used to simulate the
effects of the internal arcing fault in MV and HV switchgear. The motivation for this work was multifaceted:
a. To provide methods for pressure rise calculations and allow benchmarking with performed tests
b. To provide methods for calcuation of other effects of the internal arc
c. To verify design modifications by simulations and reduce the number of internal arc tests for environmental
reasons
d. To verify validity of replacing SF6 with air during internal arc testing
The working group reviewed existing software tools for calculating the effects of an internal arc fault, focusing on 3
main effects of an internal arc: pressure rise, mechanical stress on enclosure/buildings, and burn-through.
There was a general consensus in the Working Group that software tools can be used for calculating the effects of
the internal arc under following conditions:
• the energy input (namely arc voltage) was taken from the actual test on the similar switchgear design. The
WG is not recommending calculating arc voltage without running actual test.
• the approriate models are used for the appropriate applications. For example, simplified model shouldn’t be
used for the switchgear with complex geometry and large installation rooms where pressure will not be
uniform.
The authors further agreed that simulations cannot replace type tests, but they could be used for interpolation
between the known tests, make good predictions and thus reduce the number of type tests.
Following 4 sections summarizes the WG findings:
a) Methods for pressure calculation and benchmarking with performed tests
This review included various “home made” software tools, ranging from simple spreadsheets in Excel that most
engineers could use with a little effort, to a complex 3-D Computational Fluid Dynamic (CFD) software
package whose application remain limited to small number of experts due to the complexity and cost of the
software.
Three different models for calculating pressure rise are proposed, with each one having different levels of
complexity and its own limitations, as shown in the table below.
Page 120
Approach /model Appropriate Application Limitations
 Assumes uniform gas parameters

To quickly calculate uniform  Not applicable for large relief openings
pressure rise inside an arc  Calculations are not reliable, when gas
Basic compartment and the exhaust temperature exceeds approx. 2000 K for SF6 and
1)
(low complexity) volume in typical MV 6000 K for air.
switchgear and HV GIS  Doesn’t consider gas mixtures in the exhaust
applications. compartment.
3
 Not applicable for large rooms (>50m )
To calculate uniform pressure  Assumes unifrom gas parameters

rise as under 1) adding further  Limitations and applications depend on the
Enhanced
2) (medium approximations to better implemented approximations.
complexity) match test results and
calculation.
For calculating spatial  High effort for the modeling and meshing of the
CFD pressure distribution and gas rooms and switchgear
3)
(High complexity) flow in odd shapes geometry  Requires large computing power and time.
and large rooms.
Table 8-1: Models for calculating pressure rise during internal arc fault
The basic model is fully developed and described in this technical brochure. It has been tested on more then 80
cases, for which the measured pressure rise results were compared with the simulation results. The results were
encouraging: agreement between calculations and measurements of the pressure rise were within +/- 20 % for the
arc compartment after adjusting the k factor and the discharge coefficient α. The comparison also indicates that
most arrangements can be successfully simulated by applying common input parameters: the coefficient kp of 0.5
for air and 0.7 for SF6, the discharge coefficients α between 0.7 and 1.0.
The pressure rise inside the compartments during an internal arc fault test can be successfully predicted as long
as the input arc energy is well known. This means that arc voltage should be taken (not calculated) from the
previous internal arc test on the similar switchgear design. Due to stochastic nature of the arc, accurate
calculation of the arc voltage in complex switchgear designs is still to be desired; hence recorded values should
be used for more accurate predictions of pressure rise and mechanical stress.
The basic model also helps to understand which parameters are contributing more to pressure rise then others.
For example, arc voltage has much more influence on maximum pressure then level of asymmetry of the fault
current.
Pressure rise calculations are less accurate for the exhaust compartment. Enhanced models can be used for
more accurate results in the exhaust compartment.
For large installation rooms and arc/exhaust compartments with complex geometry where pressure isn’t uniform,
CFD calculation should be used.
Page 121
It shall be also noted that existing IEEE and IEC arc resistance standards doesn’t require measuring pressure
rise during the test. The authors of this Technical Report recommend that pressure rise is measured during the
test in both arc compartment and ehxaust compartement. Having the pressure measurements along with the
input arc energy, all 3 models can be developed and Kp factor calculated. IEC and IEEE arc resistance
standards should consider adding the pressure measurent requirement in the next standard revision.
b) Methods for calculating other effects of the internal arc.
Namely, three additional effects of the internal arc on structures have been analyzed in this technical brochure:
 Mechanical Stress on switchgear due to overpressure

 Mechanical stress on building walls due to overpressure
 Burn-through
Mechanical deformation (and von Mises stresses) of the switchgear enclosure can be reasonably accurately
calculated with off-the-shelf FEA software once the pressure curve is known.
Also, the walls in Air-Insulated Substations can be dimensioned by calculating the pressure rise using the basic
model (with some limitations that need to be taken into account). However, for Gas-Insulated Substations, the
simplified method may lead to underestimation of the pressure. With CFD calculations the localized time-
dependent pressure is obtained, which can be used to determine the size and placement of pressure relief
openings in the building.
Also, it has been shown how the burn-through time can be evaluated using different empirical formulas and
how the test results agree with calculations.
It should be noted that a large scatter in the test results is possible since the energy released in one test may
be different in another test depending on the erratic arc movement for example. For example, repeating internal
arc tests on the same piece of switchgear equipment can result in +/-20% difference in maximum pressure, as
well as in burn-through time. This scatter should be considered in order to allow a sufficient safety margin in
relation to the calculation results.
c) Verifying design modification by simulation methods and reducing number of tests
In order to reduce number of internal arc tests, this technical brochure provides guidance for internal arc
simulation review.
The switchgear user might accept the replacement of the internal arc withstand test of the specific switchgear
by a design analysis carried out by the manufacturer based on tests of a similar design. In order to be able to
perform this review the user should receive the necessary data to be able to reproduce at least an estimate of
the analysis presented by the manufacturer.
If the analysis is based on test results of similar switchgear, it is important to define the boundaries of this
similarity by comparison of the design parameters affecting the internal arc withstand capability of the
switchgear.
Page 122
For High Voltage Switchgear rated over 52 kV, tests are much less frequent and it might be probable that no
test results of similar switchgear are available. In such cases the user can accept results of calculations which
satisfy the manufacturer’s design rules for internal arc withstand capability. These design rules have to be
validated by the previous design experience and test evidence provided by the manufacturer.
d) Replacing SF6 with air during internal arc testing
After thorough examination of existing test data, the authors agreed that replacement of SF6 with air during
internal arc testing provides mixed results. It’s been concluded that there is no “silver bullet” recommendation
and that each case must be evaluated separately. Some observations are listed below:
 Arc compartment: Pressure development and resulting mechanical stresses in air are in most cases
higher than in SF6. Burn-through might happen faster with SF6 than with air under the same conditions.
 Exhaust compartment: Pressure development and the resulting mechanical stresses in SF 6 are in most
cases higher than in air. Burn-through is not applicable because there is no arc in the exhaust
compartment.
 Simulation room – indicators: The likelihood of indicator ignition might be comparable for both cases.
Ignition of the indicators would be hardest to predict even with CFD software. More work has to be
done to investigate the correlation between the flammability of the cotton samples, incident heat
energy densities, and arc flash protection requirements.
In conclusion, software tools for simulations of the internal arc in MV and HV switchgear can be used in a number
applications, including:
o Helping designer to design an arc-resistant switchgear.

o Helping civil engineer to determine wall dimensions and size of the openings of the installation room.
o Extending the validity of the type tests on a similar design and thus reduce the number of internal arc
tests. Note that Internal arc test must be done on the similar design to get the correct energy input data -
. It is important to measure the pressure rise during the internal arc test.
o Validating various installation conditions, including different exhaust channels/ducts and different size
rooms / pressure relief openings.
Page 123
ANNEX A EQUATIONS FOR PRESSURE RISE CALCULATION

A.1 In troduction
Chapter 2 provides an overview of the basic equations, assumptions and limitations of the basic model. This Annex
focuses on deriving all the equations, so a reader can use this information and create one’s own tool for pressure
rise calculations. Also, some possible enhancements of the basic model are listed in Section A.4
A.2 Equations for the basic model

The basic equations are listed in this section. They describe pressure and temperature change in a control volume.
Gas mass in the control volume is changed in time by the inflow and outflow of gas. The internal energy can be
changed by a heat source. Gas properties, like density, temperature or pressure as well as the heat source are
considered to be uniform in the control volume. An ideal gas and adiabatic gas flow are assumed.
pressure in the high pressure region
pressure in the control volume (CV)
pressure in the low pressure region
area of the opening between the high pressure

region and the CV
̇ mass flow through
area of the opening between the CV and the

low pressure region
̇ mass flow through
̇ heat source
Figure A-1: Control volume.
Figure A-1 presents an enclosed gas container called the control volume with two openings, which allow gas to
flow into and out of the volume. The gas can be heated up by a heat source.
A.2.1 Definition of specific gas parameters
For an ideal gas the specific heat capacity at constant volume is:
= (A-1)
where is the specific heat capacity at constant volume, is the specific internal energy, and is the
temperature.
The specific heat capacity at constant pressure is:
ℎ
= (A-2)
where is the specific heat capacity at constant pressure, ℎ is the specific enthalpy, and is the temperature.
The heat capacity ratio, , is given by the heat capacity ratio:
Page 124
= (A-3)
The specific enthalpy is defined as:
ℎ= + (A-4)
where is the pressure and is the specific volume.
The specific gas constant is given by the universal gas constant divided by the molar mass:
= (A-5)
where is the specific gas constant, is the universal gas constant, and is the molar mass.
A relationship between the specific gas constant and the specific heat constants is:
= − (A-6)
A.2.2 Ideal gas law
The ideal gas law is:
= (A-7)
where is the volume and is the mass.
From equation (A-5) it follows that
= (A-8)
From equation (A-6) it follows that
= − (A-9)
This can be written using equation (A-3) as
= ( − 1) (A-10)
This equation can be used to calculate the density based on the pressure, temperature, specific heat capacity and
heat capacity ratio:
= = (A-11)
( − 1)
where is the density.
A.2.3 Conservation of mass
The mass change in the volume is given by the difference between the in flowing and out flowing gas masses.
Page 125
= ̇ − ̇ (A-12)
where is the mass in the control volume, ̇ is the mass flow into the volume, and ̇ is the mass flow out of
the volume.
The mass inside the control volume (CV) may consist of different gas components. In the following equation two
components are considered but this approach can be extended to additional components.
= + (A-13)
where is the mass of the type A component in the CV and is the mass of the type B component in the CV.
The mass fractions can be calculated as:
= (A-14)
= (A-15)
where is the mass fraction of gas type A in the CV and is the mass fraction of gas type B in the CV.
It is assumed that only gas component A flows into the control volume. Then the mass changes of the different gas
components in the control volume are described by:
= ̇ − ̇ (A-16)
=− ̇ (A-17)
The sum of gas component changes gives the total mass change:
= + (A-18)
, = + (A-19)
, = + (A-20)
,
= (A-21)
,
where is the heat capacity ratio of mixed gas, is the specific heat capacity of gas type A, and is the specific
heat capacity of gas type B.
The mass fractions in the second compartment, the exhaust compartment, are calculated accordingly taking the
different components into account.
Page 126
A.2.4 Mass flow
The following equations describe the mass flow through the opening . The mass flow through the opening
can be calculated with the same equations but with adjusted indices. The mass flow is given by:
̇ = (A-22)
where is the area of the relief opening, is the gas density in the opening, is the velocity in the
opening, is the discharge coefficient of the relief opening, and ̇ is the mass flow rate out of the control
volume.
The gas density in the opening is given by:
/
= (A-23)
where is the gas density in the control volume, is the gas density in the opening, is the pressure in the
control volume, is the pressure in the opening, and is the heat capacity ratio of the gas in the control volume.
The gas density in the control volume can be calculated with equation (A-11).
The gas velocity in the opening is:
2
= 1− (A-24)
−1
where is the gas density in the control volume.
The pressure in the opening cannot fall below the critical pressure or the pressure in the low pressure region.
∗)
= max( , (A-25)
where is the pressure in the low pressure region and ∗

is the critical pressure in the opening.
The critical pressure is given by
∗
2 (A-26)
=
+1
A.2.5 Conservation of energy
The first law of thermodynamics for an open system says that the change of energy within the system boundaries is
equal to the difference of the energy entering and leaving the system.
The change of the stored energy in the system, which is given on the left side of equation (A-27), is described by
the time derivative of the sum of the specific internal energy, the specific kinetic energy and the specific potential
energy times the mass of the gas.
The right side of the equation is the sum of the heat fluxes, the technical work done by the system, the
compression or expansion work, and the specific energy transported by the incoming or outgoing mass flows.
Page 127
The left side of the equation is valid for a stationary gas, the energy of which is described by the specific internal
energy. Because the right side considers a moving gas, the internal energy has to be replaced by the specific
enthalpy; see equation (A-4).
+ 2 +
− (A-27)
= ̇+ ̇ + + ̇ ℎ+ +
2
The terms are as follows.
Left side: is the mass in the system, is the specific internal energy, is the specific kinetic energy of the
system, is the specific potential energy of the system.
Right side: ̇ is the heat transfer, ̇ is the mechanical work, is the expansion or compression work, ̇ is
the mass flow, is the specific kinetic energy of the mass flow, ℎ is the specific enthalpy of the mass flow, and
is the specific potential energy of the mass flow.
The following terms can be simplified when applying equation (A-27) to the control volume.
̇ =0 no mechanical (technical) work
=0 potential energy is zero for all volumes and regions
=0 constant control volume
Σ ̇ = ̇ one heat source
=0 inside the control volume the gas is at rest
Equation (A-27) can then be written as:
( )
= ̇+ ̇ ℎ+ − ̇ ℎ+ (A-28)
2 2
where is the mass in the control volume, is the specific internal energy of the gas in the control volume, ̇
is the mass flow, and ℎ is the specific enthalpy.
The inflow and outflow is an adiabatic process. Therefore:
ℎ = ℎ+ (A-29)
2
Page 128
ℎ = ℎ+ (A-30)
2
where ℎ is the specific enthalpy of the gas in the high pressure region and ℎ is the specific enthalpy of the
gas in the control volume. Equation (A-28) can be simplified as:
( )
= ̇+ ̇ ℎ − ̇ ℎ (A-31)
Equation (A-31) can be used to calculate the temperature.
+ = ̇+ ̇ ℎ − ̇ ℎ (A-32)
Where is the mass in the control volume and is the gas temperature in the control volume.
With equation (A-1) and equation (A-12) equation (A-32) can be written as:
̇+ ̇ ℎ − − ̇ (ℎ − )
= (A-33)
ℎ specific enthalpy of the gas in the high pressure region
ℎ specific enthalpy of the gas in the control volume
specific internal energy of the gas in the control volume
mass in the control volume (CV)
specific heat capacity at constant volume in CV
̇ heat source in CV
̇ mass flow into the CV
̇ mass flow out of the CV
Page 129
From the ideal gas law the pressure follows as:
( − 1)
= (A-34)
A.3 Basic model

For constant specific heat capacities at constant pressure (classical ideal gas, = constant) the specific enthalpy
can be expressed as:
ℎ= (A-35)
For constant specific heat capacities at constant volume (classical ideal gas, = constant) the specific internal
energy can be expressed as:
= (A-36)
If the specific heat capacities are assumed to be constant, these equations can be used to write equation (A-33)
as:
̇+ ̇ − − ̇ −
= (A-37)
gas temperature in the control volume (CV)
gas temperature in the high pressure region
specific heat capacity at constant pressure in the high pressure region
specific heat capacity at constant pressure in CV
specific heat capacity at constant volume in the high pressure area
specific heat capacity at constant volume in CV
The Euler method is used to solve differential equations (A-12) and (A-37).
Step 1: Initial conditions
( = 0) Gas pressure in the control volume
( = 0) Gas temperature in the control volume
The volume , the specific heat capacity and the heat capacity ratio are assumed to be constant. The
environment has constant ambient pressure and temperature. The initial mass in the volume follows from equation
(A-11):
( = 0)
( = 0) = (A-38)
( − 1) ( = 0)
Page 130
where is the gas mass in CV.
Step 2: Changes of mass and temperature
Δ = ̇ Δ − ̇ Δ (A-39)
where ̇ is the mass flow into the CV and ̇ is the mass flow out of the CV.
Δ +Δ − −Δ −
Δ = (A-40)
Step 3: Numerical integration
( +Δ )= ( )+Δ (A-41)
( +Δ ) = ( ) + ΔT (A-42)
Δ is the mass change and TCV is the change of temperature in the control volume.
From the ideal gas law the pressure follows as:
m (t + Δt)c {κ − 1}T (t + Δt)

p (t + Δt) = (A-43)
V
A.4 Enhancement of the basic model

In general enhanced models are based on the same set of equations as the basic model with appropriate
modifications such as:
 density dependent -factor (Section A.4.1)

 exothermic reaction energy (Section A.4.2)
 pressure dependent arc voltage (Section A.4.3)
 mixing of gas in compartments (Section A.4.4)
 metal evaporation and ablation of insulators (Section A.4.5)
 arc absorbers in the exhaust flow (Section A.4.6)
 speed of relief opening device (Section A.4.7)
 temperature dependent gas data (Section A.4.8)
These modifications do not increase the accuracy in predicting the peak pressure in the arc compartment, but do
increase the accuracy in simulating the time dependence of the gas outflow and the pressure development in the
exhaust compartment.
A.4.1 Density-dependent k p -factor
An important enhancement of the basic model is the introduction of a density-dependent -factor after the opening
of the pressure relief device. The -factor determines the pressure rise until reaching the response pressure of the
Page 131
relief device. In the basic model, is assumed constant after the relief device opening even after a large outflow of
gas. This assumption leads to rather high gas temperatures inside the arc compartment, because the same arc
power heats up an ever decreasing gas mass. Since gas temperatures above approximately 10000 K have never
been measured in free burning arcs and indeed are not realistic, other effects which keep the gas temperature
lower have to be considered.
Tests with arc exhaust into a closed room [Dullni1994] have shown that the pressure rise in the exhaust
compartment is not as high as anticipated when assuming a constant factor over the whole calculation duration.
Since a cooling effect due to the walls of the room can be neglected, one can conclude that the energy transferred
from the internal arc to the surrounding gas decreases over the duration of the arc. This effect can be simulated by
a -factor diminishing with gas density in the arc compartment [Friberg1999]. It has to be noted that the pressure
inside the arc compartment finally drops to ambient pressure after the release of most of the contained gas mass.
Unless cool gas from outside enters the compartment through unintentional gaps in the enclosure or metal
evaporation from contacts occurs, the internal gas density will become extremely low.
A good approximation for a density-dependent -factor was obtained in an experiment using SF 6 as well as air as
the insulating gas by cutting in half when the gas density decreased by a factor of 5, 25 etc [Dullni1994]. This
function of the -factor can be approximated by a power law applied to the remaining gas density in the
compartment in relation to the gas density at ambient pressure and temperature. The exponent approximating the
published tests is between 0.4 and 0.5. A pre-factor is adapted to provide a continuous transition from the -factor
determined for the initial pressure rise.
ρ( t )
k (t) = k c for ρ(t) < ρ (A-44)
ρ
k (t) = k for ρ(t) > ρ (A-45)
where ( ) is the gas density at time , is the normal gas density at ambient pressure and temperature, is the
-factor before transition, the adapted pre-factor, the gas density at transition, and is an exponent between
0.4 and 0.5.
The reduction of the -factor might start when the gas density falls below 20 % of the normal gas density as was
the case in the publication [Dullni1994], but could also be adapted to different transition densities . The
introduction of a density-dependent -factor diminishes the exhaust of gas and hence reduces the pressure in the
exhaust compartment or installation room. It keeps the gas temperature lower at later simulation times and has a
beneficial side effect of removing numerical instabilities. The variation of the pressure inside the arc compartment is
hardly changed, in particular up to the pressure peak.
A.4.2 Exothermic reaction energy
The type of material used in the conductors and walls, where the arc has its roots, influences the value of the -
factor. It is known that SF6 and to some extent also air react with aluminum with an exothermic release of energy.
In this case, > 1 may give appropriate results [Zhang2002]. Another approach is to add the exothermic energy
from the evaporated metal to the arc energy avoiding an artificially augmented -factor. From the reaction
Page 132
equations, oxygen and SF6 release a similar exothermic energy when reacting with aluminum [Bjørtuft2005].
However, since air contains only a fraction of 20 % oxygen, the total amount of released exothermic energy is a
factor of 5 smaller compared with SF6.
A.4.3 Pressure-dependent arc voltage
The voltage of an internal arc is seldom constant during a half cycle of current, and also varies in time over the arc
duration. It was observed that arc voltage is higher during the high pressure phase in the arc compartment than
later after the release of the gas. An empirical formula has been deduced for a single-phase arc burning between
two parallel copper electrodes in an aluminum container [Dullni1994]. The formula given below has been proposed
from arc initiation up to the highest pressures for air as well as for SF 6 with an adapted initial value of 500 V for
air and 950 V for SF6. After the release of the gas i.e. after reaching ambient pressure, the arc voltage varied
between 400 and 600 V for air and between 400 and 800 V for SF6.
∆
.
( )= 2− kPa (A-46)
where is the adapted initial arc voltage and Δ is the momentary pressure (relative) inside the arc compartment
in kPa.
Because of the considerable variation of arc voltage during the arc duration, it is recommended that the actual
measured arc power, i.e. the product of momentary current and phase-to-earth voltage be used as an input for the
calculation.
A.4.4 Mixing of gas in compartments
The flow of SF6 gas out of the arc into the exhaust compartment containing air modifies the gas properties in the
latter volume [Anantavanich2010]. The energy transferred from one compartment to the other is correctly
implemented in the basic model (within the limitations of the basic model). However, when the incoming mass of
SF6 is added to the mass of air in the compartment, it is assumed that the mixed gas behaves like pure air.
Because of the different adiabatic indexes of SF6 and air – considering the term ( − 1) in equation (2-16) - the
pressure rise in the exhaust compartment will be overestimated. In particular for small compartments, where the
fraction of SF6 becomes predominant, the gas temperature might easily be overestimated by a factor of 2. For
improved accuracy, the gas properties , and have to be calculated according to the fraction of the gases in
the total mass. This requires separate gas equations for all components.
Mixing of different gases can be implemented in a simplified way in an enhanced model by applying the equations
given in Section A.2.3. In particular this is recommended for the mixing of SF6 gas flowing out of the arc
compartment into the exhaust compartment filled with air. Gas mixing is also important for the correct consideration
of the effect of evaporated material on the pressure rise in the arc compartment as described in Section A.4.5.
Again, the equations given in Section A.2.3 can be applied.
A.4.5 Metal evaporation and ablation of insulators
From tests, it is obvious that hot gas is not the only substance ejected from switchgear, but also e.g. metal vapor
and carbon black. Because of the severe absorption of light by metal vapor and dust, the cloud coming out of the
Page 133
relief opening often looks black or brownish, whereas the direct exhaust jet is bright. Evaporation of metal vapor
from hot arc roots and ablation of insulating material by the hot arc column can add a considerable amount of gas
to the arc compartment. This effect counteracts the ever increasing gas temperature in the simulation, when the
gas density inside the arc compartment becomes smaller and smaller due to the exhaust of gas mass. Metal
evaporation is therefore expected to play an important role during the exhaust phase, when the pressure in the arc
compartment equals ambient pressure.
The evaporation effect can be introduced as an enhancement into the basic model by properly considering the part
of the arc energy deposited in the arc roots, . The resulting mass of metal vapor is given by the inverse of the
specific evaporation energy of the metal multiplied by the fraction of the arc energy. Alternatively, measured
values of evaporated material as a function of arc energy can be used (Table A-1).
k ΔW
Δm = (A-47)
W
Δ evaporated mass from metal electrodes
specific evaporation energy of metal
fraction of arc energy used for evaporation
ΔW electric arc energy
The amount of arc energy deposited in the arc roots is in the order of 10 to 15 % [Zhang2002]. is smaller than
this fraction, since the arc energy is also consumed for heating and melting of the electrode material. Measured
material loss of electrodes as a function of the integrated arc current provide similar values to the specific
evaporation energies, when a cathode and anode drop voltage of approximately 50 V is used. Table A-1
[Zhang2002] contains measured electrode mass loss for different materials.
Electrode loss in mg/As Al Cu Fe

SF6 8 5.8 5
Air 5.8 2.9 2.5
Table A-1: Electrode mass loss [Zhang2002].
Page 134
The mass of metal atoms, , can be simply added to the mass of gas in the arc compartment.
Δm = Δm − Δm (A-48)
where Δ is the mass flow of gas out of the volume.
For a more advanced approach, the evaporated metal atoms can be treated as a separate gas which is added to
the filling gas of the compartment in accordance with equation (A-13) in Section A.2.3. Then also the different mass
and specific heat capacities of the metal atoms can be correctly considered.
The amount of ablated insulator material, , can be estimated from the inverse of the specific dissociation
energy of the insulation material multiplied by some other fraction of the arc energy. The specific dissociation
energy can be determined from the total of all binding energies of the chain molecules of the insulation material,
subtracting all binding energies of recombined gaseous molecules, or alternatively by measuring the released
volume of gaseous molecules [Hochhaus1985]. Most often epoxy with glass fillers is involved, which requires
subtracting the non-soluble amount of the filler material. It is assumed here that most of the ablated material is
released as CH4 gas. Again, the mass of ablated material, , can be simply added to the mass of gas in the
arc compartment or implemented with more effort through separate mass equations.
Δm = Δm + Δm − Δm (A-49)
k ΔW
Δm = (A-50)
W
1
=V ρ η (1 − η ) (A-51)
W
Δ ablation mass from insulators
specific dissociation energy of insulation material
fraction of arc energy spent on ablation of solid material
volume of dissociated solid material per unit of energy
specific density of solid insulation material
fraction of gas (CH4) in ablated material
fraction of filler material in insulator
It is anticipated that the fraction of arc energy used for ablation or evaporation does not depend on time i.e. is
constant for the whole duration of the arc. It is not a volume process depending on the filling gas density, but a
surface process. Also, it has to be considered that the evaporated or ablated gas has to get into thermodynamic
equilibrium with the surrounding gas in the compartment. Since the evaporated atoms leave the electrode surfaces
already at evaporation temperature, this either constitutes a heat sink or a heat source depending on the
surrounding gas temperature being higher or lower. This energy has to be considered in the temperature balance
of the gas mass in the arc compartment assuming instantaneous mixing. For energy exchange the heat capacities
of the gaseous materials at constant volume are relevant.
Page 135
ΔQ − Δm c −c T + Δm c , (T − T ) + Δm c , T −T
ΔT = (A-52)
m c
, specific heat capacity of gaseous ablated material
, specific heat capacity of metal vapor

specific heat capacity of the mixed gas
temperature of ablated material
evaporation temperature of electrode material
Calculations confirm that the peak pressure in the arc compartment is hardly changed by the evaporated material
except in special cases, where exothermic reactions prevail. On the other hand, the course of the declining
pressure in the arc compartment and the pressure in the exhaust compartment may be indeed modified by the
presence of evaporated or ablated material. The composition of the gas during the exhaust may become
dominated by metal vapor instead of the original filling gas. For correct modelling, it is required that the composition
and mixing of the gas be taken into account to reliably simulate the evaporation effect (see Section A.2.3). Table A-
2 displays the input values for calculations.
Physical parameter Value unit

Evaporation energy of aluminum 13.7 kJ/g
Evaporation energy of copper 6.1 kJ/g
Evaporation energy of steel 8.4 kJ/g
Dissociation energy of epoxy material 95.7 kJ/g
Specific heat of methane (CH4) at constant volume 1633 J/kg K
Specific heat of gaseous copper (Cu) at constant volume 201 J/kg K
Specific heat of gaseous aluminum (Al) at constant volume 470 J/kg K
Ablation temperature of epoxy K
Evaporation temperature of Cu 2870 K
Evaporation temperature of Al 2720 K
Table A-2: Input values for calculation of pressure rise including material evaporation.
A.4.6 Arc absorbers in the exhaust flow
In some switchgear designs the release of gas is cooled down by arc absorbers. These are composed e.g. of
several layers of metallic or insulating fine-meshed grids or other obstacles in the path of the released gas. These
devices absorb some fraction of the energy of the outflowing gas, however they may also reduce the effective
cross-section of the pressure relief area. In this case the gas remains longer in the compartment and might carry
more arc energy into the installation room than would occur without the arc absorber counteracting the cooling
effect. The effect of such arc absorbers can be implemented in the equations of the models by a reduced area of
the relief opening and a concurrent cooling factor reducing the energy transported from the arc compartment into
Page 136
the exhaust compartment. The cooling effect per mesh is however limited to several percent of the energy flow and
does not linearly increase with the number of consecutive meshes [AiF2011].
A.4.7 Speed of relief opening device
Heavy relief flaps, as used sometimes for air-insulated switchgear, do not open instantaneously. The time-
dependent increase of the opening area can be calculated from the acceleration of the plate having a certain mass
under the force determined by the scalar product of gas pressure and area of the relief opening. These flaps are
often hinged on one side so that the force on the plate is reduced during the opening time. The simulation of its
motion together with the determination of the momentary opening area is usually not considered. In principle, the
speed of the plate can be calculated by integrating the acceleration of the plate over time. For the light bursting
discs in SF6-insulated switchgear, this effect is of minor importance, since the full opening area is achieved almost
instantaneously.
A.4.8 Temperature-dependent gas data
The basic pressure calculation method is based on a gas model called “the classical ideal gas”, i.e. the gas
particles are considered as being dimensionless with their mass being concentrated in points. Collisions are only
elastic and the gas properties like the specific heat capacities are constant.
A considerable improvement is already obtained, when the gas properties are based on the “ideal gas” model. In
this case the gas still consists of dimensionless mass-carrying particles, however, e.g. dissociation and ionisation
are considered by the number of particles changing with temperature. Thus, the gas properties (the specific heats,
sound velocity etc.) are no longer constant but depend on temperature, and also the mole numbers change with
temperature. The ideal gas law is still valid. Considering this gas model, the equations in Chapter 2 become much
more complex.
Some enhanced pressure calculation methods are based on real gas data. In this case the gas particles are no
longer regarded as concentrated points but have a certain volume so that interactions between particles like dipole
forces exist. The properties of real gases depend not only on their composition, but also on chemical reactions,
temperature and pressure. The generation of these data is time-consuming. That is why they are collected once in
multidimensional tables and then used in the pressure calculation method.
The specific heat capacity of SF6 increases by more than a factor of 10 around a temperature of 2000 K due to
collision-induced dissociation of molecules. A similar anomaly is observed for N2 at a temperature of 6000 K. The
temperature and pressure dependence of the heat capacity and the corresponding heat capacity ratio can be
introduced into the equations by analytic approximations of the heat capacity at constant volume during the time
steps of calculation. This approach is permitted as long as the temperature in the volume is uniform allowing for a
thermodynamic equilibrium of all molecule fractions. This might no longer be true for rapidly flowing gas in the
exhaust.
The inclusion of temperature-dependent heat capacities involve many modifications to the equations, which are not
easy to implement into the basic model and therefore are not discussed in detail here.
Page 137
A.5 Summary
A detailed description of the equations which are used for the basic model is given. The equations are based on
the ideal gas law, conservation of mass, and conservation of energy. The reader can use these equations to
develop his/her own software. The calculation of the initial conditions as well as the temperature and pressure
change for each time step is described. The code may be used to recalculate the examples given in Chapter 2.
In addition, equations to enhance the basic model are given. Enhancements consider density-dependent -factor,
exothermic reaction energy, pressure-dependent arc voltage, mixing of gas in compartments, metal evaporation
and ablation of insulators, arc absorbers in the exhaust flow, speed of relief opening device, and temperature
dependent gas data.
REFERENCES:
[AiF2011]: Jan Christoph Kahlen, G. Pietsch: “Reduzierung der Druckbeanspruchung elektrischer Anlagen im
Störlichtbogenfall” (“Reduction of the pressure stress in electrical installations due to internal arcs”), Schlussbericht,
AiF-Forschungsvereinigung, 2011.
[Anantavanich2010]: K. Anantavanich: “Calculation of Pressure Rise in Electrical Installations due to Internal Arcs
Considering SF6-Air Mixtures and Arc Energy Absorbers”, Aachener Beiträge zur Hochspannungstechnik, Band
14, ISBN 3861306778, Dissertation, RWTH Aachen University, 2010.
[Bjørtuft2005]: T. Bjørtuft, O. Granhaug, S. Hagen, J. H. Kuhlefelt, G. Salge, P. K. Skryten, S. Stangherlin. “Internal

arc fault testing of gas insulated metal enclosed MV switchgear”. CIRED 2005 proceedings. Turin, 6-9 June 2005.
[Dullni1994]: E. Dullni, M. Schumacher, G. Pietsch, “Pressure rise in a switchroom due to an internal arc in a
switchboard”, 6th Int. Symposium on Short Circuit Currents in Power Systems, Liege, 1994.
[Friberg1999]: G. Friberg und G. Pietsch: “Calculation of pressure rise due to arcing faults”, IEEE Transactions on
Power Delivery, Vol. 14(2): S. 365-370, 1999.
[Hochhaus1985]: H. Hochhaus, „Untersuchung der Wechselwirkungen zwischen Schaltlichtbögen und Isolierstoff-

wänden“ (“Investigation of the interaction between switching arcs and insulating walls“), PhD Thesis in German,
Technical University Braunschweig, 1985.
[Zhang2002]: Xiang Zhang: “Modellierung der Auswirkungen von Störlichtbögen in elektrischen Anlagen”
(“Modelling of the impacts of internal arcs in switchgear”), Ph.D Dissertation, RWTH Aachen, 2002.
Page 138
ANNEX B PRESSURE SENSORS: TYPES AND ACCURACIES

B .1 In troduction
An important point in the evaluation of internal arc tests for later comparison with simulations is a correct
measurement of the pressure in compartments. Here not only is the type of pressure sensor decisive, but also the
recording equipment, the post-processing of the signal and the data format for storage. Also, the position of
sensors during the test has to be considered.
B .2 Pressure sensors
Most pressure sensors are based on the piezo-electric or piezo-resistive effects which generate a voltage or modify
a resistance when a crystal or a semiconductor is mechanically deformed e.g. by the overpressure in a switchgear
compartment caused by an internal arc.
On the other hand, only a minor fraction of the market is taken up by capacitive-type sensors. The reasons for the
lack of a breakthrough in this type of sensor can be found in the design complexity and the requirements for a
matched sensing circuit [Puers1993].
Piezoelectric pressure sensors are primarily recommended for dynamic pressure measurements, however some
quartz pressure sensors have long discharge time constants (DTC) that extend their low-frequency capability to
permit static calibration and measurement of quasi-static pressures over a period of a few seconds. DTC is defined
as the time required for a sensor or measuring system to discharge its signal to 37 % of the original value from a
step change of measure.
Piezo-resistive sensors are also good for fast transient dynamic pressure measurement. The remarkable
advantages of the piezo-resistive type are: a higher sensitivity (larger signal), cheaper price, and simpler amplifier
unit compared with the piezoelectric type. However, the main disadvantage is its weakness under thermal shock.
Dynamic pressure, generated by an internal arc, is translated into static pressure when it flows towards a wall and
stops [Bjørtuft2005]. Therefore, what is measured depends on the location of the sensor with respect to the gas
flow direction. Accurate mounting of pressure sensors is essential for good pressure measurements. It is important
to check the installation drawings supplied in the manual for the sensor, or to contact the pressure sensor supplier
to request detailed mounting instructions.
On the other hand, sensors have to be selected according to the expected pressure and frequency range. The
sensors need an auxiliary d.c. voltage supply in the range of 10 to 30 V. The output signal can be directly
connected to the input impedance of a transient recorder or via some resistive load. The temporal response should
be in the millisecond range in order to provide sufficient time resolution. Following the Nyquist–Shannon theorem,
the sampling rate should be at least twice the bandwidth [Shannon1998].
A bridge circuit for temperature compensation and reference to atmospheric pressure is normally implemented in
the sensor. The sensor is embedded in a steel or plastic housing and can be connected to the switchgear
enclosure by a short insulating tube in order to provide galvanic insulation for the measuring system. As the
connecting tube is exposed to external electromagnetic influences, it should be designed and installed conforming
to Electromagnetic Compatibility practical installation guidelines ["EMC"], e.g. reducing its length.
Page 139
Furthermore, the sensor signals might include some level of noise. Since piezo-electric sensors are sensitive to
mechanical shocks and vibrations, this noise may come from the internal arc itself. Most often, however, an
oscillation with a regular frequency around 1000 Hz is observed. This oscillation may originate from standing waves
in the short connecting tube between compartment and sensor with a dependence on the length of the connection.
A proper filtering of this frequency e.g. by digital means removes these disturbances and provides a smooth
pressure curve, which can be compared with calculations (see Figure B-1). Furthermore, the influence of solid-
borne vibration should be minimized e.g. by appropriate sensor adapters.
Figure B-1: Measured pressure curves (in SF6) before filtering (black) and after digital filtering (red, blue).
B .3 Accu racy o f the measuring system

Accurate measurement of the static overpressure created by an internal arc is extremely challenging. When
acquiring pressure measurements from transducers mounted on test structures, it is often desirable to quantify the
rate of rise (rise time) of this overpressure, the relative timing between overpressure and/or structural response
events, or both. Regardless of which, it is important to select transducers with adequate rise times to acquire these
measurements with fidelity - pressure rise times demand an extremely high-frequency response from the
measuring pressure transducers and their associated signal conditioning.
Therefore, it is necessary to select measurement system components (amplifiers, filters, displays, etc …) capable
of maintaining the measurement acquisition fidelity/reliability. However, measurement system components are
typically specified in terms of the upper frequency at which they provide -3 dB signal attenuation. A challenge then
exists to infer the rise time capability of an entire measurement system based on the -3 dB specifications of its
individual components [Walter2004].
In addition, concurrent transient temperatures, pressure waves, intense light radiation, fragment impact, ionized
gases, and other undesirable environmental effects all attempt to couple into the transducer and its mount, the
instrumentation cabling, and other measurement system components.
Nevertheless, every manufacturer’s transducers will respond to these undesirable environmental effects, and
thermal transient responses (such as to intense radiation) can be mitigated by application of ceramic or RTV (room
temperature vulcanizing) coatings (silicone elastomer) on the face of the transducer diaphragm [Walter2005].
These provide a thermal delay, hopefully until the blast event is over. [Hilten1978] provides one such quantitative
study of time delays that are achievable.
Page 140
If the influence of these undesirable environmental effects is not compensated for or isolated, the signal output
from the measurement system can be severely corrupted [Walter2010].
For example, strain, acceleration and temperature can all interact with the piezoelectric crystal within the
transducer to result in an erroneous pressure indication [Walter2005]. Furthermore, thermoelectric, photoelectric,
electromagnetic, triboelectric, and other energy-induced effects can result in additive electrical signals that create
errors in the transducer output. All of these extraneous signals can be viewed as noise, which contaminates the
desired pressure measurement. To validate that the transducer output signal is uncontaminated (i.e., it is solely
attributable to pressure), a combination of placebo [IEST] and “check” channels [Stein1992] could be used.
The placebo transducer can be applied in the test object in the same manner as any of the operational transducers,
but it will not respond to mechanical inputs (pressure, acceleration, strain). Any electrical output from it identifies
signal contamination due to thermoelectric, photoelectric, electromagnetic, and/or triboelectric effects. Besides, to
identify the combined effect of acceleration and strain on the piezoelectric element, an operational transducer can
be taken and isolated from the desired pressure environment. It becomes a “check” channel. Any signal output
from the check channel in excess of that produced by the placebo transducer would be noise induced by strain
and/or acceleration.
The combination of the placebo transducer and “check” channels allows almost all of the aforementioned undesired
responses to be documented with the noted exception of thermal effects due to transient temperature.
B .4 Summary
Summarizing, accuracy in the measurements is crucial because the typical value of acceptable tolerance of the
results to be obtained in the overpressure simulations if compared to the real laboratory test results are 5 % to 10
% [Feitoza2010].
In addition, every pressure measuring system must undergo type and routine tests followed by consecutive
performance tests and checks throughout its service life, as IEC advises e.g. IEC 61298-2. These performance
tests and checks shall prove that the measuring system can measure the intended test overpressures within the
uncertainties given in the International Standards, and that the measurements are traceable to national and/or
international standards of measurement. The measuring system distributor must guarantee the reliability/exactness
of his products, and the customer should verify and guarantee the accuracy of the pressure system device along
time.
Page 141
REFERENCES:
[Bjørtuft2005]: T. Bjørtuft, O. Granhaug, S. Hagen, J. H. Kuhlefelt, G. Salge, P. K. Skryten, S. Stangherlin. “Internal
arc fault testing of gas insulated metal enclosed MV switchgear”. CIRED 2005 proceedings. Turin, 6-9 June 2005.
["EMC"]: Electromagnetic Compatibility, "EMC". Practical Installation Guidelines. Groupe Schneider.

http://www.global-download.schneider-electric.com
[Feitoza2010]: S. Feitoza, "Guidelines for the use of simulations and calculations to replace some tests specified in
international standards". COGNITOR Guide 2010.
http://www.cognitor.com.br/GUIDE_Simulations_v0_October2010.pdf
[Hilten1978]: Hilten, John, Vezzetti, Carol, Mayo-Wells, J. Franklin, Lederer, Paul, Experimental Investigation of
Means for Reducing the Response of Pressure Transducers to Thermal Transients, NBS Tech Note 961, January
1978.
[IEST]: Shock and Vibration Transducer Selection, Institute of Environmental Sciences and Technology, IEST RP-
DTE011.1, Sec. 7.9, Oct. 2002.
[Puers1993]: Puers, R. "Capacitive sensors, When and how to use them". Sensors and Actuators A, 37-38 (1993)
93-105 93-105.
[Shannon1998]: C. E. Shannon, "Communication in the presence of noise", Proc. Institute of Radio Engineers,
vol.37, no.1, pp.10–21, Jan.1949. Reprint as classic paper in: Proc.IEEE, vol.86, no.2, (Feb.1998).
[Stein1992]: Stein, P. K., The Unified Approach to the Engineering of Measurement Systems, Stein Engineering
Services, Phoenix, AZ, April 1992.
[Walter2004]: P.L. Walter, “Shock and Blast Measurement - Rise Time Capability of Measurement Systems?”,
Technical Note, PBC Group (www.pcb.com), 2004.
[Walter2005]: P.L. Walter, “Introduction to Air Blast Measurements. Part III: Guaranteeing that Validated Pressure
Measurements are Aquired". Technical Note, PBC Group (www.pcb.com), 2005.
[Walter2010]: P.L. Walter, “Measuring Static Overpressures in Air Blast Environments”, Technical Note, PBC
Piezotronics (www.pcb.com), 2010.
Page 142
ANNEX C COTTON INDICATORS: ENERGY ABSORPTION AND

FLAMMABILITY
C .1 In troduction
Two primary reasons for arc fault testing are safety and environmental concerns with SF 6 (if appropriate).
Calculations and simulations are an important part of understanding the internal arc phenomena to: better design
the equipment (for safety in case of arc flash) and to reduce the need for testing with SF6 (for reducing
environmental impact).
C .2 Testing for resistance against internal arcing

Cotton indicators are the primary tools for pass-fail of internal arcing tests. They are designed to simulate
personnel clothing. They are positioned in relative proximity to the switchgear equipment (for example at 30 cm in
case of testing for authorized personnel access only and at 10 cm in case of general public access equipment). If
the cotton indicators catch fire during the simulated arc test the test fails. Withstand testing of the switchgear to
internal arcing is typically accomplished by following one or more of the following standards:
 IEC 62271-200 (Ed. 2.0, 2011) Annex A - Internal fault. Method for testing the metal enclosed switchgear
and controlgear under conditions of arcing due to an internal fault
 Electrical Equipment Manufacturers Association of Canada published EEMAC G14-1 in 1987

 Type A – arc-resistant construction at the front only
 Type B – arc-resistant construction at the front, back, and sides
 Type C – arc-resistant construction at the front, back, and sides, and between compartments
 IEEE C37.20.7-2007 IEEE Guide for Testing Medium-Voltage Metal-Enclosed Switchgear for Internal
Arcing Faults includes
 Type 1 – similar to EEMAC Type A above

 Type 2 – similar to EEMAC Type B above
 Annex A addresses suffixes “B” and “C”
 Type 1C – Type 1, but also with arc-resistance designs or features between adjacent
compartments
 Type 2B – Type 2 with LV instrument compartment door open – relay and maintenance
personnel survive
 Type 2C – Type 2 with arc-resistance features between adjacent compartments –
switchgear survives with minimum damage
 Type 2BC – The ultimate in protection – combines types 2B and 2C
 These flammable cotton indicators are positioned around the switchgear to detect the escape of hazardous
gases, plasma, arcing, foreign objects, etc. from the equipment tested.
 The following Pass/Fail criteria are used:

 Door, covers, etc. do not open. Bowing or other distortion is permitted except on those surfaces
which are to be used for mounting relays, meters, etc.
Page 143
 That no parts are ejected into the vertical plane defined by the accessibility type
 There are no openings caused by direct contact with an arc
 That no cotton indicators ignite as a result of escaping gases or particles
 That all grounding connections remain effective
Two types of cotton indicators are used: thick (so called Baumville, 150 gr/m2) and thin (so called Filtex, 40 gr/m2).
Flammability of cotton indicators is a complex phenomenon therefore understanding the repeatability and
consistency of testing using cotton samples is important. This WG performed some testing on the two kinds of
cotton samples.
C .3 Appara tus o verview

Flammability testing of the cotton indicators was performed at WPI (Worchester Polytechnic Institute in
Massachusetts) in the controlled environment of a laboratory setting. A cone calorimeter was used. It is a fire
instrument based on the principle of oxygen consumption calorimetry. The core of the instrument is the cone-
shaped radiant electrical heater, which irradiates a horizontal sample. The heat flux level is set and controlled via a
temperature controller, which uses three thermocouples attached to the heating element (see Figure C-1).
An intermittent spark igniter, powered by a 10,000 V transformer and a 3 mm gap, located 13 mm above the
sample, provides the piloted ignition. The following properties are measured: rate of heat release rate, total heat
released, effective heat of combustion, time to ignition, mass loss rate, total mass loss, and smoke obscuration.
C .4 Ove rview of the test procedure

The heating element is turned on and allowed to heat to the proper heat flux. The heat flux is measured using a
calibrated Heat Flux Gauge. The sample is then placed on the load cell (Figure C-2). The spark is placed and the
shutter opened. Observations for ignition and flameout are made.
C .5 Test results
Although the WG has not performed exhaustive testing, several different samples were tested for time to ignition at
different levels of the heat energy applied in the calorimeter. Since the reason for placing cotton samples in the
actual internal arc testing is to simulate the human clothing and the degree of heat exposure that could be harmful
to personnel, the samples should respond consistently to heat exposure and ignite faster at higher heat flux levels.
Page 144
Figure C-1: Overall test apparatus.
Figure C-2: Test specimen mounted on the edge frame and specimen under test.
Page 145
The degree of harm that heat exposure could cause to humans is often calculated by the so called Stoll limit. The
5
Stoll limit defines the energy density which causes a second degree burn on humans. It is expressed as:
.
= ∙ (C-1)
where in kJ/m2 is the energy density, is the Stoll constant = 50.204 kW/m2, and is the time to reach
the maximum temperature .The time to reach depends on many factors such as the arc, arc
configuration, arc electrodes, gas, flammable/ combustible materials, calorimeters, distances, dimensions, etc. A
typical result in reaching the Stoll limit is shown in Figure C-3 as Temp rise (Δ ) versus time [Schau2011]. When
the Stoll limit crosses over the test result for a given case a second degree burn occurs to a human.
Figure C-3: Stoll limit and test result as a function of time.
One has to note that the Stoll limit is a dynamic function of time, i.e. even a smaller incident heat energy applied for
a longer period of time can cause burns as much as a high incident energy applied for a short time.
Figure C-4, Figure C-5 and Figure C-6 summarize results of the testing for the two types of the cotton samples.
2
They show the time to ignition (in seconds) as a function of the heat flux density in kW/m . Individual points show
the specific sample tests whereas the line is a linear regression function showing the trend. As expected the higher
the heat flux density applied the shorter the time to ignition. However, it is also clear from the testing that there is
visible scatter of results reaching up to 30-40 %. Considering that the safety is the primary objective of the testing
the lowest data point should always be taken into account to account for the worst case scenario. This scatter
sheds some doubts on the repeatability of the tests both in the laboratory environment as described above where
the test conditions are highly controllable as well as with the actual switchgear subjected to internal arcing tests
where the test conditions, particularly the arc behavior and the exact amount of arc energy are impossible to
control.
5
Definition from [NFPA 70E-2012]: A second degree burn is possible by an exposure of unprotected skin to an electric arc flash above the
incident energy level of 5 J/cm2 (1.2 cal/cm2).
Page 146
Figure C-4: Time to ignition for thin fabric (actual test points and regression line are shown).
Figure C-5: Time to ignition for thick fabric (actual test points and regression line are shown).
Page 147
Time to ignition
3,5
2,5
Thin
Time (sec)
2 Thick
1,5 Regression1
Regression 2
1
0,5
0
50 70 90 110
Heat Flux (kW/m2)
Figure C-6: The same time to ignition for both thin and thick fabric tested repeated at different dates with
the same apparatus and samples as in Figure C-4 and Figure C-5.
It is advisable to compare the testing of the cotton samples used in the switchgear internal arcing tests with the
electrical safety standard prescribed for protection against arc flash. One has to remember at this point that the
internal arcing tests are prescribed for equipment whereas the safety standards such as IEC/EN 61482-1-2 (Ed.
1.0, 2007), NFPA 70E (2012), IEEE 1584 (2002) and others deal with the personal protection equipment (PPE) and
calculation of safety zones at different PPE applied to humans. The “connection” between the two types of
standard is the cotton samples used in switchgear testing that are supposed to simulate the human clothing.
Before comparing the test results we have to recall the heat energy density and heat power density defined for
PPE in the IEC 61482-1-2, NFPA 70E standards.
Table C-1: Summary of NFPA 70E (2012) hazard/risk categories and IEC/EN 61482-1-2 (Ed. 1.0, 2007)
classes.
Page 148
Table C-2 and Table C-3 below represents the results of the cotton sample testing and compare with the arc flash
risk categories prescribed in NFPA 70E/IEEE 1584 standards.
Applied
Heat
Flux Thin - test 1 Thick - test 1
NFPA 70 E NFPA 70 E
Hazard Hazard
kW/m2 J/cm2 cal/cm2 Category J/cm2 cal/cm2 Category
60 8,1 1,94 1 13,32 3,18 1
70 4,83 1,15 1 12,67 3,03 1
80 5,28 1,26 1 11,12 2,66 1
90 3,87 0,92 1 9,81 2,34 1
Table C-2: Comparison of cotton sample flammability testing from Figure C-4 and Figure C-5 with the arc
flash risk categories per NFPA-70E (2012) standard.
Thin Thick
NFPA 70 E NFPA 70 E
Hazard Hazard
J/cm2 cal/cm2 Category J/cm2 cal/cm2 Category
6,12 1,46 1 17,64 4,22 2
6,93 1,66 1 14,35 3,43 1
8,24 1,97 1 7,52 1,80 1
11,25 2,69 1 17,55 4,19 2
9,4 2,25 1 12 2,87 1
Table C-3: Comparison of cotton sample flammability testing from Figure C-6 with the arc flash risk
categories per NFPA-70E (2012) standard.
In Table C-2 all the results of the cotton sample testing fall in the Hazard category 1, which is to be expected and
agrees with the premise of NFPA 70E in relation to hazard category 16. In Table C-3 the situation is different when
the tests were repeated on the same types of cotton samples. The scatter of the results means that two of the thick
cotton samples (line 1 and line 4 on the right hand side of the Table C-3) withstand the energy densities above
Hazard category 1 without igniting. This means, that the switchgear internal arc test with such a cotton sample can
pass the flammability criterion and yet produce heat energy levels unsafe for Hazard category 1.
There could be several reasons for the scatter of the flammability test result inconsistency of the cotton samples
(textile) and / or the fibers used for producing the samples, humidity levels in the air where samples were stored
prior to calorimeter testing, pollution of samples, or other chemical reactions (oxidation, exposure to a different
6
In NFPA 70E Hazard category 1 defines Arc-Rated Clothing with a minimum arc rating of 4 cal/cm2.
Page 149
atmosphere with different gases, etc.) that the samples might have been exposed to. Irrespective of the reasons,
cotton samples used for internal arc testing are exactly the same as those used in the laboratory testing and are
not controlled for these possible factors. Caution has to be taken in interpreting the results of the internal arc testing
with cotton samples as indicators and translate these results to the arc flash protection requirements described in
safety standards. More work has to be done to investigate the correlation between the flammability of the cotton
samples, incident heat energy densities, and arc flash protection requirements.
C .6 Summary
Internal arc tests on switchgear equipment rely mainly on the cotton indicators placed in front of the equipment
during testing. If the cotton samples catch fire the test fails. Therefore understanding the flammability of the cotton
samples is important in understanding the test outcomes. Although the cotton samples realistically reflect the kind
of clothing that a person might wear around the switchgear equipment, using the cotton samples as testing
indicators causes some statistical uncertainty. On the other hand, the protective clothing designed for personnel is
2
classified by its resistance to the energy density that switchgear internal arcing could produce (Joules/m or
2
cal/cm ). Although there was limited data available, the scatter of the results may cause the cotton samples to
withstand applied energy densities above what would be considered safe (hazard category 1) without igniting.
REFERENCES:
[Schau2011]: Holger Schau, Herbert Bessei, The Influence of Fuses on Arcing Fault Energy And Personal
Protective Clothing Required, 9th INTERNATIONAL CONFERENCE on ELECTRICAL FUSES AND THEIR
APPLICATIONS, ICEFA 2011.
Page 150
ANNEX D ROOM PRESSURE CALCULATION USING CFD EXAMPLES

D .1 In troduction
Experience of room pressure calculations highlights some main points about how to handle pressure relief
openings or how to position them. There are some concerns about the first pressure peak. The first pressure peak
is generally the main reason for damage in the room. Taking account of these issues enables the switchgear room
to be made safer in the case of an arc fault. Following all these guidelines is not always possible, but this
information can be helpful to prevent greater damage in the case of an arc fault.
D .2 Pressure development and p ressure relie f

As a result of the internal arc, the increased pressure spreads out more or less spherically from the ignition location
and will be reflected from walls, the ceiling, the ground and other objects in the room. The first pressure wave, due
to the spherical spread of the pressure causes a first high pressure peak at the near walls. This can only be
influenced by a pressure relief opening located immediately at the location. Otherwise a pressure relief opening
has no influence on the first pressure peak (Figure D-1 and Figure D-2).
First pressure peak
Figure D-1: First pressure peak 36 mbar at time 20 ms in a closed switchgear room.
First pressure peak
Figure D-2: First pressure peak 36 mbar at time 20 ms in the same switchgear room with an opening
(green).
The first pressure peak can be decreased by increasing the distance of switchgear from the walls (Figure D-3 and
Figure D-4). Thus, the surface area of the pressure wave increases when the wave strikes the wall. The first
Page 151
pressure peak has a high amplitude, but there is only a small area of impact at the wall. Thus this first pressure
peak often is less critical than a pressure on the whole wall. The peak acts rather like a point load on the wall. (In
this case there is also the possibility to integrate the pressure at this time point over a wall area to get the
equivalent force on the whole wall and to use it to evaluate the wall load.)
First pressure peak
Figure D-3: Switchgear directly at the wall - first peak 146 mbar after 25 ms.
First pressure peak
Figure D-4: Switchgear with distance to the wall - first peak 70 mbar after 25 ms.
Average pressure at t = 25 ms from both simulations is about 5 mbar at the wall behind the switchgear.
The pressure relief in the room cannot be effective until the first pressure wave reaches any opening. In special
cases an increase of the pressure relief opening cross section does not reduce the pressure amplitude, but the rate
of decrease. (For instance if the pressure peak amplitude is higher than the later existing room filling pressure).
In long rectangular rooms the released gas tends to generate longitudinal pressure waves. Pressure relief
openings are less effective if they are located on the long sides. In this case the pressure wave passes across the
opening. In long rectangular rooms it is recommended that the pressure relief openings are located at the small
sides. In that case the velocity of the pressure waves spreads out the gases perpendicular to the opening area and
so the pressure decreases faster.
In long or large rooms it is recommended to place some more pressure relief openings at different distributed
locations. In large rooms one or more pressure waves may spread out. These waves oscillate in different
directions, depending on the room geometry, and pass the pressure relief openings only temporarily.
Page 152
Pressure relief
Figure D-5: Switchgear room with 6 openings at the long side. The first pressure peak is 3 mbar at 15 ms.
Pressure relief
Figure D-6: Switchgear room with 4 openings at the long side and 2 openings at the small. The first
pressure peak is 3 mbar at 15 ms.
With only 4 openings (2 openings at the small sides and 2 openings at the long sides) there is a longitudinal
pressure wave, where the first pressure peak is 3 mbar at 15 ms, see Figure D-6. In this case, the area of the
openings is reduced to 66 % of the opening area in the case described in Figure D-5 with 6 openings at the long
side. However, the pressure relief over time is nearly the same for the two cases.
D .3 How to determine worst case scenarios for simulations .

The selection of the damaging panel (position of arc fault) should be the worst case, i.e. the case with the highest
pressure impact in the room. Therefore the following statements help in selecting the worst position of the panel:
In general a panel having the longest distance to a pressure relief opening should be selected, so that the first
pressure relief time is late. Further, the first pressure peak is not decreased by the pressure relief opening.
Page 153
Figure D-7: Panel with the greatest distance to the opening has an arc fault.
The first spherically-spreading pressure wave causes a first pressure peak on the wall and the ceiling. The static
pressure due to reflection is a maximum if the panel is located in a room corner. In that case the pressure wave is
reflected from the ceiling and the two adjacent walls and has its maximum pressure value (Figure D-7).
Pressure peak of ceiling corner
Figure D-8: Pressure in the left front ceiling corner (MP5) is 6.6 mbar, more than the first pressure peak.
There is a longitudinal pressure wave and 6 openings at the long side of the switchgear room. The first pressure
peak is 3 mbar at 15 ms, but there is a greater pressure peak of 6.5 mbar in the ceiling corner front left (MP5) after
the time delay of 70 ms due to the time the wave needs to reach that point (Figure D-8).
Switchgear with a pressure relief duct at the top, discharging into the room, are to simulate with the highest number
of channel openings. So the pressure can flow fast from the duct through many openings into the room. To
simulate a realistic first pressure peak, the damaged panel should be located directly at a channel opening. Then a
share of the gas flows directly through the opening into the room without direction change (Figure D-9).
Page 154
Realistic first pressure peak
Panel directly located under a channel opening
Panel has the greatest distance to the opening
Figure D-9: Switchgear with a channel at the top and absorber openings in this channel (green).
Figure D-10: Example of a power curve for a switchgear with 25 kA and a peak value of 62 MW at 10 ms.
In determining the arc fault power, a three phase arc fault having the highest power peak and so the highest energy
exchange is assumed. The simulation assumes (unless otherwise agreed) a three phase circuit breaker arc fault.
Further a generator remote short circuit is assumed. This current becomes constant after the decay of direct
current terms after the first milliseconds (as in the arc fault test) (Figure D-10).
Page 155
D .4 Summary
The issues discussed in this section are based on many arc fault simulations for customer switchgear rooms and
different switchgear. In an arc fault accident, the pressure must escape into the environment to prevent damage to
the switchgear building. It is necessary to locate pressure relief openings to let the pressure escape from the
housing. In these projects the cross section of the pressure relief opening and its position are the main concern for
the customer. The arc fault pressure simulations show how to handle the pressure relief openings. Using the above
information, users may also influence the maximum value of the first pressure peak after the arc fault beginning.
Page 156
ANNEX E EFFECTS OF NEUTRAL GROUNDING ON THE INTERNAL

ARC
E.1 In troduction
For internal arc testing two different approaches exist, regarding the grounding (BR: earthing) of the supply circuit
neutral:
 IEC 62771-200 (ed.2, 2011) prescribes in clause A5.1(Three phase tests):

"the neutral point of the supply circuit may be either isolated or earthed through an impedance, in such a
way that the maximum earth current is less than 100 A. In this situation, the arrangement covers all
situations of neutral treatment".
 Alternatively, in IEEE C.37.20.7 cl.5.2.6 it is stated that:
 "the neutral of the supply system must be grounded or connected to the ground bus of the switchgear
assembly by a separate bus. If the neutral is grounded, the ground bus of the switchgear assembly must
be grounded. If the neutral of the supply system is connected to the switchgear assembly by a separate
bus, the switchgear assembly may be isolated from ground as required by the laboratory" followed by a
note stating that: "this requirement recognizes that test laboratories may not allow the flow of intentional
ground fault current due to safety and/or instrumentation concerns".
In other words, where IEC prescribes a floating situation (without significant neutral current), IEEE prescribes an
"extended neutral", with a low impedance, where significant neutral current may develop. It is unclear, how the
grounding regime affects the impact of an internal arc.
The difference between both test-circuits is shown in Figure E-1: switch NS closed is the "IEEE case" whereas NS
open is the "IEC test case".
G: generator;
MB: master breaker;
MS: making switch;
L: current limiting reactor;
PT: test transformer;
U, I: measurement of voltage and current;
TO: test-object;
NS: switch in extended neutral.
Figure E-1: Test circuit “IEEE case” and “IEC case” defined by NS.
E.2 Test Comparison

In order to investigate the consequences of both test-circuits, tests were carried out under identical conditions, the
only difference being the closed/open position of the switch NS in Figure E-1.
3
The test object (TO) was an arc compartment of 0.36 m , filled with air at ambient pressure with an open exhaust
2
(225 cm diameter) to the environment, refer to Figure E-2. The electrodes (Ø 25 mm, length 50 mm three in line at
110 mm distance), were replaced after each test. A steel burning plate (connected to the grounding point via an
2
120 mm Litze) was placed around the conductors at a distance of 110 mm from the electrodes.
Page 157
Figure E-2: Sketch of test object (TO).
These two positions of NS were tested at 24 kV driving voltage. The three-phase test-current was set at 18 kA, the
arc duration was 0.3 s. Arc initiation was always three phase. The neutral impedance was set and verified for 18 kA
single phase fault current in case of the extended neutral (switch NS closed, “IEEE case”).
In Figure E-3, the currents in the 4 relevant conductors (three phases and neutral) are plotted for the test with the
extended neutral situation at 24 kV.
From the test results:
 it is obvious that the extended neutral current has an RMS value below 500 A, i.e. less than 3 % of the
phase current value. The current measurement system has been checked by verifying that the sum of the
three phase currents is within 5 % of the neutral current.
 Arcing voltages are about 300 Vrms to earth
 The measured pressures for both situations (floating and neutral extension) are compared in Figure E-4.
As can be seen, the difference in the pressure build-up pattern remains within the limits of reproducibility.
 The arc energy is 4.01 MJ in the situation of neutral extension where 3.95 MJ was measured in the floating
situation.
Page 158
50
-50
-50
-50
0
current in extended neutral (50 kA/div)
-50
0
current in extended neutral (2 kA/div)
-2
250 300 350 400 450 500 550 600
Figure E-3: Currents in extended neutral situation. Top three traces: phase currents; lower two traces:
Current in neutral (at same vertical scale as phase current and enlarged vertically).
1.8
ungrounded
1.6
1.4 extended neutral

pressure (b), current (100 kA)
1.2
0.8
0.6
current
0.4
0.2
260 280 300 320 340 360 380

time (ms)
Figure E-4: Pressure rise with and without extended neutral present.
Subsequently an internal arc test was also performed in the same test laboratory at 63 kA -0.5 s. The test object
was a large air-filled switchgear cubicle, with an extended neutral (with current limited to 20 kA because of test-
plant limitations). This test also confirmed that only a small fraction of the phase current returns through the
extended neutral (in the mentioned 63 kA case the neutral current was below 0.4 kA, i.e. less than 1 % of the
phase value).
E.3 Discussion
Earlier observations from several internal arc tests in various laboratories suggest that the neutral return current (in
the case of the low-impedance extended neutral) can have values up to about 20 % of the three-phase current.
This may then lead to currents across the enclosure from arc foot points to the neutral bus connection, causing
dynamic / thermal stresses to bolted connections, hinges of doors and other parts possibly not designed to
withstand high current flows. This may result in failure to pass the test criteria.
Page 159
However, in the examples presented above, a significant neutral current is absent due to the balance in the supply
circuit (phase currents almost equal in amplitude). If the balance is disturbed in a severe way, e.g. by initiating the
arc between two phases only leaving the non-used phase de-energized, a large neutral current results (see Table
E-1 third test situation). In Table E-1, the test results are summarized, including the measured deviation (in %) of
each phase current with respect to the measured RMS value averaged over the phases.
current duration arc neutral phase current unbalance (in %) current in arc energy max
initiation (rel . to average RMS value of 3 ph) neutral pressure
kA s U V W kA MJ bar (rel)
18 0.3 3 phase floating -0.4 0.7 -0.4 - 3.95 1.79
18 0.3 3 phase extended -0.4 0.2 0.2 0.5 4.01 1.73
18 0.3 2 phase extended - 55.9 44.1 16.1 1.33
63 0.5 3 phase extended (20 kA max) 1.0 -0.1 -1.0 0.4 44.70 -
Table E-1: Test parameters and results.
So, the severe neutral currents that have been observed during 3-phase internal arcing tests, may result from an
unbalance of the supply circuit. Because the interior of the arcing tank is a good conductor in the presence of the
arc(s), it is obvious that a neutral current will arise in an unbalanced situation.
Unbalanced test-circuits are common. Some test stations have the possibility to tune the impedance per phase,
some cannot. If the latter is the case, supply circuits will be unbalanced, as mutual inductances create unbalance in
the three-phase circuits that otherwise would have identical positive sequence impedance. In the test-cases of
Table E-1, the phase impedances were tuned individually to have an unbalance below 1 %.
Discussion is required on the degree of unbalance that circuits in service might have. Tests with well-balanced test
circuits may be less severe for the switchgear having an extended neutral, because of the absence of considerable
current across the tank surface. On the other hand, every ampere that is conducted away through the neutral, will
not contribute to the arc-energy anymore.
E.4 Summary
It can be concluded that in case of an extended neutral, the severity of internal arc tests depend heavily on the
balance of the supply circuit.
In low-voltage (LV) testing significant neutral current is a well-known [Dunki1972] phenomenon. In LV testing, or in
MV switchgear testing with a supply voltage far lower than rated switchgear voltage, the arc voltage may be of the
same order as the supply voltage. Due to differences in (momentary) arc voltages, a considerable neutral current
may flow. This unbalance in the arc voltages can occur even in perfectly balanced supply circuits.
REFERENCES:
[Dunki1972]: J.R. Dunki-Jacobs, "The Effect of Arcing Ground Faults on Low-Voltage System Design", IEEE Trans.
On Ind. Appl., Vol. IA-8, Nr.3, 1972, pp. 223-230”.
Page 160
ANNEX F FIGURES FROM SENSITIVITY ANALYSIS

F.1 In troduction
Chapter 3 covered the sensitivity analysis of the pressure rise calculated with the basic model. ANNEX F is an
extension of Chapter 3. It consists of two parts:
Section F.2 provides a series of figures illustrating the influence of different parameters on the pressure rise curve.
The simulations are performed for both SF6 and air, utilizing the basic model, as explained in Section 3.2.3.
Section F.3 contains the equations for proving the overshoot condition (3-5) introduced in Section 3.3.2. The
definition of constant is also provided.
F.2 Sensitivity analysis fo r selective cases

In order to illustrate the impact of different parameters on the pressure curve, several simulations are performed
with a chosen parameter varying within a range from half to double of the reference value for both air and SF6. The
reference value is the original parameter value defined by a test case (Table F-1). The results, shown in the figures
below, are the families of pressure curves which include the simulation of the reference case (black bold line), the
pressure curve in the arc compartment (continuous lines) and the pressure curves in the exhaust compartment
(dashed lines). In addition, the ratios of variation of , , and relative to the reference value are
shown by the lines with markers: each marked point corresponds to one simulation and the crossing point
corresponds to the reference value. The abscissa is the relative variation of the studied parameter and the
ordinate is the ratio of the calculated result to the reference value.
case gas 1 1 2 2
3 2 ,1 ,1 3 2
[m ] [m ] [A] [V] [MPa] [MPa] [m ] [m ] [Hz] [s]
1 2.00 1.00 SF6 0.390 0.0060 65.0 450 1 0.49 1.00 50 0.34
2 0.90 1.00 SF6 1.217 0.0620 25.0 1800 3 0.17 1.50 60 1.34
3 0.40 0.70 Air 0.284 0.0445 24.0 364 3 0.12 0.12 1.400 0.100 50 1.18
4 0.55 1.00 Air 0.416 0.0490 38.0 320 3 0.12 0.20 0.730 0.400 50 1.18
5 0.80 0.80 SF6 0.509 0.0046 14.0 366 1 0.14 0.28 1.275 0.012 60 1.05
6 0.40 1.00 Air 1.190 0.1200 45.7 430 3 0.10 0.08 1.880 0.409 50 0.10
7 0.60 0.80 SF6 0.100 0.0020 20.0 500 1 0.50 1.00 1.000 0.010 50 0.30
8 0.35 0.90 SF6 0.300 0.0250 19.6 1200 3 0.10 0.33 50 1.00
9 0.65 0.75 SF6 0.300 0.0250 19.9 900 3 0.10 0.05 50 1.00
Table F-1: Reference parameters used for the simulations.
Page 161
0.8 0.8
AIR 0.8 Ex SF6 0.8 Ex
0.4 0.4
0.9 0.4 Ex 0.9 0.4 Ex
0.52 0.52
0.8 0.52 Ex 0.8 0.52 Ex
0.64 0.64
0.7 0.64 Ex 0.7 0.64 Ex
0.76 0.76
0.6 0.6
pressure [MPa]
pressure [MPa]
0.76 Ex 0.76 Ex
0.88 0.88
0.5 0.5
0.88 Ex 0.88 Ex
1 1
0.4 0.4
1 Ex 1 Ex
0.3 1.12 0.3 1.12
1.12 Ex 1.12 Ex
0.2 1.24 0.2 1.24
1.24 Ex 1.24 Ex
0.1 1.36 0.1 1.36
1.36 Ex 1.36 Ex
0 1.48
0 1.48
0 0.5 1 1.5 1.48 Ex 0 0.5 1 1.5 1.48 Ex
1.6 1.6
time [s] time [s]
1.6 Ex 1.6 Ex
3.00 2.50
2.50
2.00
2.00 t burst t burst
1.50
s burst s burst
1.50
t max t max
1.00
1.00 p max p max
0.50
0.50
0.00 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-1: Pressure development for different –factors (case 5).
Page 162
1
14000 14000
AIR 14000 Ex SF6 14000 Ex
7000 7000
0.9 7000 Ex 0.9 7000 Ex
9100 9100
0.8 9100 Ex 0.8 9100 Ex
11200 11200
0.7 11200 Ex 0.7 11200 Ex
13300 13300
0.6 0.6
pressure [MPa]
pressure [MPa]
13300 Ex 13300 Ex
15400 15400
0.5 0.5
15400 Ex 15400 Ex
17500 17500
0.4 0.4
17500 Ex 17500 Ex
0.3 19600 0.3 19600
19600 Ex 19600 Ex
0.2 21700 0.2 21700
21700 Ex 21700 Ex
0.1 23800 0.1 23800
23800 Ex 23800 Ex
0 25900
0 25900
0 0.5 1 1.5 25900 Ex 0 0.5 1 1.5 25900 Ex
28000 28000
time [s] time [s]
28000 Ex 28000 Ex
3.00 2.50
2.50
2.00
1.50
s burst s burst
1.50
t max t max
1.00
1.00 p max p max
0.50
0.50
0.00 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
2 Figure F-2: Pressure development for different fault currents [A] (case 5).
Page 163
366 366
AIR 366 Ex SF6 366 Ex
183 183
0.9 183 Ex 0.9 183 Ex
237.9 237.9
0.8 237.9 Ex 0.8 237.9 Ex
292.8 292.8
0.7 292.8 Ex 0.7 292.8 Ex
347.7 347.7
0.6 0.6
pressure [MPa]
pressure [MPa]
347.7 Ex 347.7 Ex
402.6 402.6
0.5 0.5
402.6 Ex 402.6 Ex
457.5 457.5
0.4 0.4
457.5 Ex 457.5 Ex
0.3 512.4 0.3 512.4
512.4 Ex 512.4 Ex
0.2 567.3 0.2 567.3
567.3 Ex 567.3 Ex
0.1 622.2 0.1 622.2
622.2 Ex 622.2 Ex
0 677.1
0 677.1
0 0.5 1 1.5 677.1 Ex 0 0.5 1 1.5 677.1 Ex
732 732
time [s] time [s]
732 Ex 732 Ex
3.00 2.50
2.50
2.00
1.50
s burst s burst
1.50
t max t max
1.00
1.00 p max p max
0.50
0.50
0.00 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-3: Pressure development for different arc voltages [V] (case 5).
Page 164
AIR SF6
1.8 1.8
1.6 25000 1.6 25000

12500 12500
1.4 16250 1.4 16250
1.2 20000 1.2 20000
pressure [MPa]
pressure [MPa]
23750 23750
1 27500 1 27500
0.8 31250 0.8 31250
35000 35000
0.6 38750 0.6 38750
0.4 42500 0.4 42500
46250 46250
0.2 0.2
50000 50000
0 0
0.05 0.15 0.25 0.35 0.45 0.55 0.05 0.15 0.25 0.35 0.45 0.55
time [s] time [s]
2.50 2.50
2.00 2.00
t burst t burst
1.50 1.50
s burst s burst
t max t max
1.00 1.00
p max p max
0.50 0.50
0.00 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-4: Pressure development for different fault currents [A], without overshoot (case 2).
Page 165
AIR SF6
2 2
1.8 0 1.8 0
0 Ex 0 Ex
1.6 1.6
0.52 0.52
1.4 0.52 Ex 1.4 0.52 Ex
pressure [MPa]
pressure [MPa]
1.2 1.05 1.2 1.05
1 1.05 Ex 1 1.05 Ex
1.57 1.57
0.8 1.57 Ex 0.8 1.57 Ex
0.6 2.09 0.6 2.09
0.4 2.09 Ex 0.4 2.09 Ex
2.62 2.62
0.2 0.2
2.62 Ex 2.62 Ex
0 3.14 0 3.14
0 0.1 0.2 0.3 0.4 0.5 3.14 Ex 0 0.1 0.2 0.3 0.4 0.5 3.14 Ex
time [s] time [s]
[s] [MPa/s] [s] [MPa] [s] [MPa/s] [s] [MPa]

0.000 0.021 29.18 0.113 1.85 0.000 0.108 5.56 0.300 1.44
0.524 0.027 22.55 0.115 1.85 0.524 0.111 5.38 0.295 1.43
0.785 0.021 29.18 0.113 1.85 0.785 0.108 5.56 0.300 1.44
1.047 0.030 20.11 0.136 1.84 1.047 0.116 5.16 0.296 1.42
1.571 0.027 22.11 0.138 1.81 1.571 0.118 5.09 0.298 1.42
2.094 0.025 23.57 0.129 1.83 2.094 0.115 5.21 0.300 1.43
2.618 0.024 25.20 0.131 1.85 2.618 0.108 5.53 0.300 1.44
3.142 0.021 29.18 0.113 1.85 3.142 0.108 5.56 0.300 1.44
1 Figure F-5: Pressure development for different fault initiation angles [radian] for slow process (case 7).
Page 166
AIR SF6
0.8 0.8
0.045 0.045
0.7 0.7
0.0225 0.0225
0.6 0.02925 0.6 0.02925
0.036 0.036
pressure [MPa]
pressure [MPa]
0.5 0.04275 0.5 0.04275
0.0495 0.0495
0.4 0.4
0.05625 0.05625
0.3 0.063 0.3 0.063
0.06975 0.06975
0.2 0.0765 0.2 0.0765
0.08325 0.08325
0.1 0.1
0.09 0.09
0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
time [s] time [s]
2.50 1.10
2.00
1.05
t burst t burst
1.50
s burst s burst
1.00
t max t max
1.00
p max p max
0.95
0.50
0.00 0.90
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-6: Pressure development for different time constants [s] (case 9).
Page 167
AIR SF6
3.5 3.5
0.34 0.34
3 0.17 3 0.17
0.221 0.221
2.5 0.272 2.5 0.272
pressure [MPa]
pressure [MPa]
0.323 0.323
2 2 0.374
0.374
0.425 0.425
1.5 1.5
0.476 0.476
0.527 0.527
1 1
0.578 0.578
0.5 0.629 0.629
0.5
0.68 0.68
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
time [s] time [s]
1.10
1.00
1.05 0.80
t burst t burst
s burst 0.60 s burst
1.00
t max t max
0.40
p max p max
0.95
0.20
0.90 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-7: Pressure development for different fault current durations [s] (case 1).
Page 168
0.284 0.284
AIR 0.284 Ex
SF6
0.284 Ex
0.142 0.142
0.35 0.142 Ex 0.35 0.142 Ex
0.1846 0.1846
0.1846 Ex 0.1846 Ex
0.3 0.2272 0.3 0.2272
0.2272 Ex 0.2272 Ex
0.2698 0.2698
0.25 0.25
pressure [MPa]
pressure [MPa]
0.2698 Ex 0.2698 Ex
0.3124 0.3124
0.3124 Ex
0.2 0.2 0.3124 Ex
0.355 0.355
0.355 Ex 0.355 Ex
0.15 0.3976 0.15 0.3976
0.3976 Ex 0.3976 Ex
0.4402 0.4402
0.1 0.4402 Ex 0.1 0.4402 Ex
0.4828 0.4828
0.4828 Ex 0.4828 Ex
0.05 0.5254
0.05 0.5254
0.05 0.15 0.25 0.35 0.45 0.5254 Ex 0.05 0.15 0.25 0.35 0.45 0.5254 Ex
0.568 0.568
time [s] time [s]
0.568 Ex 0.568 Ex
2.50 2.50
2.00 2.00
t burst t burst
1.50 1.50
s burst s burst
t max t max
1.00 1.00
p max p max
0.50 0.50
0.00 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-8: Pressure developments for different arc compartment volumes [m3] (case 3).
Page 169
AIR SF6
0.8 0.8
0.025 0.025
0.7 0.7
0.0125 0.0125
0.6 0.01625 0.6 0.01625
0.02 0.02
pressure [MPa]
pressure [MPa]
0.5 0.02375 0.5 0.02375
0.0275 0.0275
0.4 0.4
0.03125 0.03125
0.3 0.035 0.3 0.035
0.03875 0.03875
0.2 0.0425 0.2 0.0425
0.04625 0.04625
0.1 0.1
0.05 0.05
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
time [s] time [s]
2.50 3.00
2.50
2.00
t burst 2.00 t burst

1.50
s burst s burst
1.50
t max t max
1.00
p max 1.00 p max
0.50
0.50
0.00 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-9: Influence of arc compartment opening area [m2] (case 8).
Page 170
AIR SF6
0.8 0.8
0.33 0.33
0.7 0.7
0.165 0.165
0.6 0.2145 0.6 0.2145
0.264 0.264
pressure [MPa]
pressure [MPa]
0.5 0.3135 0.5 0.3135
0.363 0.363
0.4 0.4
0.4125 0.4125
0.3 0.462 0.3 0.462
0.5115 0.5115
0.2 0.561 0.2 0.561
0.6105 0.6105
0.1 0.1
0.66 0.66
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
time [s] time [s]
2.00 2.50
2.00
1.50
t burst t burst
1.50
s burst s burst
1.00
t max t max
1.00
p max p max
0.50
0.50
0.00 0.00
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-10: Influence of arc compartment bursting pressure [MPa] (case 8).
Page 171
0.049 0.049
AIR 0.049 Ex
SF6 0.049 Ex
0.0245 0.0245
0.0245 Ex 0.0245 Ex
0.03185 0.03185
0.45 0.45
0.03185 Ex 0.03185 Ex
0.0392 0.0392
0.4 0.4 0.0392 Ex
0.0392 Ex
0.04655 0.04655
0.35 0.35
pressure [MPa]
pressure [MPa]
0.04655 Ex 0.04655 Ex
0.0539 0.0539
0.3 0.0539 Ex 0.3 0.0539 Ex
0.06125 0.06125
0.25 0.06125 Ex 0.25 0.06125 Ex

0.0686 0.0686
0.0686 Ex 0.0686 Ex
0.2 0.2
0.07595 0.07595
0.07595 Ex 0.07595 Ex
0.15 0.15
0.0833 0.0833
0.0833 Ex 0.0833 Ex
0.1 0.09065
0.1 0.09065
0.035 0.055 0.075 0.095 0.115 0.135 0.09065 Ex 0.035 0.055 0.075 0.095 0.115 0.135 0.09065 Ex
0.098 0.098
time [s] time [s]
0.098 Ex 0.098 Ex
2.50 2.50
2.30
2.00
2.10
t burst 1.90 t burst
1.50
s burst s burst
1.70
t max t max
1.00 1.50
p max p max
1.30
0.50
1.10
0.00 0.90
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-11: Influence of arc compartment opening area [m2] (case 4).
Page 172
0.409 0.409
AIR 0.409 Ex SF6 0.409 Ex
0.2045 0.2045
0.2 0.2045 Ex 0.2 0.2045 Ex

0.26585 0.26585
0.19 0.26585 Ex 0.19 0.26585 Ex
0.3272 0.3272
0.18 0.3272 Ex
0.18
0.3272 Ex
0.17 0.38855 0.17 0.38855
pressure [MPa]
pressure [MPa]
0.38855 Ex 0.38855 Ex
0.16 0.4499 0.16 0.4499
0.4499 Ex 0.4499 Ex
0.15 0.15
0.51125 0.51125
0.14 0.51125 Ex 0.14 0.51125 Ex
0.5726 0.5726
0.13 0.5726 Ex 0.13 0.5726 Ex
0.63395 0.63395
0.12 0.12
0.63395 Ex 0.63395 Ex
0.11 0.6953 0.11 0.6953
0.6953 Ex 0.6953 Ex
0.1 0.75665 0.1 0.75665
0.02 0.04 0.06 0.08 0.1 0.75665 Ex 0.02 0.04 0.06 0.08 0.1 0.75665 Ex
0.818 0.818
time [s] time [s]
0.818 Ex 0.818 Ex
1.10 1.10
1.05 1.05
t burst t burst
s burst s burst
1.00 1.00
t max t max
p max p max
0.95 0.95
0.90 0.90
0.00 0.50 1.00 1.50 2.00 2.50 0.00 0.50 1.00 1.50 2.00 2.50
1 Figure F-12: Influence of exhaust compartment opening [m2] (case 6).
Page 173
F.3 Pressure o vershoot equation

The pressure curve in the arc compartment after the bursting of the arc compartment relief device is described by
the ideal gas law (A-34) with the mass and the temperature obeying the energy balance equation (A-37). It is
assumed that no mass is flowing into the arc compartment. The mass flowing out of the arc compartment is
expressed by (A-22). If the volume of the exhaust compartment is infinite in comparison to the volume of the arc
compartment, then equation (A-26) can be used to calculate the pressure in the opening. These equations can be
simplified if the power input rate is considered to be constant ̇ ( ) = ̇ . The overshoot happens if the pressure at
the instant of pressure relief device bursting increases i.e. the derivative (slope) is positive. This slope can be
calculated using the ideal gas law (A-34).
( − 1) ( − 1)
= then ′ = ( + ′) (F-1)
′ ′
> 0 ⇔ >− (F-2)
After several transformations and using the above-mentioned equations, this condition can be expressed as
follows:
2 ( )
̇ √ +1 (F-3)
>
−1 ,
This is equivalent to equation (3-5) with the constant
2 ( )
√ +1 (F-4)
=
−1
/
The unit of is .
F.4 Summary
The impact of various parameters on the pressure curve in the arc and exhaust compartments is depicted in a
visual way. These include factor , fault current, arc voltage, fault initiation angle, time constant, arc duration, arc
compartment volume, arc and exhaust compartment opening area, and arc compartment bursting pressure. The
derivation of the overshoot condition for a slow process from the basic method equations is described.
Page 174
ANNEX G EFFECT OF REPLACING SF 6 WITH AIR ON BURN-

THROUGH
G.1 In troduction
Section 6.4.4 presented the qualitative approach for comparing the burn-through times in SF6 gas and in air. From
Table 6-15 it can be seen that most parameters indicate that burn-through times are shorter in SF6 than in air. This
annex takes a closer look at some of those parameters, and their dependency on the type of gas.
G.2 Arc voltage

As for , it is known that an electrode voltage drop region exists in an arc, which is a specific, quite thin, area
close to the electrode and provides a significant voltage drop as schematically shown in Figure G-1. Joule heat
provided in an electrode voltage drop region can be considered to be the most dominant energy source available
for melting through an enclosure, because it releases significant power in an area quite close to the enclosure.
Figure G-2 indicates electrode voltage drops for SF6, argon, and air for various electrode materials measured by a
laboratory test [Yokomizu1996]. As seen in Figure G-2, electrode voltage drops depend on electrode material much
more than on the type of gas, and they differ among the gases by only a few percent for the same electrode
material.
Arc Column
Figure G-1: Voltage drop along an arc column. Electrode region appears as a dominant power source for
melting the enclosure.
Page 175
Figure G-2: Electrode voltage drops for SF6, argon and air for various electrode materials at atmospheric
pressure [Yokomizu1996].
G.3 Heat of chemical reactions

The chemical reaction between vaporized metal and gas should be also investigated for both SF6 and air, because
it may cause a significant difference in the temperature that the enclosure would be exposed to, and in the
pressure rise in the enclosure. It is known that aluminum reacts with both SF6 and air (oxygen) and these reactions
provide an exothermic energy release. Thermo-chemical formulas between Al and SF6/O2 are as follows, which
shows that exothermic energies could be comparable (850 kJ for SF6, 837 kJ for air) if the same amount of
aluminum (1 mol) completely reacts with SF 6 and O2. However, the O2 content in air is only 21 %, therefore more
exothermic energy could be released in SF 6 than in air under the same filling pressure.
One mole of aluminum + SF6:
3 3
Al + SF6 → AlF3 + SF4 + 850 kJ (G-1)
2 2
One mole of aluminum + air (oxygen):
3 1
Al + 02 → Al2 O3 + 837 kJ (G-2)
4 2
G.4 Radius of Arc root

The radius of the arc root is one of the influencing factors for burn-through time, because it corresponds to the
concentration of input power at the erosion spot on the enclosure. Although some papers have reported
comparison of arc radii of wall-stabilized arcs with forced gas flow for different gases [Yokomizu JRED2000], none
was found for a free-burning arc in an enclosure such as an internal fault arc. Figure G-3 shows an analytical
temperature distribution of wall-stabilized gas arcs with forced gas flow as in circuit breaker arcs (wall radius 5 mm,
current 50 A, gas flow rate 5 liters/min). As seen in Figure G-3, it is known that the arc radius in SF6 is thinner than
Page 176
that in air in such a small current mainly due to the dissociation of SF6 molecules around 2,000 K. On the other
hand, Figure G-4 shows that the difference in arc radius (defined as the zone carrying 99 % of the total current)
becomes closer in a higher current condition even with forced gas flow. Consequently, arc radii could be
considered probably comparable between SF6 and air for high current free-burning internal arcs.
Figure G-3: Analytical temperature distribution of wall-stabilized arcs with forced gas flow [Yokomizu,
JRED2000]
Figure G-4: Conductive radius dependence on arc current [Yokomizu, JRED2000]
G.5 Velocity of Arc Motion

The first stage after the arc ignition is the displacement of the arc along the conductor until it meets a spacer which
could act as a barrier against its movement. This time is included in the total time to burn-through.
2∙ ∙ ∙
= [Lutz1983] (G-3)
∙ ∙
Page 177
0.28 ∙
= [Chu1985] (G-4)
∙
According to equation (G-3) and (G-4), replacing SF6 by air at the same pressure will lead to an acceleration of the
axial velocity because of the ratio of their gas densities. The arc will then spend less time axially blocked against
the partition barrier which has the effect of decreasing the theoretical time to burn-through.
Referring to the requirements of the standards the arc should be initiated in the vicinity of a partition furthest from
the point of injection (IEC 62271-203 (Ed. 2.0, 2011)). Due to this recommendation, it appears that the time where
the arc is moving axially is usually less than a tenth of millisecond for SF6 and should be shorter for air but without
a significant impact on the total time to burn-through.
Azimuthal motion speed will be also impacted by the change of gas, depending on the design of the barrier and
then the moving character of the arc; this could greatly affect the time of burn-through.
G.6 Pressure exerted on the tank wall

Subsection 6.2 deals with the evaluation of the pressure development for air or SF6. Burn-through is a
phenomenon that will normally happen well after bursting disk opening. It is clear that the pressure exerted by the
gas on the weakened structure plays an important role because the ablation process decreases the thickness of
the wall and this is combined with a loss of mechanical properties of the material.
Even if the reader must be careful and conduct its own pressure/temperature calculation for a particular case, one
can generally expect a higher pressure some hundreds of ms after disk opening inside an SF6-filled compartment
than in an air-filled compartment.
G.7 Summary
This annex presents discussions on some influencing factors for burn-through comparison between SF6 and air, as
a complement to Table 6-15 of Section 6.4.4. Even if the consequence of SF6 replacement by air is not obvious, we
theoretically could expect shorter burn-through times in SF6 than in air. This conclusion is mainly drawn from the
change of behavior regarding the heat of chemical reactions, the difference between azimuthal velocities, and the
change in the pressure exerted on the tank walls after the bursting disk opening.
Page 178
REFERENCES:
[Chu1985]: Chu, F.Y and Lutz, F. and Braun, J.M and Stuckless, H.A, “Effects of power arc faults in gas-insulated
substations”, CIGRE–Symposium, Brussels, Contribution 340-04, 1985.
[Lutz1983]: Lutz, F. and Chu, F. Y., “Burn-through of GIS Enclosure Due To Power Fault Arcs”, EEE Transactions
on Power Apparatus and Systems, 1983.
[Yokomizu1996]: Y Yokomizu and T Matsumura and R Henmi and Y Kito, “Total voltage drops in electrode fall
regions of SF6, argon and air arcs in current range from 10 to 20 000 A”, Journal of Physics D: Applied Physics, Vol
29.
[Yokomizu JRED 2000]: Yasunobu Yokomizu, "Physical mechanism of arc interruption in several gaseous
mediums", Trans on The Institute of Engineers on Electrical Discharges in Japan, vol. 166, 2000.
Page 179

602 Tools For The Simulation of The Effects of The Internal Arc in Transmission and Distribution Switchgear

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

602 Tools For The Simulation of The Effects of The Internal Arc in Transmission and Distribution Switchgear

Uploaded by

Copyright:

Available Formats

602

Tools for the Simulation of the Effects of the

TOOLS FOR THE SIMULATION OF THE

T ABLE OF CONT ENTS

TABLE OF CON TEN TS ....................................................................................................... 1

EXECUTIVE SUMMARY ............................................................................................................ 3

2 CALCULATION OF PRESSURE USING A BASIC MODEL.............................................. 14

3 SENSITIVITY ANALYSIS OF THE BASIC MODEL........................................................... 36

4 CALCULATION OF PRESSURE WITH CFD SOFTWARE................................................ 49

5 INTERNAL ARC TESTING ............................................................................................... 58

6 EFFECT OF INTERNAL ARC ON STRUCTURES ............................................................ 72

7 INTERNAL ARC SIMULATION REVIEW ........................................................................ 114

7.1 Introduction ................................................................................................................ 114

8 CONCLUSION ................................................................................................................ 120

ANNEX A EQUATIONS FOR PRESSURE RISE CALCULATION ....................................... 124

ANNEX B PRESSURE SENSORS: TYPES AND ACCURACIES........................................ 139

ANNEX C COTTON INDICATORS: ENERGY ABSORPTION AND FLAMMABILITY ......... 143

ANNEX D ROOM PRESSURE CALCULATION USING CFD EXAMPLES.......................... 151

ANNEX E EFFECTS OF NEUTRAL GROUNDING ON THE INTERNAL ARC .................... 157

ANNEX F FIGURES FROM SENSITIVITY ANALYSIS ....................................................... 161

ANNEX G EFFECT OF REPLACING SF6 WITH AIR ON BURN-THROUGH ....................... 175

Figure 1-1: A 13 kA electric arc moving between two conductors, at 5 cm distance.

Figure 1-2: Internal arc fault test.

I typically occur due to following reasons:

Approach /model Appropriate Application Limitations

 Doesn’t consider spatial non-uniformity of gas

Basic compartment and the exhaust in relation to the compartment volume.

 Doesn’t consider spatial non-uniformity of gas

Table 1-1: Model selection table.

1.2 Abb re viations

AIS air insulated switchgear

AISI American Iron and Steel Institute

CAD computer-aided design

CFD computational fluid dynamic

d.c. direct current

EEMAC Electrical Equipment Manufacturers Association of Canada

EMC electromagnetic compatibility

FDM finite difference method

FEA finite element analysis

FVM finite volume method

GIS gas insulated switchgear

IAC internal arc classification

NFPA National Fire Protection Association

PDE partial differential equation

PPE personal protection equipment

RMS root mean square

RMU ring main unit

SF6 sulfur hexafluoride

WPI Worchester Polytechnic Institute

Arc plasma: Thermal plasma generated by an arc.

Arc voltage: Voltage which appears between the electrodes of an arc.

Exhaust channel: See arc duct.

Joule heating: Heating caused by an electric current through a resistive material.

Overpressure: Pressure above ambient pressure within a pressurized enclosure.

Pressure withstand: Maximum pressure which can be withstood by an enclosure.

1.4 Referred standards

2 CALCULATION OF PRESSURE USING A BASIC MODEL

2.2 Equations of the basic model

Arc Compartment Exhaust Compartment Installation Room

2.2.2 Mass flow

2.2.3 Gas temperature

The corresponding temperature change in is:

The summation of all temperature changes Δ provides the temperature at time .