A3 14 (SC) 01 (B)

TOOLS FOR THE SIMULATION OF
EFFECTS OF THE INTERNAL ARC IN

TRANSMISSION AND DISTRIBUTION
SWITCHGEAR
WG A3.24, rev 11.1, January 26, 2014
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 2011
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 : (To be completed by CIGRE)

1
TOOLS FOR THE SIMULATION OF EFFECTS OF THE INTERNAL ARC IN TRANSMISSION AND DISTRIBUTION SWITCHGEAR, REV 11.1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
ISBN : (To be completed by CIGRE)
28
Page 1
TOOLS FOR THE

SIMULATION OF THE
EFFECTS OF INTERNAL ARC
IN TRANSMISSION AND
DISTRIBUTION SWITCHGEAR
TABLE OF CONTENTS
7
8
T AB L E O F CO NT E NT S ............................................................................................................ 2
EXECUTIVE SUMMARY ................................................................................................................. 5
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
1.1
1.2
1.3
INTRODUCTION ..................................................................................................................... 8
Overview ............................................................................................................................. 8
Definitions and Abbreviations ........................................................................................... 12
Referred standards ........................................................................................................... 15
2.1
2.2
2.3
2.4
2.5
2.6
2.7
CALCULATION OF PRESSURE USING A BASIC MODEL ................................................ 17

Introduction ....................................................................................................................... 17
Equations of the basic model............................................................................................ 17
Input parameters ............................................................................................................... 20
Application limits of the basic model ................................................................................. 23
Application of the basic model to selected test cases ...................................................... 25
Enhancements of the basic model ................................................................................... 36
Summary........................................................................................................................... 36
3.1
3.2
3.3
3.4
3.5
SENSITIVITY ANALYSIS OF THE BASIC MODEL ............................................................. 38

Introduction ....................................................................................................................... 38
Model parameters ............................................................................................................. 38
Model results .................................................................................................................... 39
Sensitivity analysis ............................................................................................................ 44
Summary........................................................................................................................... 50
4.1
4.2
4.3
4.4
CALCULATION OF PRESSURE WITH CFD SOFTWARE .................................................. 52

Introduction ....................................................................................................................... 52
Motivations for CFD .......................................................................................................... 52
Workflow in CFD ............................................................................................................... 53
Summary........................................................................................................................... 59
INTERNAL ARC TESTING ................................................................................................... 60

5.1
Introduction ....................................................................................................................... 60
5.2
Standardisation and Test Experience ............................................................................... 61
Page 2
5.3
Summary........................................................................................................................... 72
2
3
4
5
6
7
EFFECT OF INTERNAL ARC ON STRUCTURES ............................................................... 74

6.1
Introduction ....................................................................................................................... 74
6.2
Mechanical stress on switchgear due to the overpressure .............................................. 75
6.3
Mechanical stress on building walls due to overpressure ................................................ 85
6.4
Burn-through ................................................................................................................... 105
6.5
Summary......................................................................................................................... 115
8
9
10
11
12
13
INTERNAL ARC SIMULATION REVIEW ........................................................................... 118

7.1
Introduction ..................................................................................................................... 118
7.2
Information for the design review ................................................................................... 118
7.3
Comparison of actual and reference switchgear using design parameters ................... 119
7.4
Design Evaluation ........................................................................................................... 123
7.5
Summary......................................................................................................................... 123
14
CONCLUSION ..................................................................................................................... 124
15
16
17
18
19
20
ANNEX A: EQUATIONS FOR PRESSURE RISE CALCULATION ........................................ 128

A.1 Introduction ..................................................................................................................... 128
A.2 Equations for the basic model ........................................................................................ 128
A.3 Basic model .................................................................................................................... 133
A.4 Enhancement of the basic model ................................................................................... 135
A.5 Summary......................................................................................................................... 141
21
22
23
24
25
ANNEX B: PRESSURE SENSORS: TYPES AND ACCURACIES ......................................... 143

B.1 Introduction ..................................................................................................................... 143
B.2 Pressure sensors ............................................................................................................ 143
B.3 Accuracy of the measuring system ................................................................................. 144
B.4 Summary......................................................................................................................... 145
26
27
28
29
30
31
32
ANNEX C: COTTON INDICATORS: ENERGY ABSORPTION AND FLAMMABILITY ......... 147

C.1 Introduction ..................................................................................................................... 147
C.2 Testing for resistance against internal arcing ................................................................. 147
C.3 Apparatus overview ........................................................................................................ 148
C.4 Overview of test procedure ............................................................................................. 148
C.5 Results of the testing ...................................................................................................... 148
C.6 Summary......................................................................................................................... 154
33
34
35
36
37
ANNEX D: ROOM PRESSURE CALCULATION USING CFD EXAMPLES .......................... 155

D.1 Introduction ..................................................................................................................... 155
D.2 Pressure development and pressure relief ..................................................................... 155
D.3 How to determine worst case scenarios for simulations. ............................................... 157
D.4 Summary......................................................................................................................... 160
38
39
40
41
42
ANNEX E: EFFECTS OF NEUTRAL EARTHING ON THE INTERNAL ARC ........................ 161

E.1 Introduction ..................................................................................................................... 161
E.2 Test Comparison ............................................................................................................ 161
E.3 Discussion....................................................................................................................... 163
E.4 Summary......................................................................................................................... 164
43
44
45
46
47
ANNEX F: FIGURES FROM SENSIBILITY ANALYSIS.......................................................... 165

F.1 Introduction ..................................................................................................................... 165
F.2 Sensitivity analysis for selective cases ........................................................................... 165
F.3 Pressure overshoot equation .......................................................................................... 178
F.4 Summary......................................................................................................................... 178
Page 3
1
2
3
4
5
6
7
8
9
ANNEX G: EFFECT ON REPLACING SF6 WITH AIR ON BURN-THROUGH....................... 178

G.1 Introduction ..................................................................................................................... 178
G.2 Arc voltage ...................................................................................................................... 179
G.3 Heat of chemical reactions ............................................................................................. 180
G.4 Radius of Arc root ........................................................................................................... 180
G.5 Velocity of Arc Motion ..................................................................................................... 181
G.6 Pressure exerted on the tank wall .................................................................................. 182
G.7 Summary......................................................................................................................... 182
10
Page 4
EXECUTIVE SUMMARY
2
3
Recognizing the increasing role of commercial and home made modelling 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, can accurately predict stresses and can provide good performance
extrapolation where test data is available on similar designs (interpolation). The scope for pure performance
prediction (utilizing extrapolation) remains limited.
WG A3.24 has continued the analysis of the use of simulation as verification tools with a specific focus on internal
10
arc testing of Medium and High Voltage SF6 and air-filled equipment.
11
The main goal was to reduce the number of tests and - for environmental reasons - to eliminate testing where SF6
12
is released to the environment. The international standard for MV metal-enclosed switchgear, IEC 62271-200,
13
permits SF6 to be replaced by air, while the standard for HV Gas-Insulated switchgear, IEC 62271-203, allows the
14
extension of test results by calculation methods. IEEE standard C37.20.7 for internal arc testing does not address
15
SF6 at all, as the standard is only dedicated to air insulation. Figure 0-1 shows the test setup for internal arc testing.
16
17
Figure 0-1: Arc resistance test per IEC 62271-200.
18
19
20
WG A3.24 started work by reviewing the existing literature (100+ white papers and applicable IEEE and IEC
21
standards, a number of which are referenced later in this technical brochure), and collecting the test data from
22
numerous Internal Arc Tests. Test data was collected for more than 70 different cases; with tank sizes ranging
23
from small 5 l test tanks to large 1200 lGIS tanks, with fault currents ranging from 12 kA to 63 kA, with fault
24
durations ranging from 10 ms to 1.2 s, including single compartment and multi-compartment equipment and
25
including both SF6 and air-insulated switchgear (Figure 0-2).
26
The WG then reviewed existing software tools for calculating the effects of an internal arc fault, focusing on 3 main
27
effects of an internal arc:
Page 5
a. Pressure rise
b. Mechanical stress on enclosure and buildings
c.
Burn-through
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.
7
8
9
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
10
11
Pressure rise: After calculating the pressure rise in a number of cases with simple home made tools, the
12
WG realized that the calculated pressure peak was within 10-20% of the measured peak, which indicated
13
that usage of the simpler tool should be explored. Encouraged by that finding, the WG has developed a
14
set of basic equations and validated this mathematical model for all 70+ cases. It has been found that
15
calculation of pressure curves inside the arc compartment during an internal arc fault gives good
16
agreement between test and simulation as long as the input arc energy is known. These findings are
17
covered in Chapter 2, which lists the equations for the basic model and identifies its benefits and
18
limitations. Also, it shows different applications of the basic model for both MV and HV switchgear. Detailed
19
set of equations for the basic model is provided in ANNEX A: this can be used to create ones own home
20
made tool. In addition to the basic model, this Technical Brochure also covers the enhanced model and
Page 6
CFD in Chapter 2 and Chapter 4. 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.
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.
6
7
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.
8
9
Lastly, in effort to minimize Internal Arc tests, the working group created guideline for Internal Arc Simulation
10
review. Chapter 7 provides the guideline to replace the internal arc withstand test of the specific switchgear by
11
performing a design analysis based on tests of a similar design
12
13
14
15
Page 7
1 INTRODUCTION
1.1 Overview
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.
7
8
Figure 1-1: 13 kA electric arc moving between two conductors, at 5cm distance.
Two ongoing trends in the power industry are causing the possible damage from internal arc faults to increase.
10
One is the increase in the available fault current levels, resulting in an increase in the available arc energy. The
11
other is the evolution towards more compact switchgear, which results in smaller enclosures. These reduced tank
12
volumes result in a higher rate of pressure rise, higher temperatures and larger electro-magnetic forces on the
13
conductors. On the other hand, minimizing SF6 gas release to the environment is becoming a more and more
14
important issue and IEC 62271-200 actually permits SF6 to be replaced by air in an internal arc test. Pressure rise,
15
temperature rise and arcing behaviour, however, might be quite different between SF6 and air, because the
16
relevant physical properties of these gases, such as specific heat, density, etc. are significantly different. Therefore
17
any reasonable way to achieve an equivalent demonstration of the performance of the test object using air should
18
be explored.
19
The incidence of internal arc faults in MV and HV switchgear is very rare, but when an arc fault occurs in an
20
electrical installation it may seriously damage the electrical equipment and the switchgear buildings (see Figure
21
1-2), and endanger personnel. Failures typically occur due to:
22
external influences
23
material or mechanical defects
24
incorrect operation
Page 8
Figure 1-2: Internal arc fault test.
2
3
4
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 second is comparable to the energy
8
9
released by exploding 2 kg of dynamite.
10
Sound levels can reach 160 dB. By comparison, shotgun noise is measured to be in the range of 150 160
dB.
11
Debris may travel at speeds up to 1000 km/h.
12
Vapourized copper expands to 67,000 times its solid volume (1 cm of copper vaporises into 67 L of
13
14
vapor). For comparison: the conversion of water into steam has an expansion factor of 1670.
15
16
The resultant force of the expelled gases following rupture may reach several tones 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.
17
18
If a fault arc occurs in an electrical installation, the electrical energy of the arc plasma is transferred to its
19
surroundings by various different mechanisms. The main pressure rise is due to heat transfer. The energy input
20
into the fault arc by Joule heat is balanced by the interactions of the arc column with the electrodes, the arc length
21
and by several energy exchanges.
22
compartment where the arc occurs, and also through relief openings in the enclosure of the compartment. The
23
convective transfer of heat and the mass of the gas cause a change in the internal heat of the surrounding gas, and
24
is therefore part of the overall pressure rise. In addition, evaporated metal from the arc roots together with chemical
25
reactions play an important role in the energy transfer from the fault arc to the surrounding gas. As a result, during
26
internal arcing a rather great amount of energy is released in the cubicle and into the environment within a short
27
period of time.
This includes heat conduction, radiation and gas convection inside the
Page 9
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 between them. Gas
properties are assumed to be independent of temperature and pressure.
10
2. Enhanced models:
11
These models are based on the same basic equations, effective volumes and openings. Some of them
12
consider temperature and pressure dependent gas properties. They may be extended by including further
13
effects such as exothermic reactions, ablation of material and mixing of gases.
14
3. CFD models:
15
The calculation of gas pressure and temperature is based on the fluid-dynamic equations describing the
16
conservation of mass, momentum and energy of the gas in each finite volume element. The system of
17
equations is solved three-dimensionally with a computational fluid dynamics (CFD) solver.
18
The calculation approach in Chapter 2 describes the basic model in detail and provides a comparison with
19
measured data. Enhanced and CFD models are presented and discussed in the brochure (Section 2.6 and Chapter
20
4). Table 1-1 summarizes the application range and the limitations of the three approaches.
21
22
23
24
25
26
27
28
29
30
31
Page 10
1
Approach /model
Appropriate Application
Limitations
Doesnt consider spatial nonuniformity of gas parameters

(pressure, temperature, density) in
each volume part.
To quickly calculate uniform pressure

1)
Not applicable if the relief opening
Basic
rise inside an arc compartment and the
area is too large in relation to the
(low complexity)
exhaust volume in typical MV switchgear
compartment volume.
and HV GIS applications.
Calculations are not reliable, when

gas temperature exceeds approx.
2000 K for SF6 and 6000 K for air.
Doesnt consider gas mixtures in the

exhaust compartment.
2)
Enhanced
(medium
complexity)
Doesnt consider spatial non-
To calculate uniform pressure rise
uniformity of gas parameters
as under 1) adding further
(pressure, temperature, density) in
approximations to better match test
each volume part.
results and calculation.
Limitations and applications depend

on the implemented approximations.
3)
CFD
(High complexity)
High effort for the modeling and
For calculating spatial pressure
meshing of the rooms and
distribution and gas flow in odd shapes
switchgear
geometry and large rooms.
Requires large computing power

and time.
Table 1-1: Model selection table.
Page 11
1.2 Definitions and Abbreviations
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. Relatively small
openings necessary for interconnection, control or ventilation may be present.
10
Arc duct: Channel connected to the arc-exhaust intended to lead the arc products to another place.
11
Arc exhaust: The expulsion of hot gases from an arc fault through the relief opening of an arc compartment.
12
Arc fault: A high power discharge of electricity caused by a breakdown of insulation or flashover generating
13
excessive heat.
14
Arc fault single phase: Arc fault occurring between one conductor and ground.
15
Arc fault three phase: Arc fault occurring between three conductors or between three conductors and ground.
16
Arc flash: Direct physical phenomenon such as flame due to the hot plasma expansion of an arc fault. This can
17
cause burns and fire, and impacts humans as well as equipment and surroundings.
18
Arc plasma: Thermal plasma generated by an arc.
19
Arc power: Active (electrical) power of an arc given by the product of momentary current and voltage measured at
20
the terminals of the test object.
21
Arc voltage: Voltage which appears between the electrodes of an arc.
22
Arcing time: The time elapsing from the ignition of an arc to the interruption of the current.
23
Available (prospective) current: The current that would flow in a circuit if each pole of the switching device was
24
short-circuited by a link of negligible impedance without any other change in the circuit or the supply.
25
Basic model: Mathematical approach for the calculation of pressure rise due to an internal arc using simplified
26
equations under basic assumptions.
27
Burn-through: A hole burnt through the walls of the equipment enclosure or compartment by an arc.
28
Computational fluid dynamics (CFD): A branch of fluid mechanics that uses numerical methods and algorithms
29
to solve and analyze problems that involve fluid flows.
30
Deflection: The degree to which a structural element is displaced or bent under a mechanical load. It may refer to
31
an angle or a distance.
32
Deflectors: Plates placed in the flow of exhaust gas to deflect the stream of gas.
Page 12
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.
10
Exhaust compartment: Enclosed volume adjacent to the arc compartment which receives the arc exhaust.
11
Exothermic: In thermodynamics, the term exothermic ("outside heating") describes a process or reaction that
12
releases energy from the system.
13
Fault-shorting switch (arc killers): Eliminates arc faults by creating a metallic short circuit. It generally has a
14
sensor to detect the arc and an earthing device to extinguish it. The sensor can be either sensitive to the light
15
generated by the arc or to the pressure reached in gas sealed tank.
16
Finite element analysis (FEA): A numerical technique for finding approximate solutions of partial differential
17
equations (PDE) as well as integral equations.
18
Finite volume method (FVM): A method for representing and evaluating partial differential equations in the form of
19
algebraic equations. "Finite volume" refers to the small volume surrounding each node point of a mesh.
20
Heat capacity ratio: Ratio of the specific heat of a gas taken at constant pressure to that taken at constant volume
21
also termed adiabatic index.
22
Heat conduction: A mode of transfer of energy within and between bodies of matter, due to a temperature
23
gradient. Conduction means collisional and diffusive transfer of kinetic energy of particles of tangible matter (as
24
distinct from photons).
25
Heat convection: Heat transfer by the flow of a fluid in regions with different temperatures. Convective heat and
26
mass transfer take place through both diffusion (the random Brownian motion of individual particles in the fluid) and
27
by advection, in which matter or heat is transported by the larger-scale motion of currents in the fluid.
28
Heat flux: Heat flux or thermal flux, sometimes also referred to as heat flux density or heat flow rate intensity is a
29
flow of energy per unit of area per unit of time.
30
Heat flux sensor: A transducer that generates an electrical signal proportional to the total heat rate applied to the
31
surface of the sensor. The measured heat rate is divided by the surface area of the sensor to determine the heat
32
flux.
33
Heat radiation: Emission and propagation of energy in the form of electromagnetic waves.
34
Heat transfer: Transfer of heat e.g. from an arc to its surroundings. Heat transfer is classified into various
35
mechanisms, such as heat conduction, convection, thermal radiation, and transfer of energy by phase changes.
Page 13
Infrared thermography: Thermal imaging cameras detect radiation in the infrared range of the electromagnetic
spectrum (roughly 900014,000 nanometres or 914 m) 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
10
motion of the wave.
11
Metal-enclosed switchgear: A switchgear assembly completely enclosed by sheet metal (except for ventilation
12
openings and inspection windows) containing primary power circuit switching or interrupting devices, or both, with
13
buses and connections, which may also include control and auxiliary devices. Access to the interior of the
14
enclosure is provided by doors or removable covers.
15
Metal evaporation: Heating a metallic part up to a temperature, at which a considerable amount of metal vapour is
16
released from its surface.
17
Navier-Stokes equation: The NavierStokes equations describe the motion of fluid substances. These equations
18
arise from applying Newton's second law to fluid motion, together with the assumption that the fluid stress is the
19
sum of a diffusing viscous term (proportional to the gradient of velocity), plus a pressure term.
20
Net compartment volume: Effective volume of a compartment after subtraction of the volume of all built-in
21
components.
22
Overpressure: Pressure above ambient pressure within a pressurized enclosure.
23
Pad-mounted: A general term describing switchgear equipment positioned on a surface-mounted pad located
24
outdoors. The equipment is usually enclosed with all exposed surfaces at ground potential.
25
Pressure relief device: A device which opens on overpressure, releasing gases from a compartment into the
26
ambient atmosphere. A pressure relief device can be a loose flap or even a constant opening to the outside world.
27
Pressure withstand: Maximum pressure which can be withstood by an enclosure.
28
Relief opening area: Area provided by a pressure relief device to expel hot gases.
29
Response (bursting) pressure: Pressure at which a pressure relief device is ruptured or opens.
30
Rupture (bursting) disc: A non-reclosing pressure relief device that, in most uses, protects a pressure vessel,
31
equipment or system from overpressurisation. A rupture disc is a type of sacrificial part because it has a one-time-
32
use membrane that fails at a predetermined differential pressure, either positive or negative.
33
Short circuit: An abnormal connection (including an arc) of relatively low impedance, whether made accidentally or
34
intentionally, between two points of different potential.
Page 14
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
10
begins to deform plastically. Prior to the yield point the material will deform elastically and will return to its original
11
shape when the applied stress is removed. Once the yield point is passed, some fraction of the deformation will be
12
permanent and non-reversible.
13
Youngs modulus: It is defined as the ratio of the uniaxial stress over the uniaxial strain in the range of stress in
14
which Hooke's Law holds. It can be experimentally determined from the slope of a stress-strain curve created
15
during tensile tests conducted on a sample of the material.
16
1.3 Referred standards
17
EEMAC G14-1, Procedure For Testing The Resistance Of Metal Clad Switchgear Under Conditions Of Arcing Due
18
To An Internal Fault, EEMAC G14-1, 1987.
19
EN 61482-1-2, IEC Standard 61482-1-2. Live working Protective clothing against the thermal hazards of an
20
electric arc. Part 1-2: Test methods - Method 2: Determination of arc protection class of material and clothing by
21
using a constrained and directed arc (box test).
22
IEC 60076-5, IEC Standard 60076-5. Third edition 2006. Annex A: Theoretical evaluation of the ability to withstand
23
the dynamic effects of short circuit.
24
IEC 60298, IEC 60298 A.C. metal-enclosed switchgear and controlgear for rated voltages above 1 kV and up to
25
and including 52 kV.
26
IEC 62271-200, High-voltage switchgear and controlgear Part 200: AC metal-enclosed switchgear and
27
controlgear for rated voltages above 1 kV and up to and including 52 kV. Ed.2.0 , 2011
28
IEC 62271-201, High-voltage switchgear and controlgear - Part 201: AC insulation-enclosed switchgear and
29
controlgear for rated voltages above 1 kV and up to and including 52 kV.
30
IEC 62271-203, High-voltage switchgear and controlgear Part 203: Gas-insulated metal-enclosed switchgear for
31
rated voltages above 52 kV.
32
IEEE 1584-2002, IEEE 1584 Standard. IEEE Guide for Performing Arc-Flash Hazard Calculations.
33
IEEE Standard C37.20.7-2007, IEEE Guide for Testing Medium Voltage Metal-Enclosed Switchgear for Internal
34
Arcing Faults.
Page 15
NFPA 70E-2012, Standard for Electrical Safety in the Workplace, NFPA, 1 Batterymarch Park, Quincy, MA 02169-
7471.
3
4
Page 16
2 CALCULATION OF PRESSURE USING A BASIC MODEL
2.1 Introduction
The first part of this chapter (Sections 2.2-2.4) focuses on describing of 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
10
2.2.1 Basic arrangement and quantities
11
Figure 2-1 shows schematically an installation consisting of arc compartment, exhaust compartment, and
12
installation room/environment. The arc represented by the temporal development of energy input 1 is ignited in the
13
14
15
16
arc compartment with volume 1 . A pressure relief opening with cross-section 12 connects the arc to the exhaust
compartment with volume 2 . When the pressure 1 in the arc compartment reaches the response pressure, the
relief device opens and gas flows into the exhaust compartment 2 . From there, gas flows through the opening with
cross-section A23 into the installation room or environment with volume 3 .
Exhaust Compartment
Arc Compartment
V1
V2
A12
p1
m
12
T1
Installation Room
V3
p2
A23
m 23
T2
p3
T3
Q1
17
Figure 2-1: Principal arrangement and quantities used for pressure calculation.
18
19
The type of insulating gas in each volume is characterized by the corresponding heat capacity ratio (adiabatic
20
index) and the specific gas constant . In the basic model, these quantities are assumed to be constant. The initial
21
22
23
state of the gas is defined by pressure and temperature . While volume 1 may be filled with air or SF6, volumes
2 and 3 are typically always filled with air.
Mass, density and specific heat constants of the gas in volumes 1 and 2 are (see ANNEX A.2.1).
Page 17
1 =
1 1
1 1
2 =
2 2
2 2
1 =
1
1 1
2 =
2
2 1
1 =
2
3
4
1
1
2 =
1 = 1 1
2
2
2 = 2 2
The thermal energy 1 as a part of the electrical energy heats up the gas.
1 =
(2-1)
(2-2)
(2-3)
(2-4)
(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 equations for difference of thermal
energy (1 = ) are provided 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 . The mass flow from
8
9
10
11
the arc compartment into the exhaust compartment is given by:
12 = 12 12 12 12
12 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 (12 12 is the effective opening).
Reaching the response pressure in 1 , the pressure relief device opens. 12 and 12 are the gas density and gas
velocity within the opening 12 , which are different from the values in 1 and 2 [Schmidt1958].
12
12
12
13
(2-6)
12 1
= 1
1
(2-7)
1 1
1
21 1
12
=
1
1 1 1
1
(2-8)
If the ratio of pressure in 1 and 2 i.e. 1 /2 exceeds a value of 1.89 for air and 1.70 for SF6, respectively, 12 is
determined by the critical pressure 1 ; for smaller ratios 12 is equal to the pressure in 2 .
12 = (2 , 1 )
Page 18
(2-9)
1
1 1
2
1 = 1
1 + 1
(2-10)
The mass in volume 1 is reduced after the time step by the mass 12
1 = 12
(2-11)
23 = 23 23 23 23
(2-12)
The mass from the exhaust compartment flowing into the installation room (3 ) is given by:
Density and flow velocity are calculated using the equations provided above with all indices incremented by one.
The change of mass in volume 2 within is the difference between the incoming mass m12 and the outgoing
mass m23 during .
2 = 12 23
(2-13)
2.2.3 Gas temperature
The temperature change in the arc compartment with volume 1 after the time step is determined by the
8
9
10
11
12
13
14
difference between the thermal energy input by the arc (1 ) and the energy loss due to gas flow out of the
compartment (see ANNEX A.2.4 for details):
1 =
1 12 1 1 1
1 1
The corresponding temperature change in 2 is:
2 =
12 1 1 2 2 23 2 2 2
2 2
(2-14)
(2-15)
The summation of all temperature changes provides the temperature at time .
2.2.4 Gas pressure
With given gas mass and temperature, the pressure in 1 and 2 at time is given by the ideal gas law:
1 =
(1 1)
1 1 1
1
2 =
(2 1)
2 2 2
2
Temperature 3 and pressure 3 in the installation room are calculated correspondingly.
Page 19
(2-16)
2.3 Input parameters
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
5
6
and are given in Table 2-1 taken from publications. The value of is calculated directly from and using
equations (2-3) and (2-4). The assumption of being independent of temperature is acceptable up to
temperatures where gases start to dissociate (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
9
10
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
[Mende1975]
SF6
[Solvay]
716
608
J kg /K
1005
665
J kg /K
kg/m
specific heat capacity at constant volume
gas density
1.205
6.07
heat capacity ratio (kappa)
1.403
1.0936
29
146
kg/kmole
universal (molar) gas constant
8314
8314
J K kmole
287
56.9
J K kg
specific heat capacity at constant pressure

molar mass
11
-1
specific gas constant
-1
-1
-1
-1
-1
Table 2-1: Basic gas quantities at normal conditions (20 C and 101.3 kPa).
12
For SF6 insulated switchgear, the flow of SF6 out of the arc compartment leads to an SF6/air mixture in the exhaust
13
compartment. Gas mixing is not considered in this model, i.e. the gas properties in the exhaust compartment are
14
taken to be those of pure air. However the mass and energy exchange is considered correctly. This simplification is
15
reasonable as long as the SF6 concentration is low (e.g. in large exhaust compartments). For high concentrations,
16
gas data from mixtures have to be considered (see Section A.4.4).
17
2.3.2 Volume
18
All volumes in the model are net volumes i.e. volume of compartment minus volume of built-in components. The
19
shape of the compartment is not considered. The volume of the built-in components might reach 10 to 20 % of that
20
of the arc compartment. The energy supplied by the arc is taken as homogeneously distributed inside the arc
21
compartment. That is why the model does not cover pressure waves, which might play a role in long, narrow
22
compartments (channels).
Page 20
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.
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.
10
The response pressure of the relief device is most often given as a static value. If the value is determined from
11
arcing tests, it is a dynamic value. Experience shows that the difference between both is of the same order as the
12
scatter from sample to sample.
13
2.3.4 Arc current
14
For pressure calculation the temporal development of the single or three-phase fault current must be known. This
15
current development can be taken from test or from simulation. The asymmetry of the short circuit current might
16
influence the initial pressure build-up, however, has little influence on the later pressure curve. As long as the
17
supply voltage is much higher than the arc voltage, the arc resistance does not influence the current asymmetry
18
(see Section 5.2.4). Knowing the d.c. time constant of the circuit .. , of the source circuit, the temporal
19
20
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, .
21
2.3.5 Arc voltage
22
The arc voltage is one of the most important parameters determining arc energy and hense the pressure rise. In
23
tests the arc energy is determined from measured line currents and phase-to-ground voltages:
= ( + + )
(2-18)
24
The term in brackets is the arc power. The arc voltage is the voltage drop of an arc between its roots. For pressure
25
calculations (and the comparison with measured values) averaged arc voltages are used. In a three-phase
26
system, the phase-to-ground voltage may not be identical with the arc voltage. If the arc appears between the
27
phases only two arcs appear simultaneously commutating between the phases. In this case the arc energy during
28
is given by [Welich1984]:
= 0.5 (| | + | | + | |)
Page 21
(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 is given by:
= (| | + | | + | |)
(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
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 inserted. 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
10
material. The arc voltage differs between an arc ignited in an empty or in a fully equipped compartment. Some
11
experiments show that the arc voltage increases with rising pressure in the compartment and decreases later after
12
pressure reduction [Dullni1994].
13
If arc voltage data is missing, basic formulas, which have been extracted from three phase internal arc tests with
14
MV metal enclosed switchgear separately for air and SF6 and copper electrodes might be used [AiF2011]. These
15
voltages have to be applied together with the energy equation (2-19) for arcs between phases.
V
1
V
V
= 30
+
40
cm 2
cm kA
cm
16
V
1
V
V
= 40
+
50
cm 2
cm kA
cm
(air)
(SF6 )
(2-21)
(2-22)
Here is the arc voltage between phases, is the distance between pole centres, and is the effective short
17
circuit current.
18
For single phase HV aluminium enclosures filled with 1 to 4 bar of SF6 the following formula has been derived from
19
a survey of available data for certain specific conditions (details see [Knig1984]).
20
= 250 +
50 + 4
mm
kA
(2-23)
Here 0 is the arc voltage including the -factor, is the clearance between conductor and wall, and is the
21
effective short-circuit current.
22
Internal arc faults in electrical switchgear in 3-phase arrangements typically start as a phase to ground or as a
23
phase to phase fault. If this arc fault cannot be eliminated automatically in a short time (in the millisecond range),
24
and no single pole solid insulation is present, it will most likely develop into a three-phase arc fault. A three phase
25
arc fault consists of two parallel arcs with arc roots changing among the three phases. In the case of close
26
proximity of the enclosure, the fault may rapidly develop into three parallel phase-to-ground arcs.
27
Depending on the network voltage, the time constant of the asymmetrical arcing current will decrease to about half
28
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
Page 22
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.
3
4
Figure 2-2: Momentary arcing powers. Asymmetrical ( = 45 ms) first 5 periods (100 ms) and symmetrical
one period.
2.3.6 k p - factor
The heat transfer coefficient determines that 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
calculated pressure rise to the measured one before the operation of the pressure relief device. If exothermic
10
reactions occur, e.g. between aluminium and SF6, additional energy might heat up the gas resulting in a -factor
11
12
e.g. [Zhang2002] exist, however, this approach is not really feasible. In practice, is determined by fitting the
possibly larger than 1.
It is highly recommended that -factors are determined from tests under similar conditions. If no experiments are
14
available, -factors taken from publications should be used with caution. It has been found that in general in air
15
conditions with copper electrodes ranges from 0.40 to 0.65 and in case of SF6 from 0.50 to 0.70. For aluminium
13
16
17
is lower than in SF6 [Dullni1994, Friberg1995]. Based on these experimental investigations, for air at standard
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
18
compartments with relief openings, where gas density is considerably reduced during the exhaust of gas (see
19
Section A.4.1). In general, decreases with gas density [Dullni1994, Zhang2002].
20
2.4 Application limits of the basic model
21
Due to the assumptions implemented in the basic model, the user should be aware of how the application range is
22
limited. Typical applications with reliable results are:
Page 23
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
4
5
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).
10
The model does not consider the evaporation of metal or insulation material, which influences gas
11
composition, density and temperature. This influence is negligible as long as the density of the insulating
12
gas (air or SF6) is larger than the density of the vapour. Due to the exhaust of gas from the arc
13
compartment, the density of the insulating gas drops within this compartment, and the proportion of gas
14
resulting from evaporation becomes more and more important. For long arcing times the gas in the arc
15
compartment may practically consist only of evaporated material. Because evaporation is not considered in
16
the model, the simulation results become uncertain when the insulating gas density in the arc compartment
17
drops significantly.
18
If considerable gas flow occurs in any compartment (e.g. in elongated rooms or channels) the approach
19
with spatially averaged quantities is not applicable. As a consequence, reliable results are only achieved
20
for pressure relief opening areas limited to not more than 10 % of the side surface of the arc compartment
21
(see Section 2.3.3).
22
The energy transfer out of the arc compartment determines the pressure rise in the exhaust compartment.
23
The model assumes a constant gas type in the exhaust compartment. Typically the exhaust compartment
24
is filled with air. In case of SF6-insulated switchgear the gas in the exhaust compartment will be a mixture
25
of SF6 and air. Therefore the assumption of only air in the exhaust compartment is violated when the SF6
26
portion becomes remarkable. This might be the case for small exhaust volumes. These considerations do
27
not affect air-insulated switchgear.
28
Some of these limitations can be overcome to some extent by additional approaches (e.g. enhanced models, see
29
Section 2.6). These include for example the application of real gas data, the evaporation of metal and insulation
30
31
material, gas mixtures, the density dependence of the -factor and exothermic reactions. Such modifications do
32
calculation for longer arcing times and a calculation of the pressure rise in the exhaust compartment or installation
33
room. Spatial resolution, if necessary, can be achieved using CFD tools. Anyhow these are in principle subject to
34
the same limitations as discussed above.
not really increase the accuracy of the pressure calculation in the arc compartment, but allow the extension of the
Page 24
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 SF6
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
Section 2.5.2. These cases can be used by readers to benchmark their own calculation programs.
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;
10
discharge coefficient is chosen between 0.7 to 1 (adapted to the measured pressure decay);
11
energy input is based on the measured currents and averaged phase-to-ground voltages using equation
12
(2-18) (see Section 2.3.5);
13
14
-factor results from adapting the calculated to the measured pressure rise (slope
response pressure of the relief device);
/ up to the
15
response pressure pburst of the relief device is taken from measured pressure curves;
16
the limitations described in Section 2.4 are respected.
17
The input parameters and initial values of the selected cases are listed in Table 2-3, Table 2-4 and Table 2-5
18
respectively. Some details of the test arrangements or switchgear configurations are given in Section 2.5.2.
19
Calculated and measured pressure curves are shown in Figure 2-10 to Figure 2-17. The red curves show the
20
calculated pressure in the arc compartment, the blue curves the pressure in the exhaust compartment, if available.
21
Grey curves present measured data. The simulation stops when the gas temperature in the arc compartment
22
reaches 20000 K. The changes in the gas pressure in the arc and exhaust compartments are characterised by
23
values determined from measured pressure curves. These values are depicted in Figure 2-3 and listed in Table
24
2-2.
Peak pressure
Maximum pressure measured after opening of the relief device
Response pressure
Pressure value, at which the relief device opens
Time to peak
Time between start of pressure rise and maximum pressure
Response time
Time, at which the relief device opens
Pressure slope
Initial rise of pressure until
Decay time
25
26
27
Duration of pressure drop starting from peak pressure
Table 2-2: Characteristic values for the pressure curve.
The change of gas pressure in the exhaust volume can be characterised 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.
Page 25
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.
3
4
5
Figure 2-3: Characteristic values determined from calculated or measured pressure curve.
2.5.2 Test arrangements
Side view of container

6
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
Figure 2-4: Test cubicle used for cases A and E.
Page 26
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.

1
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
Front view of bus bar compartment

2
Exhaust into Channel 2
Figure 2-6: Switchgear used for cases D and G.
Page 27
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)

1
Figure 2-7: Switchgear used for case C.
Two welded steel housings
Arc ignited in volume 1
Top view arc and exhaust volumes

3
Exhaust into volume 2
Figure 2-8: Experimental arrangement used for case B.
Page 28
Cross-section of encapsulation
1
2
Cast aluminum housings
Single phase circular Al bar
Coaxial electrode arrangement
Arc ignited at spacer
Figure 2-9: Arrangement of single-phase HV components used for case H.
2.5.3 MV switchgear with air insulation

Case No.
Volume of arc comp. (1 )
0.509
0.509
0.648
0.27
Volume of exhaust comp. (2 )
>1000
1.275
>1000
0.58
Volume of installation room (3 )
n/a
>1000
n/a
>1000
Initial filling pressure in 1
150
160
100
120
kPa abs air
100
100
100
100
kPa abs air
Area of the relief opening 12
0.00456
0.00456
0.0763
0.049
Discharge coefficient of 12
0.7
1.0
0.7
1.0
Response pressure of relief device
276
285
35,3
220
kPa rel
Area of the opening 23
0.010
0.195
Short-circuit current
14.5
14.5
14.5
38.8
kA rms
Number of phases
Averaged phase-to-ground voltage
314
424
400
250
-factor
0.4
0.55
0.7
0.6
Table 2-3: Input parameters and initial values for MV switchgear cases with air insulation.
4
5
6
Page 29
Case A 22
0.5
0.45
P1 calculated
0.4
P1 measured
pressure [MPa]
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
time [s]
1
2
3
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
pressure [MPa]
0.35
P2 measured
0.3
0.25
0.2
0.15
0.1
0.05
0
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
time [s]
4
5
Figure 2-11: Case B Calculated pressure developments in V1 and V2 in air and comparison with test.
Page 30
1
Case C 70
0.15
P1 calculated
0.14
P1 measured
pressure [MPa]
0.13
0.12
0.11
0.1
0.09
0.08
0
0.05
0.1
0.15
2
3
4
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
P1 calculated
0.35
P2 calculated
0.3
P1 measured
pressure [MPa]
P2 measured
0.25
0.2
0.15
0.1
0.05
0
0
5
6
7
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.
8
9
Page 31
For air as the filling gas in the compartment, calculation with the basic model and measured results show good
agreement. 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. 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
10
one volume in order to achieve the best agreement. This measure is recommended if there are small (intermediate)
11
volumes with large openings between volumes.
12
Overall, the agreement between experiment and calculation is better than 10% considering that the arc voltage has
13
14
been provided from the tests and the -factor as well as the discharge factor have been adapted to give the best
15
cases.
16
2.5.4 MV switchgear filled with SF 6 gas
results. This result is supported by the range of deviations of the peak pressure shown in Section 2.5.6 for all test
Case No.
0.509
1.217
0.27
>1000
>1000
0.58
NA
NA
>1000
150
166
120
kPa abs SF6
100
100
100
kPa abs air
0.00456
0.062
0.049
1.0
1.0
1.0
310
1400
220
NA
NA
0.195
14.2
25
38
Number of phases
350
1700
400
-factor
0.75
0.7
0.76


17
kPa rel
m
kA rms
Table 2-4: Input parameters and initial values for MV switchgear cases with SF6 insulation.
18
Page 32
Case E 24
0.5
0.45
P1 calculated
0.4
P1 measured
pressure [MPa]
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
0.2
0.4
0.6
0.8
1
2
1.2
1.4
1.6
1.8
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
pressure [MPa]
1.2
1
0.8
0.6
0.4
0.2
0
0
3
4
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 33
Case G 13
0.4
P1 calculated
0.35
P2 calculated
pressure [MPa]
0.3
P1 measured
P2 measured
0.25
0.2
0.15
0.1
0.05
0
0
0.05
0.1
0.15
1
2
3
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.
4
5
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
7
8
peak pressure, but also a much longer residence time of the gas in the arc compartment. The -factor is taken 0.7
pressure curves was obtained with discharge coefficients of the exhaust openings of 1.0, whereas for air as filling
10
gas the anticipated factor of 0.7 was prevailing. For SF6, even better agreement would be obtained for discharge
11
coefficients larger than 1. In particular in case G, the calculated residence time of SF6 in the arc compartment is
12
much longer than measured. Better agreement with test results would be achieved by increasing the gas outflow by
13
a factor of 2, which is equivalent to a discharge coefficient of 2. Since for air (see case D), the outflow is simulated
14
correctly, this cannot be attributed to a falsely determined outflow area, but rather seems to be a peculiarity of SF6.
15
Since the gas temperature in this case exceeds 2000 K, SF6 is already dissociated to a large amount suggesting
16
that decomposition products are responsible for the higher flow velocity. For air, this effect might not be visible,
17
since decomposition starts at higher temperatures. Other effects like the observed reduction of the phase-to-
18
ground voltage after opening of the relief device or the reduction of the -factor as discussed in Section A.4.1
to 0.76 consistently for all cases, which is within the range of published data. The best fit of the decaying part of the
19
might also result in a shorter residence time, however, should be effective for both air and SF6.
20
Summarizing, the fit between experiment and calculation for SF6 is not as good as for air in particular with respect
21
to the decay of pressure after opening of the relief device.
22
2.5.5 HV switchgear filled with SF 6 gas

Case No.
Page 34

>1000
NA
490
kPa abs SF6
100
kPa abs air
0.006
1000
1.0
0.39
kPa rel
NA
65
kA rms
Number of phases
450
-factor
1.6
Table 2-5: Input parameters and initial values for HV switchgear cases with SF6 insulation.
2
Case 01
H
P1 calculated
2.5
P1 measured
pressure [MPa]
1.5
0.5
0
0
3
4
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
6
7
achieved adapting to a high value of 1.6. This is most probably due to excessive evaporation of electrode and
energy available for gas heating. This case is better reproduced by an enhanced model, which takes electrode
evaporation into account. The discharge coefficient of the exhaust opening is again taken as 1.0 and the exhaust of
10
housing material in particular aluminium. The associated reaction is exothermic and increases the amount of
gas is simulated correctly though the gas temperature also exceeds 2000 K.
Page 35
2.5.6 Evaluation of key parameters for all test cases
2
3
For all test cases, the peak pressure in volume 1 was calculated and compared with the result from tests.
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
ageement 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
The values are shown in Figure 2-18 irrespective of any differences of the tests with respect to geometry, electrical
an adaptation of the arc energy input can only be done after the tests.
9
10
11
Figure 2-18: Deviation between measured and calculated peak pressures in V1 irrespective of geometrical and
electrical parameters and kind of filling gas.
12
13
2.6 Enhancements of the basic model
14
Enhanced models are based on the same gas equations as are used for the basic model. However, some
15
simplifications introduced in the basic model can be removed and some parameters may obtain dependencies
16
deduced from theory or experiments. The modifications allow a better adaption of measured data to the calculated
17
data, but also however introduce a higher degree of freedom with the uncertainty that the included effect might be
18
more or less relevant in the general case. These modifications do not increase the accuracy in predicting the peak
19
pressure in the arc compartment, but do increase the accuracy in simulating the time dependence of the gas
20
outflow and pressure development in the exhaust compartment. Details of possible modifications are provided in
21
ANNEX A.4.
22
2.7 Summary
23
The attempt to re-calculate pressure curves measured inside the arc compartment during an internal arc fault gives
24
good agreement between test and simulation as long as the input arc energy is known. Pressure peaks and decay
Page 36
can be simulated within a deviation of 10 %. The comparison also indicates that for most arrangements common
2
3
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
of predictive calculations is larger than 10%. Enhanced models with adapted input parameters, temperature-
dependent gas properties or separate gas equations for different gas species also including evaporation of
electrode material are appropriate to improve the agreement between simulation and test results.
0.7 and an approximated arc voltage determined from phase distance. Clearly one has to expect that the deviation
7
8
REFERENCES:
[AiF2011]: Final Report of a German research project, AiF-Projekt 15657N, Reduzierung der Druckbeanspruchung
10
elektrischer Anlagen im Strlichtbogenfall, Projektleiter: G. Pietsch Reduction of the pressure stress in electrical
11
installations due to internal arcs, 2011.
12
[Anantavanich2008]: K. Anantavanich, G. J. Pietsch, Inclusion of SF6-air mixtures in the calculation of pressure
13
rise due to internal arcing in switchgear installations, Proc. XVIIth Int. Conf. on Gas Discharges and their Appl.,
14
Cardiff, UK, pp. 165-168, 2008.
15
[Dubbel1997]: Editors W.Beitz, K.H. Grote, Taschenbuch fr den Maschinenbau, Springer Verlag, 19th edition,
16
1997, section B (Mechanik) 6.2.
17
[Dullni1994]: E. Dullni, M. Schumacher, G. Pietsch, Pressure rise in a switchroom due to internal arc in a
18
switchboard, Proc. 6th Int. Symp. on Short-Circuit Currents in Power Systems, pp. 4.5.1 - 4.5.7, Lige, Belgium,
19
1994.
20
[Friberg1995]: G. Friberg, G. Pietsch u. M. Schumacher, On the description of pressure rise in the surroundings of
21
high current arcs in metal enclosed compartments with pressure relief, Proc. 11th Int. Conf. on Gas Discharges
22
and their Applications, Tokyo, Japan, Vol. I pp. 18-21, 1995.
23
[Knig1984]: D. Knig, Th. Facklam, Pressure rise in metal enclosed SF6-insulated HV switchgear of single-phase
24
enclosure type due to internal arc, Electra no. 93, pp. 25-52, 1984.
25
[Mende1975]: D. Mende, G. Simon, Physik Gleichungen und Tabellen, VEB Fachbuchverlag Leipzig, 1975.
26
[Schmidt1958]: E. Schmidt, Technische Thermodynamik, 7th edition in German, published by Springer, 1958
27
Technical Thermodynamics.
28
[Solvay]: Schwefelhexafluorid, Company Brochure SOLVAY GmbH.
29
[Welich1984]: D. Welich, Zur Ausbildung mehrpoliger Strlichtbgen und zur Innendruckentwicklung von
30
luftisolierten, metallgekapselten Mittelspannungsschaltzellen, PhD thesis in German, Technical University
31
Dresden, 1984
32
On the formation of multi-phase internal arcs and pressure rise in air-insulated, metal enclosed MV switchgear.
33
[Zhang2002]: X. Zhang, G. Pietsch, E. Gockenbach, Investigation of the thermal transfer coefficient by the energy
34
balance of fault arcs in electrical installations, IEEE trans. On power delivery, 2006.
35
Page 37
3 SENSITIVITY ANALYSIS OF THE BASIC MODEL
3.1 Introduction
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 continuous parameters
are varied around the 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
10
selected parameters and the effects of replacing SF6 with air are described in Section 3.4.
11
The sensitivity analysis is limited to the basic model (Section 2.2).
12
3.2 Model parameters
13
The input parameters are defined in Chapter 2. This chapter uses the same definitions and notations, with a few
14
exceptions which are explicitly pointed out in the text. The subscripts 1, 2 and 3 below refer to arc compartment,
15
exhaust compartment and installation room respectively.
16
For the sensitivity analysis the parameters are regrouped in the following way:
17
energy input: effective short circuit current - , angle of fault initiation ; d.c. time constant ; mean
19
arc voltage - ; fraction of arc energy producing the pressure rise - kp,SF6 and kp,Air; number of phases -
20
Chapter 2
18
21
(1or 3); duration of short circuit current - ; the two last symbols are not introduced explicitly in
exhaust compartment relief opening areas - 12 and 23 , discharge coefficient for arc and exhaust
22
compartment relief openings - 12 and 23 ; arc and exhaust compartment response relative pressures -
23
,1 and ,2 ; arc and exhaust compartment initial pressures and initial temperatures - ,1 , ,1 ,
24
,2 , ,2 ; the symbols for initial values are not explicitly introduced in Chapter 2, in ANNEX A: they are
25
26
27
switchgear design and initial conditions: arc and exhaust compartment volumes - 1 and 2 ; arc and
noted as ( = 0)
gas type: heat capacity ratios - 6 and ; specific gas constants - ,6 and ,
28
The subscripts related to gas type (SF6 and Air) may be avoided in some equations.
29
3.2.1 Energy input
30
The energy () is integrated according to formula (2-20) in Section 2.3.5. For symmetrical current (=90) and
31
using (2-5), the part of energy absorbed by the gas () within a number of entire periods of power frequency can be
32
expressed by:
Page 38
1
2
( ) =
22
(3-1)
where = , is the number of cycles, is the frequency (50 or 60 Hz). Thus the parameters , and
22
have exactly the same effect on the provided energy. Let =
. We can consider this
expression as the average input power (heat source). The dimension of is watts.
influence on the pressure curve (Section 3.4.1).
3.2.2 Switchgear design and initial conditions
7
8
The parameters , , , ,1 , ,2 , ,1 , ,2 , and ,1 , ,2 define the main mechanical characteristics of
The absolute (measured) response pressure of the arc compartment relief device is the sum of the relative bursting
10
11
pressure and the absolute initial pressure in the exhaust compartment: = ,1 + ,2 . In the simulation,
12
relief device opening.
13
14
The initial gas density in the arc compartment equals = ,1 / ,1 . It does not depend on the volume.
15
temperature are the same.
16
3.2.3 Gas type
17
There are only two independent parameters in the model strictly related to the gas type: the heat capacity ratio
18
and the specific gas constant . The two remaining parameters related to gas type i.e. and are also related to
Equation (3-1) does not apply for asymmetrical current. However, the asymmetry parameters and have limited
the switchgear and the conditions at the beginning of the simulation.
the time when the absolute pressure in the arc compartment reaches this value is considered as the instant of the
However it does depend on gas type: it is more than five times higher for SF6 than for air if the pressure and the
20
the switchgear. The factor is discussed in Section 3.2.1 as part of the energy input equation. The sensitivity
21
(equation (A-22), ANNEX A:).
19
22
23
24
25
study is not relevant for the coefficient because its sole effect is to reduce the value of the opening area
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.
26
3.3 Model results
27
The most important output of the model is the pressure curve in the arc compartment and the exhaust
28
compartment during the process. The simulation finishes in two ways: it either stops normally when the simulation
29
time elapses or else it is aborted when the gas temperature reaches 20000 K.
30
A typical arc compartment pressure curve has the following intervals (Figure 2-3):
Page 39
Rapid pressure rise ( ) until the opening of the pressure relief device ( ).
Further pressure increase until the peak pressure is reached (,1 , ,1 ), called the overshoot period.
there is no overshoot. The difference between maximum pressure and absolute response pressure is
called the overshoot level (see Section 3.3.2).
For some conditions, it is possible that the pressure decreases immediately after the opening; in this case
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 (,2 , ,2 ).
Essentially, the model resolves the following four types of energy and flow equations:
10
11
before the bursting of the arc compartment relief device: arc energy is provided to the arc compartment, no
mass flows
12
13
after the bursting of the relief device: energy is still provided and a mass flows from the arc compartment to
the exhaust compartment
14
after the bursting of the exhaust pressure relief device: mass flows between all compartments
15
16
after the end of arc ( ) there is no more energy provided to the arc compartment but mass flow
continues
17
3.3.1 Before bursting of the pressure relief device
18
Before the bursting of the relief device and assuming constant gas parameters as well as constant energy input
19
rate () , the pressure curve in the arc compartment is a linear function of time:
() =
1
1
,1 + () = ,1 +

21
where is the gas mass in 1 and is the heat capacity at constant volume (ANNEX A:).
22
the beginning of pressure rise is due to the current asymmetry neglected by equation (3-2).
20
(3-2)
Figure 3-1 compares the pressure curve calculated with equation (3-2) to the basic model. The slight difference at
Page 40
0.49
0.44
Burst
3
0.39
[MPa]
2.8
0.34
2.6
2.4
0.29
2.2
0.24
2
1.8
0.19
1.6
0.14
0
0.2
0.4
0.6
1.4
0.8
0.07
1
[s]
0.12
1.2
0.17
1.4
1.6
1.8
0.22
1
2
3
4
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).
The approximated values of and can be calculated with the following equations:
( 1)
1
1 ,1
( 1)
(3-3)
(3-4)
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
10
constant. Therefore equation (3-2) is a helpful approximation of the pressure rise for slow processes.
11
3.3.2 After pressure relief bursing with a large or infinite exhaust compartment
12
If the exhaust compartment is infinite compared to the arc compartment, it will not affect the processes in the arc
13
compartment. In this case for slow processes, the pressure curve in the arc compartment after the burst can have
14
one of two forms: with overshoot or without overshoot. Without the overshoot, the pressure in the arc compartment
15
16
drops immediately after the opening of the relief device and ,1 = .
17
parameters:
It is shown in ANNEX F.3 that for slow processes the overshoot condition can be directly derived from the input
18
Page 41
3
> ,1
12 12
,1
(3-5)
where = 40.42 and 6 = 55.27.The constants depend only on and . 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 (,1 = 293 K).
The influence of on the overshoot condition of the pressure curve is shown in Figure 3-2.
0.5
0.45
289V, r=1
157V, r=0.543
0.4
251V, r=0.868
392V, r=1.356
487V, r=1.685
[MPa]
0.35
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]
7
8
9
10
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, > 1), intermediate (dark-green,
= ) and without overshoot (orange & red, < 1).
11
3.3.3 After bursting of the relief device with a limited size exhaust compartment
12
If the pressure in the exhaust compartment is greater than approximately half of the pressure in the arc
13
compartment, the flow of gas out of the arc compartment will be slowed down and the pressure curve will be
14
altered. More exactly, in Section 2.2.2 a gas dependent critical ratio is deduced from equation (2-10), relating the
15
16
gas pressure in the exhaust compartment (2 ) to that in the arc compartment (1 ) for which the latter is not
17
2

1
influenced by the former. The reciprocal value of the critical ratio is denoted here by .
18
where the constant depends only on : 6 = 0.586 and = 0.528.
19
results
(3-6)
At the instant of the pressure relief device bursting 1 = ,2 + ,1 and 2 = ,2 , so the following equation
Page 42
1
,1
1 ,2
(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).
0.5
simulation for air
0.45
0.42 MPa
0.4
0.35
[MPa]
0.3
0.25
0.22 MPa
0.2
arc compartement with infinit exhaust volume

arc compartement with exhaust volume of 1.275m
exhaust compartement
critical ratio of 0.528
0.15
0.1
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]
5
6
7
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:
10
A.
11
fraction of the initial pressure in the exhaust compartment:

,1 > 0.706 ,2 for SF6 and
12
13
14
the relative bursting pressure of the relief device of the arc compartment is greater than the following
B.
,1 > 0.893 ,2 for air
the pressure in the arc compartment is sufficiently high to ensure that the condition (3-6) holds
15
The second condition should be verified by simulation.
16
17
Figure 3-4 shows the case where the condition A above holds (,1 = 0.27 MPa vs ,2 = 0.1 MPa) but the
18
pressure in the arc compartment (pink line) before the maximum is reached.
condition B does not. The pressure in the exhaust compartment (green line) increases relatively fast and affects the
Page 43
0.45
simulation for SF6

0.41 MPa
0.4
0.35
[MPa]
0.3
0.25
0.24 MPa
0.2
arc compartement with infinit exhaust volume

arc compartement with exhaust volume of 1.275m
exhaust compartement
critical ratio of 0.586
0.15
0.1
0.05
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
[s]
1
2
3
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
7
8
limited to the evaluation of arc compartment results: ( , ) and ( , ).
expressed by equations (3-3) and (3-4).
The sensitivity is expressed by the variation related to the reference value. For ( , ) the sensitivity is also
10
3.4.1 Sensitivity to power input
11
The power provided to the system has a great influence on the pressure curve for both the arc and the exhaust
12
compartments. From equation (3-1), the parameters , , are equivalent i.e. varying any of them has the
13
same effect on the pressure curve, e.g. Figure F-1, Figure F-2, and Figure F-3 in ANNEX F:.
14
General observations
15
Increasing power:
16
is proportional to the power input and the time to burst is inversely proportional to the power
17
18
19
accelerates the process before bursting of the pressure relief device; for slow processes the pressure slope
input
increases the maximum pressure in arc compartment if overshoot is present. The peak pressure growth is
20
generally higher for air than for SF6. In the absence of an overshoot condition the peak pressure remains
21
the same and equals the bursting pressure (Figure F-4 in ANNEX F:).
22
increases the maximum pressure in the exhaust compartment
23
favors the overshoot condition (criterion (3-5))
24
Page 44
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
of pressure rise for SF6 switchgear, whereas for air current duration is hardly relevant, since all effects happen
instantaneous pressure at this moment (Figure F-7 in ANNEX F:). Shorter current durations may relieve the impact
10
rapidly after arc initiation.
11
3.4.2 Sensitivity to volume
12
The peak pressure does not depend on the arc compartment volume as long as the process can be considered as
13
a slow process and the exhaust compartment is relatively large. Moreover, the time to reach the peak pressure
14
is proportional to the volume (Figure 3-5). This means that increasing the volume will slow down the process
15
but will not change the type of process: the overshoot level and the maximum pressure will remain the same.
0.506 m
0.253 m
0.6
0.329 m
0.405 m
0.55
0.481 m
0.5
0.557 m
0.45
0.633 m
[MPa]
0.708 m
0.4
0.784 m
0.35
0.860 m
0.936 m
0.3
1.012 m
burst 0.42 MPa
0.25
0.5
0.15
0.4
0.1
0.08
[s]
0.2
0.18
0.28
0.38
0.48
0.58
0.3
0.2
0.1
[s]
0.68
0.78
0.88 time peak

0.98 pressure
0
0.2
0.4
0.6
0.8
[m]
16
17
18
Figure 3-5: Pressure development over time with varying volume of arc compartment [m3] (simulation for
air).
19
On the other hand decreasing the volume will accelerate the process and for smaller volumes the peak pressure
20
may increase notably (the process can no longer be considered as slow), especially with respect to an
21
asymmetrical current (Figure F-8 in ANNEX F:).
Page 45
3.4.3 Sensitivity to opening area and bursting pressure
Arc compartment opening area 12 and bursting pressure have an important influence on the pressure curve.
It follows from criterion (3-5) that increasing 12 or reduces the overshoot level.
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
10
pressure, the lower the maximum pressure. For the case shown in Figure F-10 in ANNEX F:, the maximum
11
pressure reduction is 6 % for air and 15 % for SF6 for 50 % of bursting pressure reduction. It should be noted that
12
increasing the bursting pressure beyond a certain level, for which the overshoot is absent, leads to the trivial case
13
where the maximum pressure equals the absolute bursting pressure.
14
3.4.4 Sensitivity to exhaust volume
15
As explained in Section 3.3.3, the exhaust volume influences the pressure curve in the arc compartment. Figure
16
3-6 shows that the maximum pressure may also be affected. There is no overshoot in this case so the pressure
17
drops directly after the burst and increases again about one hundred of milliseconds afterwards. The arc must last
18
a sufficiently long time in order to buildup the gas pressure in the exhaust compartment, so that it starts affecting
19
the pressure in the arc compartment.
The opening area of arc compartment 12 is one of the principal parameters controlling the gas outflow and thus
1.275
0.4
0.638
0.35
0.828
[MPa]
0.3
1.02
1.211
0.25
1.403
0.2
1.594
0.15
1.785
1.976
0.1
0
20
21
22
0.2
0.4
0.6
0.8
1
[s]
1.2
1.4
1.6
1.8
2.168
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).
23
It is difficult to formulate more precise rules to describe the mutual interaction between the pressures in arc and
24
exhaust compartments. It should be noted that the influence of the latter is more important for SF6 than for air.
Page 46
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.
0.45
0.4
0.35
1.5
[MPa]
0.3
9.75
1.2
1.425
0.25
1.65
1.870
0.2
2.1
2.325
0.15
2.55
2.775
3
0.1
0.05
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
[s]
Figure 3-7: Influence of gas initial pressure (air).
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
10
temperatures there is no overshoot. If the initial temperature decreases, the maximum pressure and the time to
11
peak increase slightly. Considering that the initial temperature varies only within the ambient temperatures
12
permitted by the standard, the initial temperature has a limited effect on the pressure curve.
Page 47
0.43
0.38
[MPa]
0.33
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
[s]
1
2
Figure 3-8: Influence of gas initial temperature (air).
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.
0.43
[-38 C, 0.12 MPa]
0.38
[ -5 C, 0.14 MPa]]
[ 20 C, 0.15 MPa]]
[MPa]
0.33
[ 50 C, 0.16 MPa]]
0.28
0.23
0.18
0.13
0.08
0
10
11
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
[s]
Figure 3-9: Initial pressure and temperature combined as gas density (air).
Page 48
3.4.6 SF 6 vs air
Replacing SF6 by air modifies the pressure curve in many aspects.
a. The rate of pressure rise depends on an adiabadic constant: for air is more than four times
b. After the bursting of the pressure relief device, the process is slower for SF6: time to peak for air is about
higher than for SF6 (precisely 4.27); thus is shorter for air than for SF6.
6
7
8
55 % of the time to peak for SF6 under overshoot conditions.

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, 0.73466 (see criterion (3-5)). For any
10
SF6 case with the ratio greater than 0.7346 and smaller than 1 (0.7346 < < 1), replacing SF6 by air will
11
0.7346 the overshoot is absent for both air and SF6.
12
turn on the overshoot condition. If overshoot is present for SF6 ( > 1) then it will be higher for air. If <
d. The peak pressure is lower by about 25 % on average for SF6 than for air as long as the overshoot is
13
present. If there is no overshoot in the arc compartment, the maximum pressure remains unchanged and
14
equals the absolute response pressure ( ) for both gases.
15
16
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.
17
3.4.7 Impact of input parameters (summary)
18
The sensitivity analysis also allows the evaluation of the impact on results caused by the lack of precise test data
19
and tolerances of input data, including the switchgear geometry. Table 3-1 illustrates the worst errors on peak
20
pressure for all test cases when the input parameters are varied by 10 % around the reference value giving a
21
bandwidth of 20 % variation.
Page 49
Peak pressure variation
Parameter
Slow process Fast process

20 % of variation ( 10 %)
Arc power ( )
Air
SF6
Air
SF6
14 % 13 % 18 % 19 %
Arc compartment volume

Arc compartment relief opening
Arc compartment response
pressure
7%
3%
19 % 19 %
13 % 12 % 15 % 12 %
19 % 19 % 20 % 10 %
Arc compartment initial

temperature
Arc compartment initial pressure
Exhaust compartment relief
opening
Exhaust compartment
temperature
6%
6%
8%
6%
6%
6%
8%
15 %
0%
4%
10 % 11 %
0%
1%
2%
1%
Table 3-1: Variation of peak pressure when input parameters are varied (worst of test cases).
1
2
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.
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
7
8
includes ,1 and ,1 and the absolute bursting pressure includes ,1 and ,2 .
The most sensitive parameters are:
10
energy input rate (combining arc voltage, arc current and thermal transfer coefficient ),
11
gas type
12
burst characteristics (including bursting pressure and opening area)
The parameters with limited impact are:
13
asymmetry of current (d.c time constant and angle)
14
initial conditions (density of gas)
15
There are two types of processes: slow and fast. A process is considered fast if the pressure relief device bursts
16
within one period of the power frequency. The impact and importance of the input parameters differ for slow and
17
fast processes.
18
For slow processes, the arc compartment volume has practically no impact on the maximum pressure in the arc
19
compartment. However an increase in arc compartment volume slows down the process before and after the
20
opening of the relief device. An easy method is provided in 3.3.2 for the assessment of the overshoot condition for
Page 50
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 51
4 CALCULATION OF PRESSURE WITH CFD SOFTWARE
4.1 Introduction
While the basic model described in Chapter 2 delivers spatial averaged pressure values (one pressure value per
compartment), Computational Fluid Dynamics (CFD) software provides spatial resolved 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
10
switchgear experiencing an internal arc, especially for MV switchgear.
11
4.2 Motivations for CFD
12
CFD allows the following to be considered:
13
Space-dependency of an arc model (energy input, vaporization..) and results
14
Dynamic processes like reflection, diffraction and interference of gas flows (wave theory)
15
Compressible gas flow
16
17
CFD is typically used in:
18
Assessing the actual geometry of the switchgear and installation room (simulating actual electrical
installations when they differ from the manufacturers requirements or from the test conditions).
19
Analysing the influence of the location of pressure relief openings in rooms.
20
Analysing the influence of specific flap design, the influence of grids and absorbers.
21
In all these cases the expected short-circuit conditions of the actual power system are usually considered instead
22
of the rated short-circuit values.
23
24
For example: A switchgear tested with a straight pressure relief duct on the top of the installation must be used with
25
a rectangular duct (Figure 4-1). It is expected that the duct angle creates a jump in the pressure, therefore creating
26
pressure increases upstream in the switchgear. CFD is used to evaluate the extent to which this happens.
27
28
Note: A methodology for comparison between an actual and a reference switchgear is defined in Section 7.3.
29
CFD requires a dedicated workflow, which is described in Section 4.3.
30
Page 52
(a)
1
2
(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
9
10
USER-INTERFACE
11
PROGRAMMING
12
13
INTERFACE
Specification of boundary conditions

and calculation settings
PROGRAMMING
INTERFACE
Gas Data
Arc model
14
15
EQUATION SOLVER
16
Transient solution of equation system, which is
17
solved by iteration.
18
19
20
21
PROGRAMMING
POST-PROCESSOR
PROGRAMMING
INTERFACE
Presentation of results e.g. spatial
INTERFACE
e.g. Absorber
resolved pressure developments
Future effects
22
23
Figure 4-2: Workflow for CFD.
Page 53
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 SF6 or mixture) volume is
called the fluid domain. It is usually modelled 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)
6
7
8
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
10
(also called finite elements). Each (stationary) volume will have its own values of gas velocity (vector in 3 space
Page 54
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 modelled. 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
10
space distribution of a short pressure peak. This has to be considered for wave reflection, diffraction and
11
interference. For instance - after the rupture of a disk in a GIS or the flow of a pressure wave in an AIS switchgear
12
room.
13
There is no need to define a discharge coefficient for pressure relief device opening, as long as it is a free opening
14
(e.g.: hole in the wall).
15
When pressure relief is made using a grid, a labyrinth or all types of porous media [Besnard2008], the pressure
16
drop across this device must be modeled. Formula (4-1) can be implemented in CFD:
17
18
19
20
21
1
= + 2 2
(4-1)
Where [Pa = N/m = kg/(ms )] is the pressure jump, is the device thickness, the gas velocity, [kg/(ms)]
2
the fluid viscosity, the fluid density, [m ] and 2 [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 2 is
sufficient to describe the porous media. This coefficient is called the head loss coefficient.
22
Figure 4-4 shows an example of two-dimensional CFD simulation results around pressure relief openings with a
23
labyrinth. It shows the pressure relief opening device and the simulated pressure drop across it in 2D. The air flows
24
from right to left; the red color indicates that the pressure is increased upstream. The use of the head loss
25
coefficient removes the need for a full geometric description of the device, calculating the same pressure drop
26
using the head loss coefficient as a boundary condition.
27
Page 55
(a)
1
2
(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
6
7
thermal transfer coefficient described in Chapter 2.
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
10
values (11000 K in air typically), whereas the arc compartment still includes cold gas regions, as do the other
11
compartments. This accounts for the high temperature gradients existing during the internal arc event, leading to
12
high density gradients.
13
The most complete approaches, where the arc would be modeled using physical equations describing the arc
14
roots, the arc plasma column, the effect of electro-magnetic fields on the motion of the arc, the transfer of energy
15
from the arc plasma to the surrounding gas etc. have never been applied to internal arc to our knowledge.
16
All enhancements described in ANNEX A.3 and A.4 for the basic model can also be considered in CFD, especially:
In a more detailed approach [Besnard2009], the arc heating power is confined to a small number of finite volumes
17
Gas model : temperature dependance, or temperature and pressure dependence; real gas law
18
Material evaporation
19
Thermal transfer in absorbers
20
In addition, turbulence and heat transfer can also be considered. Compared to the basic method, CFD allows a
21
space-dependency to be chosen for all enhancements.
Page 56
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 onnes, 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
10
another. In a three-dimensional transient analysis, results are time- and space-dependent.
11
Initial conditions (gas temperature, gas pressure, gas velocities) shall be defined by the user, as well as boundary
12
conditions, which are appropriate to the solver used.
13
4.3.4 Examples of CFD results: post-processing activities
14
Finally the post-processor is used to present simulation results. Pressure, temperature and flow velocity are
15
available at each point of the flow domain over time. The solver can calculate average gas pressure on surfaces or
16
in volume automatically, if the respective surface or volume is defined.
17
Post-processing tools allow animations (videos) of the flow, which provide a visual support of the phenomena to be
18
created. Example of the distribution of pressure, temperature and gas velocity at 55 ms after the initiation of
19
internal arc is shown in Figure 4-5, Figure 4-6 and Figure 4-7 respectively.
Page 57
Duct exhaust
Arc location
1
2
3
Figure 4-5: Isovalue of pressure (relative) (Pa) Case of an AIS switchgear, 55 ms after arc ignition (50
kA).
4
5
Figure 4-6: Isovalues of temperature (K) Case of an AIS switchgear, 55 ms after arc ignition (50 kA).
Page 58
1
2
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
10
the influence of the location of pressure relief openings can be analysed.
11
The accuracy depends on the quality of the models used to describe the internal arc physics, as well as on an
12
appropriate meshing of the flow domain.
13
In the end, it is pointed out that the use of CFD for internal arc simulation is complex. It is recommended that a CFD
14
methodology be calibrated with actual test results, on a sufficient sample of cases. Furthermore, accuracy depends
15
also on the mesh quality and therefore on the knowledge and engineering skills of the user.
16
REFERENCES:
17
[Besnard2008]:
18
internal arcing fault, REE. 11, 35-38.
19
[Besnard2009]: C. Besnard, 2009, Internal arc simulation in MV/LV substations, CIRED 2009, paper 0536.
C. Besnard, 2008, Numerical simulation for design of safe electrical equipments in case of
20
Page 59
5 INTERNAL ARC TESTING
5.1 Introduction
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
10
of arcs burning in air and in SF6.
11
Internal faults inside metal-enclosed switchgear can occur in a number of locations and can cause various physical
12
phenomena. The arc energy resulting from an arc in any insulating medium within the enclosure will cause an
13
internal overpressure and local overheating which will result in mechanical and thermal stressing of the equipment.
14
Moreover, the materials involved may produce hot decomposition products, in the form of gases or vapours, which
15
may be discharged to the outside of the enclosure, and endanger personnel or the general public.
16
Relevant tests are defined in the IEC standard IEC 62271-203 (for GIS), IEC 62271-200, 201, and IEEE guide
17
C37.20.7 (for metal/insulation enclosed-switchgear).
18
19
20
Figure 5-1: Indicator racks located at the front and side of an MV panel for internal arc testing.
21
5.1.1 GIS equipment > 52 kV
22
Evidence of the internal arc withstand ability of the enclosure against bursting and burn-through shall be
23
demonstrated by the manufacturer when required by the user. The IEC 62271-203 standard allows this ability to be
Page 60
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
carried out with the normal insulating gas, usually SF6, at rated filling density. The switchgear is considered
adequate if no external effect 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
10
Internal arc testing of metal-enclosed switchgear is intended to offer a tested level of protection to persons in the
11
immediate vicinity of the switchgear in the event of an internal arc. Effects from an internal fault arc, such as
12
overpressure acting on covers, doors, inspection windows etc., as well as the thermal effects of arc(s), arc roots,
13
ejected gas(es) and glowing particles are included. In contrast to internal arcing in GIS > 52 kV, the standards IEC
14
62271-200, IEC 62271-201, and IEEE C37.20.7, allow no possibility of verifying internal arc withstand ability by
15
calculation, even when based on the testing of equivalent designs. For this reason, and because of the generally
16
closer proximity of medium voltage installations to the public compared to high-voltage installations, internal arc
17
testing of metal-enclosed medium voltage switchgear is very common.
18
5.2 Standardisation and Test Experience
19
5.2.1 Standardisation status
20
With the release of IEC 62271-200 in 2003 an Internal Arc Classification (IAC) was defined, taking into account
21
various levels of accessibility of the switchgear:
22
Type A: Accessible by authorised personnel only;
23
Type B: Accessible by general public;
24
Type C: Not accessible, i.e. out of reach (pole-mounted switchgear);
25
For assessment of the thermal effects of the hot gases expelled from the installation due to pressure rise caused
26
by the fault arc, special black cotton cloth indicators of size 15x15 cm are used in a steel frame to avoid mutual
27
ignition (see Figure 5-1). The indicators are mounted on a rack (vertically and horizontally) arranged in a
28
checkerboard pattern, covering 40-50 % of the area of the accessible switchgear side, or of the 3x3 m area below
29
the switchgear in case of pole-mounted apparatus.
30
An important (by test-experience the most critical) criterion to pass internal arc tests besides the mechanical
31
withstand of the enclosure is the absence of ignition of any indicators by hot gases. Ignition by glowing particles,
32
however, is allowed. High-speed video is normally used in order to make a distinction between the causes of
33
ignition. However, in many cases, the real reason of ignition (hot gases or particles) cannot be identified, for
34
example if the flight of a particle is not in the area of sight of the camera.
35
Depending on the accessibility type, two degrees of flammability of the cotton indicator cloth are required,
Page 61
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 authorised personnel and the lighter for the general public's
garments.
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.
10
Acceptance criteria to qualify the switchgear for an IAC classification are the following:
11
Criterion 1: Doors and covers must not open. Deformations must not touch the indicator racks or walls;
12
Criterion 2: No parts above 60 g must be projected; enclosure must remain intact during arcing;
13
Criterion 3: Arc must not burn-through an accessible side lower than 2 m high;
14
Criterion 4: Indicators must not ignite due to the effect of hot gases;
15
Criterion 5: The connection of the enclosure with the earthing point remains intact.
16
Although it is easy to measure and its the most important factor, IEC / IEEE standards do not request a pressure
17
measurement in the arc compartment during testing. Another point to note is that, due to the different and
18
sometimes conflicting conditions in testing ( e.g. in temperature rise tests ventilation openings are welcome and in
19
internal arc testing ventilation openings are not welcome), a proper identification of the equipment and relief areas
20
of the equipment which was tested should be included in the test reports, in the absence of guidelines from the
21
main technical standards..
22
5.2.2 Test result statistics
23
Results of internal arc tests have been analysed by a major test laboratory. Most recent statistics are based on 91
24
tests carried out in 2005 and 2006. It is the test laboratory's experience that in approximately 80 % of the tests all
25
criteria have been fulfilled, see Figure 5-2. The most prominent failure mode is related to criterion 4: the absence of
26
ignition of indicators. In 15 % of all tests, indicators ignited.
27
By comparison, earlier data (2001-2002 when IEC 60298 was in use) are also evaluated (from 137 tests): 32 %
28
did not fulfill all criteria; also in that period, in 23 % of the tests vertical indicators ignited, and in 11 % horizontal
29
ones (IEC 60298 made a distinction between ignition of vertical and horizontal indicators).
Page 62
100%
not fulfilled
fulfilled
80%
60%
40%
20%
0%
1
2
3
4
5
criterium as specified in IEC 62271-200 A6
total result
Figure 5-2: A major test laboratory's experience with achievement of criteria of IEC 62271-200
(population: 91 tests in 2005 and 2006).
2
3
4
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:
7
8
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).
10
Many test laboratories (including some belonging to the Short-circuit Testing Liaison) do not issue certificates on
11
internal arc tests only. The reasons for this are as follows:
12
Identification of the relevant parameters which determine test results in the design verification is not clear.
13
A well-defined system of documentation does not exist yet. Too little is known about what design
14
parameters determine a positive test result.
15
Relation of test voltage to rated voltage.
16
There are questions regarding the reproducibility of test results.
17
The variety of rooms in which switchgear is located in service can never be simulated in test with a single
18
representative test situation.
Page 63
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.
3
4
5
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 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
10
asymmetrical peak at reduced (low) voltages. The arc voltage counteracts the driving voltage, thereby
11
reducing the arc current (asymmetrical peak value as well as a.c. component). An example of this is shown in
12
Figure 5-3, showing current wave traces from a 36 kV (rated voltage) switchgear, tested with a source voltage
13
of 7.7 kV. The measured prospective and arc current at the applied voltage of 7.7 kV are indicated together
14
with the calculated current at a supply voltage of 36 kV. For the calculated current at 36 kV the arc voltage
15
from the measured at 7.7 kV is used. Figure 5-3 demonstrates the following asymmetrical peak values:
16
prospective peak is 2.55; the measured peak at 7.7 kV is 2.05 while the expected value at 36 kV supply
17
voltage is 2.44.
18
This reduction of peak factor at lower supply voltages is illustrated in Figure 5-4 to some more extent.
19
Guidelines can be given on the ratio of supply voltage / rated voltage ( = / ) that does not lead to
20
21
unacceptable asymmetrical peak reduction.
In Figure 5-4, the asymmetrical peak factor (as a fraction of the prospective peak) is plotted against for a
22
number of rated voltages. Following the IEC standard 62271-200, supply voltage reduction is allowed, as long
23
as the fraction of prospective peak current is not lower than 0.9. The dot shows as an example the case of
24
17.5 kV switchgear, tested with a supply voltage of 0.8x17.5 = 14 kV and leading to a peak reduction of 0.92,
25
within the limit set by the standard. It can be concluded that supply at 7.7 kV resulted in an asymmetrical peak
26
value more than 10 % below the prospected value, making the test invalid according to the standard IEC
27
62271-200.
Page 64
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.
7
8
9
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
10
assuming a constant arc voltage) equals the specified value (within 0 to 10 % tolerance). This under the
11
provision that the first three a.c. half-cycles are as specified and that the current shall not be reduced by more
12
than 50 % of the specified value at the end of the test.
13
Such calculations, however, can only be performed in hindsight, since arc behaviour cannot be predicted. This
14
leads to imprecise representation of stresses.
fraction of prospective peak factor

0.95
36 kV
24 kV
0.9
17.5 kV
prospective peak
actual peak
2.5
0.85
2
1.5
12 kV
1
0.5
0.8
0
-0.5
testvoltage
15
16
17
18
0.75
0.4
0.5
-1
0.6
0.005
0.01
0.015
0.7
0.02
0.025
0.8
0.03
0.035
0.04
0.9
0.045
0.05
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).
19
Page 65
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. 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: note2
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-sulphides and -fluorides) as well
as gaseous SF6 decomposition products (SF4, H2S, SO2, HF, CF4, S2F10, S2OF10) are mostly very poisonous,
especially in the presence of humidity. In addition, test laboratories wish to minimise their emission of clean SF6, a
10
greenhouse gas, and certainly of polluted SF6. Technically, however, it is not clear yet, that testing in air presents
11
similar conditions as testing in SF6.
12
A short overview of some past published work by manufacturers is presented in Section 5.2.5.1, based on three
13
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
14
arcing phenomena in SF6 and air (and other gases).
15
5.2.5.1 RESULTS FROM THE LITERATURE
16
The first series of tests was performed with a transformer substation model, with a 0.3 m SF6 compartment
17
[Daalder1989]. Three-phase current (15 - 20 kA) was supplied from a 7.2 kV circuit to arcs between Cu electrodes
18
with 100 mm distance. The main difference between SF6 and air was found to be:
19
A significantly higher arc voltage in SF6 (720 V) than in air (480 V);
20
The period during which pressure is above 50 % of its maximum value in case of SF6 is 2.4 times longer
21
than in air (in this test).

3
22
In a second research program a series of tests were performed with an arcing volume of 0.2 m , exhausting in a 8
23
m air-filled volume [Dullni1994]. Test current was single phase (10 - 20 kA) and arcing was in a 100 mm gap. The
24
importance of these tests is the measurement of pressure-rise in the exhaust volume, as a simulation of the room
25
in which the switchgear is located. It was concluded that:
26
Rise and fall (after pressure relief) of pressure in the arcing volume are faster in air than in SF6;
27
The pressure rise in the exhaust volume is twice as high when the arc was burning in SF6. This is attributed
28
29
to the higher heat capacity of SF6 and the longer duration of stay in the arcing volume;
30
A wide range of arc voltages has been encountered, in SF6 (500 - 1900 V) notably higher than in air (400 1000 V).
31
A third series of tests were carried out with "typical gas-insulated metal-enclosed switchgears", with three-phase
32
current of 16 kA [Bjrtuft2005]. The authors conclude on the difference between internal arcing in SF6 and in air:
33
34
35
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 66
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
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, SF6 for the highest and the
lowest current values applied in the larger tank volume (0.24 m ). It can be observed that the pressure-rise at both
10
current values is slower in SF6 than in the other gases, and thus confirms the other sources [Daalder1989,
11
Dullni1994, Bjrtuft2005].
pressure rise (b)
18.3 kA
4
4.6 kA
Ar 1.2bar
Air 1 bar
N2 1 bar
SF6 1.2 bar
1
0
100
200
300
400
500
600
700
arc duration (ms)
12
13
14
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).
15
5.2.5.3 SINGLE-PHASE TESTS IN A TANK INCLUDING PRESSURE RELIEF
16
A series of new tests were performed, purely aimed at comparing the effects of arcing in SF6 with air under
17
identical conditions of current, driving voltage, arc duration, geometry, contact material etc. Test parameters were:
18
arc current 14.2 0.3 kA, driving voltage 15.5 kV, frequency 60 Hz, tank volume 0.53 m , electrode material Cu,
19
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
20
disk set for pressure relief at approximately 2.8 bar.
21
Measured quantities were arc current and voltage, tank pressure. The exhaust gas was monitored by infrared
22
thermography; the arc was monitored with high-speed video.
23
An impression of the effect of the exhaust gas can be obtained from Figure 5-6, where tests with arcing in SF6 (left)
24
and air (right) are compared (arc duration 1 s with symmetrical current).
25
A dummy was placed 5 m from the exhaust opening (see Figure 5-6). The different effect of SF6 and air exhaust
26
gases on the T-shirt, worn by the dummy, is illustrated in Figure 5-7. This observation, however, does not allow a
27
scientific evaluation, though the clothing which interacted with SF6 is most affected.
Page 67
1
2
The different effects of arcs burning in air and SF6 are:

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):
6
7
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;
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.
10
11
12
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.
13
14
Figure 5-7: T-shirt worn by dummy and affected by SF6 gas (left) and air (right).
15
Page 68
6
5
arc energy (MJ)
air 1 s
4
SF6 1 s
pressure relief action
air 0.5 s
SF6 0.5 s
1
0
200
1
2
400
600
time (ms)
800
1000
1200
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 by Schumacher et al, 1994
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
10
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,
11
whereas the arc in SF6 shows its highest arc voltage after 0.5 s when the pressure is still rising.
12
In the literature there are quite different statements about the value of SF6 arc voltage compared to that in
13
air at the same filling pressure: In some experiments a lower voltage was obtained, in others a higher one.
14
If a pressure-dependent arc voltage is assumed, these differences may easily come from the different
15
development of pressure during arcing. The different behaviour of air or SF6 arcs on electrodes and their
16
different instabilities may also play a part in this.
17
18
2)
Cooling of exhaust gas after arcing. By monitoring the exhaust jet with time-resolved infra-red
19
thermography, an impression of the cooling process of the hot gases could be obtained. This is illustrated
20
in Figure 5-10, in which the maximum temperature in the exhaust gas volume is shown over time. In Figure
21
5-11, infrared pictures of the exhaust gas are shown for SF6 and air, recorded at comparable time after arc
22
initiation.
23
From this, it becomes clear that:
24
25
26
2000 C, since the IR camera saturated above this level).
27
28
the exhausted air is hotter than the exhausted SF6 (actually the maximum temperature in the air is above
the cooling down after arcing in SF6 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.
Page 69
1
4
pressure rise (bar)
arcing volume
3
1
exhaust volume
0
500
1000
3
4
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
time after arc initiation (ms)
1500
1800
2100
6
7
Figure 5-10: Estimated maximum temperature of exhaust gas (air, SF6) vs time after arc initiation.
Page 70
1
2
3
4
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-
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];
10
b. The pressure relief disc operates at a (somewhat) higher pressure in air than in SF6. This is (probably)
11
because the inertia of the disc bursting process causes the disc to operate at a somewhat higher
12
pressure when the rate of pressure rise is higher. This phenomenon is frequently observed in tests.
13
c.
The pressure in the exhaust volume reaches a higher value in the SF6 case than in the case with air.
14
From these tests it follows that the arc energy of arcs in air in a worst-case situation could be regarded
15
as being lower than in SF6.
16
d. Pressure inside the arc compartment. As shown by several authors the maximum pressure in a closed
17
arc compartment is higher if it is filled with air instead of SF6. This can be true even when the arc
18
energy in SF6 is higher than in air. The reason is the larger heat capacity of SF6, which compensates
19
for the higher arc energy. The smaller heat capacity of air leads to a faster pressure rise and an earlier
20
bursting of the rupture disc. When the pressure relief disk opens it limits the pressure rise in the arcing
21
compartment. There is a tendency for the opening to occur at a higher pressure in air due to inertia
22
effects. From this it follows that the arc compartment will be stressed in a comparable or even more
23
severe way by an arc in air.
24
e. Exhaust of gases via an intermediate compartment. If the overpressure of the arc compartment is
25
directly discharged into the environment (room), the hot gas stream will affect the indicators
26
immediately. However, in general, metal-enclosed switchgear consists of several compartments with
27
only the "arcing" compartment filled with SF6. In this case, hot SF6 first of all will exhaust to a neigh-
28
bouring air-filled compartment (intermediate room, e.g. cable compartment, pad mount enclosure)
29
within the switchgear before leaving it e.g. through fissures.
30
4) Ignition of indicators. Ignition of any material means exothermic reactions start to run. This happens
Page 71
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
9
10
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 SF6), the more energy will be
11
12
13
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.
14
d. Gas velocity: The higher the gas velocity (as with air), the higher is the turbulence in enlarging the
15
temperature at the boundary of the gas stream (increasing the temperature gradient to the indicator).
16
Moreover, with higher gas velocity, the boundary layer between hot gas and the surface of the indicator
17
is thinner. Higher turbulence and a thinner boundary layer improve heat conduction to the indicator and
18
can lead to easier ignition of indicators.
19
Given that these various contributing factors have both ignition-enhancing and -impeding effects in air and
20
SF6 further research is necessary to determine the overall effect.
21
5.3 Summary
22
23
Based on the results, the authors conclude that relevant differences exist in the behaviour of fault arcs in SF 6 and
24
in air, and in their effects on switchgear and the environment.
25
Replacing SF6 by air (all other parameters being equal) in internal arc testing leads to comparable or higher
26
pressure rise in the arcing compartment. Pressure rise in an adjacent compartment or a switchgear room into which
27
the exhaust gas is released may be lower in tests with an air-filled arc compartment than in tests with an SF6-filled
28
arcing compartment. No conclusions exist on other criteria required to pass an internal arc test such as the ignition
29
of the cotton indicators and enclosure burn-through. This needs further detailed investigation.
30
The following conclusions are drawn:
31
a. Arc compartment: The mechanical stress of the arcing compartment filled with air is higher than
32
with SF6, i.e. if criterion 1, 2 of IEC 62271-200 are fulfilled with an air test, this will be true for SF6
33
as well.
34
35
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.
Page 72
c.
Indicators: With the main focus directed to the ignition of the indicators, the following conclusions
are drawn if air is used instead of SF6 during internal arcing tests (criterion 4 of IEC 62271-200):
3
4
For worst-case situations (e.g. long arc duration) the arc energy in SF6 can be higher than
in air.
5
6
With air, the higher gas temperature as well as the thinner boundary layer in front of the
indicators will increase the heat flux applied to the indicators.
7
8
Thermal conductivity of the gas determines the heat flux as well, hot SF6 has a higher
conductivity than air in the relevant temperature range.
The gas stream duration is longer with SF6 and it cools down slower than with air.
10
Both gas streams with, i.e. with SF6 or air, are longer than the distances from the panel to the cotton indicators
11
even considering the reflections at walls and ceilings. Although more test results would be needed to confirm this,
12
the indications are that, in this respect, the effects are comparable.
13
Also, the authors noted that IEC / IEEE standards currently do not request a pressure measurement in the arc
14
compartment during testing. A lot of useful information which could be used in the design of the switchgear can be
15
obtained from the measurement of the overpressure curve.
16
17
REFERENCES:
18
[Bjrtuft2005]: T. Bjrtuft, O. Granhaug, S.T. Hagen, J.H. Kuhlefelt, G. Salge, P.K. Skryten, S. Stangeherlin,
19
"Internal Arc Fault Testing of Gas Insulated Metal Enclosed MV Switchgear", 18th Int. Conf. on El. Distr. (CIRED),
20
2005.
21
[CIGRE WG 23.03]: CIGRE WG 23.03, Pressure Rise in Metal-Enclosed Switchgear of Single Phase Enclosure
22
Type due to Internal Arc. Evaluation of various International Test Results and Study of Calculation Procedure,
23
Electra 93, pp. 25-52, 1984.
24
[Daalder1989]: J.E. Daalder, O. Lillevik, A. Rein, W. Rondeel, "Arcing in SF6-MV-Switchgear. Pressure Rise in
25
Equipment Room", Int. Conf. on El. Distr (CIRED), 1989.
26
[Dullni1994]: E. Dullni, M. Schumacher, G. Pietsch, "Pressure Rise in a Switchroom Due to an Internal Arc in a
27
Switchboard", Proc. 6th Int. Symp. on Short-Circuit Currents in Power Systems, Sept. 6-8, 1994.
28
[Pettinga1989]: J.A.J. Pettinga, "Pressure-rise tests due to a High-Current Internal Arc in a MV Cubicle Model", Int.
29
Conf. on El. Distr. (CIRED), 1989 and KEMA internal report 00880-DZO 88-2046 (88-10), 1988.
30
[Trinh1992]: N. Giao Trinh, Risk of Burn-Through a Quantitative Assessment of the Capability of Gas Insulated
31
Equipment to Withstand Internal Arcs, IEEE Trans. On Pow. Del., vol. 7 no. 1, 1992.
32
33
Page 73
6 EFFECT OF INTERNAL ARC ON STRUCTURES
6.1 Introduction
The internal arc is initiated at a particular place in the switchgear during a type-test (see IEC62271-203 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
9
10
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
11
on the enclosure structure causing deformation of walls, blown out doors, etc. Generic constructions are
12
shown in Figure 6-1.
13
a. In Figure 6-1 (a), a typical construction of air insulated switchgear (AIS) with grounded enclosure
14
used in medium voltage is shown. The walls of the enclosure are made of a sheet metal with a
15
certain thickness connected by bolts spaced by a certain distance. For simplicity, the doors are not
16
shown in the figure. The pressure relief devices are flaps which are structurally weaker than the
17
main enclosure, so that in case of high pressure inside the panel they will be first to rupture and
18
release the excess pressure. Usually the flaps start to open at an internal pressure around 20 to 30
19
% higher than the external pressure.
20
b. Figure 6-1 (b) shows a typical construction of a gas insulated enclosure used in medium voltage
21
switchgear (GIS). The walls of the enclosure are made of a sheet metal with a certain thickness
22
and the walls of the enclosure could be welded or bolted together. The pressure relief device is
23
usually a rupture disk which opens at a specific overpressure, for example 200-300 kPa depending
24
on the application and the protection philosophy.
25
c.
Figure 6-1 (c) shows a typical construction of a high voltage GIS busbar arrangement. The external
26
envelope is composed of a metallic cylinder with a certain thickness which contains a concentric
27
conductor. Insulating spacers support the conductors and divide the GIS into separate gas
28
compartments. The pressure relief device is usually a rupture disk which opens at a specific
29
overpressure, for example 1000-1500 kPa depending on the application and the protection
30
philosophy.
31
The second effect (see Section 6.3) has an impact on the switchgear room and building. Hot gases will be
32
ejected through the pressure relief device of the switchgear enclosure and cause pressure buildup in the
33
rooms and buildings. The pressure relief devices of the installation room start to open when the
34
overpressure reaches a certain value.
Page 74
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.
(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.
4
5
6.2 Mechanical stress on switchgear due to the overpressure
The overpressure caused by internal arc event exerts 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.
10
The present section starts with a calculation method for static loads and then moves into complex structures and
11
dynamic loads.
12
Previous sections of this brochure have shown how to calculate the overpressure values generated during an
13
internal arc. Consider that the overpressure curve is known for a non-deformed geometry and an assessment of
14
the withstand strength of the enclosure during an internal arc is to be made. In the case where the deformation of
Page 75
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.
3
4
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.
10
6.2.1 Calculated results for the failure of metal plates due to static overpressure
11
In this section typical failure details of standardized metal plates due to static overpressure are presented. The
12
plate will deform elastically up until the yield point and then deform plastically until failure. The material properties
13
governing these deformations as well as the typical load and deformation values which a plate can sustain are
14
provided for principal alloys.
15
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
16
boundary conditions, with the plate edges fixed in different ways, are analyzed:
1
The plate edges are simply-supported. This is a simplified analysis for welded edges . See Type 1
in Figure 6-3.
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.
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.
1
The choice of 50 mm bolts simulates a rigid bolted connection.
Page 76
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.
3
All three types are calculated for yield and failure using materials AISI 304 and AISI 1010 .
Since the plates have non-standard supports and a large deflection for which an analytical solution doesnt exist, a
finite-element method (FE method) is used to calculate the structural strength. The material properties used are
given in Table 6-1.
Material
Youngs
Modulus
Yield
Strength
Tangent
Modulus
Ultimate
Strength
(Pa)
(Pa)
(Pa)
(Pa)
AISI 304
1.93E11
220E6
1.8E9
520E6
AISI 1010
2E11
305E6
1.45E9
360E6
Table 6-1: Material properties of two AISI alloys.
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.
9
10
AISI 304 is a corrosion-resistant stainless steel and AISI 1040 is a wrought carbon steel.
Page 77
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
under different boundary conditions.
plate under different boundary conditions.
2
# Material
Support
2 mm Plate
5 mm Plate
Page 78
6 mm Plate
Type
Pressure
value at
yield
Maximum
deflection at
yield
Pressure
value at
yield
Maximum
deflection at
yield
Pressure
value at
yield
Maximum
deflection at
yield
(Bar)
(mm)
(Bar)
(mm)
(Bar)
(mm)
AISI 304
1.2
18
1.5
15.5
2.1
14.8
2 AISI 1010
1.3
19.8
2.6
18.4
3.6
17.3
AISI 304
0.03
0.07
3.4
0.1
3.2
4 AISI 1010
0.07
7.6
0.13
5.0
0.2
3.7
AISI 304
0.15
8.2
0.45
7.6
0.5
6.8
6 AISI 1010
0.3
9.9
0.6
8.5
0.9
Table 6-2: Typical values at yield for plates.
2 mm Plate
# Material
Support
Type
Pressure
value at
failure
Maximum
deflection
before
rupture
(Bar)
6 mm Plate
Pressure
value at
failure
Maximum
deflection
before
rupture
Pressure
value at
failure
Maximum
deflection
before
rupture
(mm)
(Bar)
(mm)
(Bar)
(mm)
AISI 304
18.4
195
48
192
56
188
2 AISI 1010
87
18.9
83
21
76
AISI 304
1.9
52.7
4.8
50.6
6.3
50
4 AISI 1010
0.6
19
1.4
18
1.9
17
AISI 304
61.5
13.5
63
15.2
59
6 AISI 1010
2.4
24
4.7
21
5.5
20.7
Table 6-3: Typical values at failure for plates.
2
3
5 mm Plate
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.
10
The ductility of material AISI 304 helps it in withstanding higher overpressures even though it has low yield
11
strength.
12
This section provided the basic understanding of the effects of different types of material and support on the
13
mechanical strength of metal plates. A basic assumption is made that the load acting on the plates is static, which
14
means it is applied slowly so that dynamic effects can be neglected, but it has provided a guideline for the strength
15
of plates.
16
In the next section, the rate of rise of overpressure will be taken into account when estimating the mechanical
17
stresses during an internal arc.
Page 79
6.2.2 Calculation results for the failure of metal plates due to dynamic
overpressure
2
3
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, like hitting the wall with a hammer.
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.
10
Equation (6-1) is the governing equation.
()
+ 20
+ 02 =
2
(6-1)
11
12
13
14
15
Where is deformation, is time, is mass, () is force, 0 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
where

{ ( ) sin ( ) cos }
( ) =

( ) = (1 ) + () cos
(6-3)
( ) = (1 ) + () sin D
(6-4)
16
17
18
(6-2)
where is the damped natural frequency and is the damping.
So, for a given time-varying overpressure the deformation of a simplified plate can be calculated in the elastic
19
regime with equation (6-2). The stresses are calculated from the resulting displacement and compared against the
20
mechanical failure properties of the material.
21
It is important to remember that equation (6-2) is valid for displacement in the elastic regime only when there is no
22
permanent deformation.
Page 80
6.2.2.2 ELASTO-PLASTIC REGIME
When the load is so high that the structure undergoes plastic deformation, the above mentioned equation (6-2) is
not valid, as it doesnt 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
9
10
+ + =
where is displacement , is velocity and is acceleration for , which is the applied force, represents the
change in value during an infinitesimal time increment at the th time step, is the mass , is damping, and is
11
stiffness.
12
The solution of equation (6-5) yields
13
14
15
(6-5)
where , are effective stiffness and force:
=
= +
= +
6 3
+
2
+ 3 + 3 +
(6-6)
(6-7)
(6-8)
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
16
time step starting from the given initial condition. The stresses are calculated from the deformation plot of the
17
above equations.
18
As mentioned earlier the applicability of above equation becomes difficult for complex structures and high plastic
19
deformations. For example, take the case of a plate which is deformed by a given time-varying load and undergoes
20
plastic deformation. However this plastic deformation is not uniform throughout the plate. A part of the plate
21
undergoes plastic deformation whereas the other areas remain elastically deformed. This presents a problem for
22
stiffness being used in the above-mentioned formula at a given time step.
23
So the next logical step is to divide the plate into small discrete subsections and define the stiffness of these
24
subsections based on an elastic or plastic regime. The division of the structure into discrete elements can be
25
handled efficiently with Finite Element Analysis (FEA). The finite element method is a numerical method for solving
26
partial differential equations in physical systems. The solution of standard differential equations exists for simplified
27
cases. When complicated geometry or non-linearity are involved, an approximate numerical solution is the right
28
way to proceed. The finite element formulation results in simultaneous algebraic equations for the degrees of
29
freedom. These algebraic equations provide a solution for the primary variable at a discrete number of points rather
Page 81
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
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
10
deformation of this tank under two different conditions; in the first simulation the tank was filled with SF6 gas. In the
11
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
AISI, 1010, 6 mm thick
AISI 1010, 6 mm thick
13.8 bars in 70 ms (from

measurement)
13.8 bars in 12 ms (from simulations,

using basic model)
Steel:
Maximum Pressure:
12
Table 6-4: Two cases that are used for the FEA analysis of the mechanical deformation and von Mises
13
stress.
14
15
16
Figure 6-6: Deformation of structure after an internal arc test in the SF6-filled switch.
17
Page 82
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 SF6-filled switch tank. The simplified
representation of the tank is divided into discrete finite elements which are connected together, as seen in Figure
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

(simulation) after an internal arc test in an SF6-filled
tank.
Figure 6-8: von Mises stress for corresponding

deformation.
10
11
The order of magnitude of the deformation is valid and the shape of the test object after the test is well predicted by
12
the finite element computation. As can be seen from the above calculation the total residual deformation is 81.5
13
mm, which matches closely with the measured deformation in the picture of 81 mm.
14
6.2.3.2 SIMULATION WITH AN AIR-FILLED TANK
15
In the second case, the FEA analysis was carried out on the air-filled switch tank. Although the calculated
16
maximum pressure in this case is identical to the measured maximum pressure in SF6-filled tank, the time to peak
17
pressure is much less in the air-filled than in the SF6-filled tank. This excites the eigen modes (fundamental
18
frequencies) of the tank differently than the pressure rise in SF6.
19
The residual deformation of the tank is shown in Figure 6-9 and the corresponding von Mises stress in Figure 6-10.
20
The maximum deflection is 85.3 mm, which is higher than the deflection in SF6 (81.6 mm). So for this particular
21
tank, testing in air is more severe than testing in SF6.
Page 83
Figure 6-9: Total deformation of structure

(simulation) after internal arc test in air.
1
Figure 6-10: von Mises stress for corresponding

deformation.
The comparison between the mechanical deformation in the air-filled and SF6 filled tanks is shown in Figure 6-11.
2
3
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 rise time 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 was justified.
10
The bottom line is that by use of theoretical simplification the failure of a switchgear structure can be predicted.
11
Simplified calculation without dynamic effects guides the designer during the design stage. Inclusion of dynamic
Page 84
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.
6.3 Mechanical stress on building walls due to overpressure
In the previous section the first effect of the overpressure caused by internal arc that exerts a severe stress on the
enclosure of switchgear was analyzed. Using FEA software, the mechanical deformation and von Miss stresses of
the switchgear tank / enclosure can be simulated once the pressure rise inside the tank is known, thus providing
10
the designer with an important input for the switchgear construction.
11
In this section the second effect of the internal arc fault is analysed i.e. the mechanical stress on the installation
12
room and building walls as a result of the overpressure caused by ejection of hot gases through the pressure relief
13
devices.
14
Section 6.3.1 shows how to estimate the maximum pressure on the room walls. Section 6.3.2 lists the allowable
15
maximum pressures for different wall structures. Sections 6.3.3 through 6.3.6 show whether the basic calculation
16
method can be used to calculate the pressure in the rooms and provide an overview of CFD simulations. Finally,
17
Section 6.3.7 discusses the sizing of the pressure relief opening.
18
6.3.1 Estimation of the maximum pressure on the room walls
19
At the start of the internal arc event, the pressure field inside the room is not homogeneous in space. It is highly
20
influenced by:
21
22
23
The location of the switchgear pressure relief compared to the adjoining wall or ceiling. The closer the gas
exhaust to the wall, the higher the local pressure.
The wave reflections, especially on room corners, which generate local pressure peaks.
24
However, the impact of the first local pressure peaks is weighted by the area involved.
25
It is possible to differentiate two behaviors, according to the power surface density P/S:
26
. P is the arc power: P = 3 x I x Uarc for three-phase fault (MW)
27
. S is the area of room pressure relief opening (m )
28
For the high power density (P/S typically above 46 MW/m from [Douchin2013]) when the peak occurs, the
29
pressure is close to being homogenous, with a distribution within 20 % around the average value. Thus, the
30
average room pressure is representative of the stress applied to walls at this instant of time.
31
For the low power density (P/S typically lower than 46 MW/m from [Douchin2013]) the pressure is not likely to
32
become homogenous in space. It is mainly the results of waves propagating, and the local pressure can vary from
33
zero to twice the average pressure on a given wall. The space average pressure is thus less representative of wall
34
stress.
Page 85
2
When the power density P/S is higher than typically 46 MW/m , the maximum pressure applied on a given wall can
then be estimated to be 1.2 times the peak of the average pressure within the room.
For all power density ratios, the maximum local peak pressure is usually several times the room average peak
pressure.
As these results are time-dependent, dynamic effects may be considered in wall sizing, according to the relevant
civil design code.
Example of pressure distribution within a room: Local peak at the beginning of the event, and almost
homogenous pressure later.
In this example, the AIS switchgear pressure relief device is located on top of a vertical chimney inside the
10
switchgear. The gases are therefore directed to the ceiling. Right above the pressure relief device, on a small
11
surface area, the pressure reaches a high value for a short time.
Local peak pressure. Note

that at this instant the
pressure is zero at the
other side of the ceiling.
Room pressure relief
Switchgear
12
13
14
Figure 6-12: Example of pressure field within a switchgear room. Beginning of the event (11 ms after fault
initiation).
15
Page 86
Switchgear pressure relief
Arc compartment
1
2
3
Figure 6-13: Example of pressure field within a switchgear room. Homogenous pressure stage (260 ms
after fault initiation).
On Figure 6-12, it can be seen that the maximum local peak is 15000 Pa, about 4 times the peak of the space-
averaged pressure.
In Figure 6-13, it can be seen that pressure increases from the right hand side (3200 Pa near the room opening) to
the left hand side (3900 Pa near the faulty cell). The pressure is then within 3550 Pa 10 %.
6.3.2 Pressure withstand for different wall types
The pressure withstand capability of different types of wall
10
construction has been analysed and proposed by number of
11
authors. This section provides a overview of 5 papers that
12
have covered this subject:
13
In [Pigler1976], a table of permitted pressure values
14
in housings dependent on the wall material is given.
15
This table is based on experience of a 24 cm thick
16
wall, with carefully executed joints between the wall
17
and adjoining members at the side and the top. This
18
is only general information, more exact information
19
has to be specified by the wall dimensions and the
Source:
TV Rheinland /
Berlin-Brandenburg
20
Figure 6-14: View of a building destroyed

joints between the wall and adjoining members. For after internal arc fault in the switchgear room.
21
example a wall with a larger surface area is less stiff
22
than a smaller wall. In the case of a larger surface area wall and a constant pressure there is also a higher
Page 87
force according to the larger surface area. That means, walls of varying surface areas but the same
material react in different ways. With this in mind Table 6-5 can be only a first indication for the expected
wall stiffness.
Wall type
Permitted overpressure in the room
bricks (solid bricks, perforated bricks,

gas formed concrete)
reinforced brick wall
30 mbar
precast concrete members
50 mbar
cast in place concrete
> 70 mbar
concrete space cell
130 mbar
Table 6-5: Source of table: 'EWT 1976, journal 3.
4
5
3-10 mbar
In [Primus1999], the standard values for permitted pressures in switchgear rooms depending on the
construction can be found (Table 6-6). The permitted pressure values from 3 to 10 mbar are given here for
a brick wall without other specifications of the brick wall.

Wall type
Brick wall (solid brick, perforated
brick, gas concrete)
3 up to 10 mbar
Armoured (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].
8
9
Permitted overpressure in the room
In contrast Table 6-7 from [Hollmann1999] give permitted pressure values for brickwork (masonry) taking
10
account of the wall thickness. A permitted overpressure of 10 mbar is given here for non- reinforced brick
11
walls with 24 cm thickness.

Wall material / Thickness
Allowed overpressure
Brick wall 24 cm (solid brick, perforated brick, gas

concrete)
approx. 10 mbar
Armoured (reinforced) brick wall 24 cm, Lime

sand stone wall (Solid brick)
approx. 25 mbar
Table 6-7: Pressure Strength of walls [Hollmann1999].
12
13
14
The withstand values of flat walls presented by D. Graf in [Graf1987] are of interest. He presented values
15
for sand-lime brick with different grout and provided explode (burst) and cracking values (Table 6-8). The
16
bursting pressure for sand-lime bricks is about 26 mbar with pulse stress, cracking at approx. 25 mbar.
Page 88
Wall Material / Thickness
Pressure stress
Gas concrete
12 cm thickness
exploded at approx. 15 mbar
Sand-lime brick
1/2 perforated brick, cement

grout
big clefts at approx. 25 mbar

exploded at approx. 60 mbar
Sand-lime brick
1/2 solid brick, lime grout
Exploded by a pressure pulse of 65

ms at 26 mbar
Table 6-8: Pressure Strength of walls [Graf1987].
1
2
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.
In [Pusch2007] the acceptable pressure stress values for walls are provided (Table 6-9).
Wall material/thickness
Permitted overpressure values
Brick wall, 24 cm
(solid brick, perforated brick, gas concrete)
up to 10 mbar
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].
5
6
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
high power density (typically above 46 MW/m ). Applying the basic method, one has to cope with the following
potential difficulties:
10
Complex Geometry: Between the arc compartment and the installation room there can be several
11
intermediate compartments and often one of them has more than one input or exhaust area (a busbar for
12
instance). In this case, it is not easy to choose the correct geometric model of the basic method provided in
13
Chapter 2.
14
15
Unreasonably high temperatures could result from the basic calculation method, when solving the
equations under the following circumstances:
16
17
when the pressure rise in the room itself lasts for a long time (sometimes up to one second) during
which the gas temperature in the arc compartment might rise to unrealistically high values.
18
19
when the pressure in the arc compartment is close to the external pressure so that the outflow
function becomes very sensitive to small variations of the gas density.
20
In order to apply the basic method to a room pressure calculation, while being conservative in overestimating the
21
pressure peak, the following approaches can be followed:
22
23
1.
The volume of the switchgear is ignored and only a single volume is considered, which is the room volume.
This approach may be used for AIS (see [Douchin2013]).
Page 89
2.
An enhanced model, where the -factor -decreases with the decreasing gas density in the arc
compartment. This is a more realistic case, which considers the encapsulation effect of the switchgear as
published in [Schumacher et al.]. An example is given in Section 0.
Approach 1 for Medium Voltage AIS:
The average pressure in the room can be calculated applying a single-volume model, with the following
parameters.
If and are known by the manufacturer, these values can be used. However, if the manufacturer
room, not the one in the switchgear compartment. It is not necessarily the same value. Furthermore,
10
material evaporation is not modeled in the basic method. The use of a lower than 0.3 could then lead to
specifies a < 0.3, then 0.3 shall be used instead. Indeed, the here is linked to the pressure rise in the
11
12
under-estimation of the room pressure.
13
14
15
16
If and are not known, typical values may be considered. For , the typical range is [0,4; 0,6] for air.
For , typical range is [300 V; 600 V] (refer to Section 2.3.5 and 2.3.6 for further details).
The room volume is calculated subtracting the switchgear volume. The discharge coefficient of the room
exhaust area shall be in the range of 0.6 to 1.
17
Note: The pressure relief device is modeled as a fully open area. If a grid or mesh exists, the relief area
may be multiplied by an efficiency factor less than 1.
18
6.3.4 Pressure results derived from CFD simulation
19
There are cases where the basic calculation method is not relevant any more: when the power density P/S is too
20
small, pressure does not become uniform within the room. Furthermore, local pressure peaks could not be
21
evaluated with the basic model, nor the effect of multiple pressure relief openings and their precise locations. CFD
22
is the right tool to overcome these issues.
23
The main idea is to model the switchgear room with all relevant flow resistances and volumes in a three
24
dimensional model. For correct sizing and placement of the room pressure relief openings, the basic model may
25
lead to an incorrect pressure because:
26
Flow blockages in the room are not considered.
27
The location of pressure relief openings and the geometry of the room are not investigated.
28
Dynamic processes like reflection, diffraction and interference (wave theory) are neglected.
29
But these conditions are often the reason for pressure on the walls in such highly dynamic processes. Often the
30
maximum pressure builds up in the corners of the room, because pressure waves are reflected and generate a
31
local maximum similar to standing waves in a tube.
32
This is illustrated in the following example. In Figure 6-15 the pressure measurement locations are marked with red
33
dots and the pressure relief openings are marked green. The orange panel with the yellow arrow is the panel where
34
the arc fault is assumed to take place. In this example there is a transformer room with big openings in the doors
Page 90
behind the switchgear. The gas flows out of the switchgear, passes the first small green openings and will leave the
building through the transformer doors.
By integrating the velocity over the opening area at the time of its maximum value 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 from their spectrum of products.
Figure 6-16 shows the pressure over time at the point MP3 of the right hand ceiling corner in the room. This
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.
Figure 6-15: Model of a switchgear room with

pressure measurement points (MP).
Figure 6-16: Pressure evolution over time on measuring

point MP3.
10
Page 91
2
3
Figure 6-17: Contour plot (left) and pressure function (right) exerted on the wall.
To get the value of the total force on the wall at a specific time it is necessary to integrate the local pressure values
over the wall or ceiling. The pressure maximum at the ceiling, as visible in Figure 6-17, is 16 mbar at time 20 ms
(left picture as contour plot, right picture as red pressure function).
The average pressure is given by the equation:
() 15300 N
=
= 1.5 mbar
104 m2
(6-9)
Finally one has only 10 % average pressure. This approach is helpful for engineers engaged in static calculations
10
to evaluate the stiffness of the switchgear room walls. It highlights the fact that, most often, a local high
11
overpressure (limited in space) does not affect the sizing of the entire wall.
12
13
14
Page 92
6.3.5 Pressure on building walls: comparison with test results (GIS case)
6.3.5.1 TESTED INSTALLATION AND RECORDED ARC POWER

Room
Arcing compartment
Room
Opening
Arc location
B
3
Valve
4
Cylinder (modelled with a

square section in CFD)
Figure 6-18: Test arrangement.
5
6
7
Pressure sensors A, B, C are located as per Figure 6-18:
Sensor A: in the cylinder, near the valve (arcing compartment)
Sensor B: in line with cylinder axis, on the room front wall (the one of the opening)
10
Sensor C: on the room side wall, 35 cm from the rear wall
11
12
Dimensions:
3
Arcing compartment
0.3 m
Cylinder volume and length
0.039 m , 0.54 m
Room volume
8 m (2 m x 2 m x 2 m)
Room opening
0.3 m
Valve area
0.049 m
13
14
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
15
voltage of 800 V, the power curve fits the theoretical one for the first two cycles. Afterwards, the arc voltage
Page 93
decreases to a lower value; however, the pressure peaks are recorded during the first two cycles, so the value of
800 V is relevant.
From the pressure sensor A curve, it can be deduced that the valve opens at 203 kPa.
eq(2-20) with Uarc = 800V
Arc power
measurement (eq (2-18))

1.0E+08
(W)
8.0E+07
6.0E+07
4.0E+07
2.0E+07
0.0E+00
0
0.02
0.04
0.06
0.08
0.1
Time (s)
4
5
Figure 6-19: Arc power, measurement versus formula.
6
7
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.
Despite being the same volume, pressures at location B and C are very different. CFD can handle this, and shows
10
that pressure, temperature and velocity fields are greatly influenced by the geometry of the installation.
11
It is also observed from CFD results that the pressure is not uniform within the room until 45 ms, which is the
12
instant of the peak. At this instant, pressures in the room are within the average value 20 %, and stay in this
13
range afterward (see Figure 6-23).
14
15
16
17
Page 94
Sensor A
250
200
50
Measurement
40
CFD
Pressure (kPa)
Pressure (kPa)
Measurement
150
CFD
30
100
20
50
10
0
0
-50
Sensor B
60
0.02
0.04
0.06
0 -10 0
0.08
Time (s)
0.04
0.06
0.08
Time (s)
-20
Sensor C
40
0.02
Temperature field (K) 21 ms
Measur
ement
35
30
Pressure (kPa)
25
20
15
10
5
0
-5 0
-10
0.02
0.04
0.06
0.08
0.1
Time (s)
Pressure field (Pa) 25 ms
Velocity field (m/s) 25 ms
Figure 6-20: Room simulation; CFD results.
Page 95
6.3.5.3 BASIC METHOD RESULTS
V2
V1
Welded cubic steel container
Arc ignited between three electrodes
Circular relief device
Exhaust into a 8 m container
Side view of container.

2
Figure 6-21: Basic method model.
Figure 6-22 shows that, for the arcing compartment (sensor A), a correct fit is achieved with = 0.22 and a
discharge coefficient () = 1.
Sensor A
2.00
Simplified method with kp=0,22
Pressure (bar relative)
2.50
measurement
1.50
1.00
0.50
0.00
0
0.02
0.04
0.06
Time (s)
0.08
0.1
Figure 6-22: Basic method, best fit for the arc compartment with = 0.22.
6
7
For pressure calculation in the room, there are two measurement points, B and C, which do not match each other.
With CFD results being correct for B and C, the average pressure within the room calculated with CFD is used as
the reference.
10
The input data of the basic method is the following:
11
Valve burst pressure:
203 kPa
12
Arc voltage:
800 V (see 6.3.5.1)
13
0.22
14
-factor:
Discharge coefficient () of the arcing compartment valve:
15
Discharge coefficient () of the room opening:
0.7
16
The result is shown in Figure 6-23. The basic method underestimates the pressure peak: it is 10 kPa, 45 % below
17
the reference value (18.3 kPa). In order to get a correct result, the -factor should be tuned from 0.22 to 0.34.
Page 96
50
Sensor B (CFD result)

Sensor C (CFD result)
room average (CFD result)
Simplified method with alfa = 0,7
Pressure (kPa)
40
30
20
10
0
0
0.01
0.02
0.03
0.04
0.06
0.07
0.08
0.09
0.1
Time(s)
-10
0.05
Figure 6-23: Basic calculation method compared with CFD.
2
3
4
5
6.3.5.4 ENHANCED METHOD WITH PRESSURE-DEPENDENT -FACTOR
Here, the following input data have been used:
Valve bursting pressure:
160 kPa
Arc voltage:
600 V
0.5
-factor:
Discharge coefficient () of the arcing compartment valve:
10
Discharge coefficient () of the room opening:
0.7
11
The results are given in Figure 6-24 and Figure 6-25 [Dullni2013].
12
13
It can be seen that the assumption made about the valve bursting pressure leads to a different value. This
14
With these settings, the enhanced method gives a good estimation of the room average pressure (sensor C);
15
however, it cannot reflect the local pressure peak recorded on sensor B.
illustrates the importance of this parameter for GIS cases.
16
Page 97
1
2
Figure 6-24: Enhanced method: Results sensor A (red) and B (blue).
4
5
Figure 6-25: Enhanced method: Results sensor C (blue).
6
7
6.3.5.5 CONCLUSIONS FOR THIS GIS CASE
Page 98
After the valve opening, the pressure field is not uniform during the first 40 ms, and pressure peaks occur locally
(see sensor B for example, and Figure 6-20). However, after 45 ms, pressure becomes homogenous, with a
distribution within 20 % around the room average. 45 ms is the instant when the room average curve reaches its
peak value.
For the different methods, one can conclude that:
CFD: the pressure development within the switchgear room is properly modelled
Basic and enhanced method: before 45 ms, the uniform pressure calculated is not relevant, and local
pressure peaks (sensor B) are not calculated. After 45 ms, the simplified method underestimates the space
averaged pressure, whereas the enhanced method gives a correct peak value.
10
In addition:
11
For all methods, results are very senitive to the bursting pressure of the tank valve.
12
13
For basic and enhanced methods, results are very sensitive to and discharge coefficients values.
In the end, one can not propose general rules for use of the simplified method for GIS rooms.
14
6.3.6 Pressure on building walls: comparison with test results: AIS case
15
Two Air-Insulated Switchgear rooms, respectively of 16 m (Figure 6-26 and Figure 6-27) and 40 m (Figure 6-30
16
and Figure 6-31), have been tested. The arc current was 20 kA. In this section, pressure measurements are
17
shown, and are compared with CFD calculation results.
18
The switchgear was equipped with an arc energy absorber. Such a device cannot be modeled with the basic
19
method, therefore only CFD results are shown.
20
6.3.6.1 16 m 3 ROOM
Room opening
21
22
Figure 6-26: 16 m3 AIS room installation and pressure sensor locations.
Page 99
1
2
Figure 6-27: 16 m3 AIS room: main data.
Note: The absorber is installed in the switchgear pressure relief opening (0.1 m ). Two models of the absorber are
calculated.
Room measurements are filtered with a low-pass 200 Hz filter (Yi corresponding to Pi). The three filtered
measurements within the room are shown in Figure 6-28:
7
8
Figure 6-28: Pressure measurement within the 16 m3 AIS room.
The measurements are within an average 20 % at the peak instant. In general, P4 is much lower than the two
10
others. CFD results are presented in Figure 6-29.
Page 100
Figure 6-29: CFD results for the 16 m3 AIS room.
Within the room (P2 to P4), CFD overestimates the pressure, which is conservative.
Page 101
1
2
6.3.6.2 40 m 3 ROOM
Room opening
3
4
Figure 6-30: 40 m3 AIS room installation and pressure sensor locations.
5
6
Figure 6-31: 40 m3 AIS room: main data.

2
Note: the absorber installed in the switchgear pressure relief opening (0.1 m ).
Room measurements are filtered with a low-pass 200 Hz filter (Yi corresponds to Pi). The three filtered
measurements within the room are shown in Figure 6-32.
Page 102
1
2
Figure 6-32: Pressure measurements within the 40 m3 AIS room.
The measurements 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 6-33. Within the room (P2 to P4), CFD overestimates the pressure, which is
conservative, and stay close to the measurements.
6.3.6.3 CONCLUSIONS FOR THE AIS CASES
These cases show again that the pressure field within the room can be far from uniform until a given instant. CFD
results are close to measured values and are conservative.
10
11
12
13
14
15
Page 103
Figure 6-33: CFD results for the 40 m3 AIS room.
1
2
3
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.
Room openings most often do not influence local pressure peaks. However, these peaks are limited in space and
time, and probably may not have much influence on the civil design. On the other hand, when pressure becomes
uniform in space (i.e. high power density P/S), the peak value of this uniform pressure is largely controlled by the
size of the room openings.
10
Basically: the larger this area, the smaller the volume average pressure inside the room. By appropriate
11
dimensioning of such relief areas, the building stress generated by an internal arc can then be minimised.
12
When increasing the size of the room opening, the room uniform pressure decreases as illustrated in Figure 6-34.
Page 104
per unit of the closed room pressure peak
1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
Room opening area
1
2
Figure 6-34: Room pressure function of room opening area.
The room pressure relief openings should be chosen in order to be in the knee of the curve, which is most
effective.
If the room is closed (without any opening), the pressure inside the room will increase continuously during the arc
fault and reach a high value. Therefore, such installation rooms should not be used unless the rooms 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.
6.3.8 Summary
10
Pressure on building walls as a result of an internal arc is a critical load for building design. It is a dynamic load,
11
spatially defined, however, CFD can calculate the whole figure of it with relative fidelity in contrast to the basic
12
model. An enhanced method can be used to calculate the room average pressure, ignoring local pressure peaks at
13
the beginning of the event. For AIS, this can also be achieved using the basic method, with particular rules.
14
The bursting pressure of the GIS valve is of major importance for all methods; in addition, basic and enhanced
15
methods are very sensitive to and discharge coefficients.
16
One can get a first idea of wall withstand capability using tables; but local building codes must be considered for
17
structural design.
18
6.4 Burn-through
19
In the previous sections the effect of the pressure on the switch enclosure and building walls has been seen. In this
20
section the third effect of the internal arc is looked at - the burn-through effect. This effect is caused by the arc
21
which can burn on a surface of the metallic enclosure (like a switchgear wall or panel, or GIS bus duct). This melts
22
and then punctures the wall.
Page 105
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
10
phase and ground. This grounded surface might be exposed to the burn-through. This section focuses only on the
11
burn-through in the single-phase enclosed constructions.
12
The arc will produce a pressure buildup within the faulty volume and an erosion process will take place at the arc
13
root location. In relation to the enclosure, a portion of the total arc power transfers to the enclosure wall and heats
14
the wall material so that it changes from solid to liquid and finally to gas (Figure 6-35) which could finally produce
15
the burn-through if the arc stays at the same place for a sufficient time (Figure 6-36).
16
The effects cannot be fully predicted by simple thermal conduction models since too many parameters play an
17
important role. It is also difficult to predict the movement of the arc and to determine the energy input at the arc
18
root.
EVAPORATION HEAT
MELTING HEAT
Q/m
19
20
Figure 6-35: Range of heat energy per mass (Q/m) fed into electrode material (schematically).
21
The time to burn-through depends mainly on the current density, the thickness of the enclosure wall, the type of
22
material and the duration of the fault. The erosion combined with pressure stress can finally cause the burn-through
23
of the enclosure wall.
24
In the case of burn-through the safety of operators or persons near the switchgear cannot be ensured since hot
25
gases and metal droplets are blown out of the hole in an uncontrolled manner. This is an unlikely event due to the
26
combination of the low rate of burn-through occurrence and the minimum presence of personnel near the GIS.
Page 106
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.
Figure 6-36 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
blown out 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.
8
9
Figure 6-36: Typical views of a HV GIS enclosure in the case of burn-through.
10
6.4.2 Evaluation of the burn-through time
11
Available studies to evaluate the burn-through time are listed in Table 6-10.
12
Rod to plane
Arrangement
Real GIS arrangement
Analytical
[Kuwahara1982]
[Kolbe1975]
[Babusci1998]
[Kuwahara1982]
Experimental
[Kolbe1975]
[Petterson1977]
[Lutz1983]
[Bernard1982]
[Trinh1992]
Table 6-10: Some literature references dealing with the burn-through phenomena.
13
Page 107
1
2
Figure 6-1: Movement of an arc in axial and azimuth direction [Babusci 1998].
Once the arc is established, unbalanced magnetic forces drive the arc in both the axial and azimuthal directions .
The axial motion of the arc along the conductor stops once the arc reaches a barrier (such as an 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 Krger [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.
10
The total time from the instant of the arc ignition to the enclosure burn-through could be divided into two parts:
= +
(6-10)
12
Where is the total time from the instant of the arc ignition to the enclosure burn-through, is the time during
13
and may cause a burn-through.
14
could be analytically determined from a model of the axial arc velocity assuming columnar arc geometry.
11
15
16
which the arc is traveling axially along the conductor, and is the time during which the arc stays axially stationary
According to Lutz and Chu in [Lutz1983] this leads to erroneous results. can be better evaluated by experimental
formulas given as following:
17
18
19
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.
20
Table 6-11 shows the comparison of arc velocities between calculations with equation (6-11) above and two test
21
cases of real GIS arrangement carried out by a manufacturer. SF6 gas pressure, current, and electrode spacing
22
were 6 bar, 40 kArms, and 122.5 mm, respectively. As seen in the table, measured arc velocities are faster than
23
those calculated from equation (6-11), showing 20 % variation.

Test shot
Velocity of arcs [m/s]
Chu [1982] mentions that a stationary arc can burn in HV GIS if the fault is supplied from two sides (two sources).
Page 108
Calculated with
equation (6-11)
Test result
#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.
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].
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
In the analytical approach, the models consider a small area of material heated up by the arc energy and the
10
of studying the influence of parameters such as current, enclosure thickness, and material.
11
The experimental approach to determining is made using rod-to-plane tests and coaxial arrangement tests:
12
6.4.2.1 ROD TO PLANE TESTS
13
Lutz and Chu [Lutz1983] proposed the use of the following formula based on the Kolbe proposal [Kolbe1975] for
14
aluminum plate:
15
on their rod to plane tests:
16
(, ) =
282
39.1
0.864
2.1
(6-12)
where is the current in kA rms, the plate thickness in mm and the time in ms. Authors found good agreement
17
between measured times and calculated times for their modification of Kolbes formula.
18
Pettersson proposed another formula for determining the burn-through time for an aluminum plate:
(, ) =
540 0.97
(6-13)
19
Figure 6-37 compares the results obtained with the two equations, for two different thicknesses. One can see that
20
for lower currents the differences between the two equations are larger but that they become smaller at higher
21
currents.
22
In the case of a GIS compartment with a coaxial arrangement, one can observe greater burn-through times than
23
those for rod to plane geometry because the azimuthal motion of the arc produces a less intensive vaporization of
24
the enclosure wall. For aluminum, the burn-through time could be between 1 and 2.5 times the burn-through time
Page 109
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.
Estimated perforation time [ms]
500
Lutz (thickness 8mm)

Pettersson (thickness 8mm)
Lutz (thickness 12 mm)
Pettersson (thickness 12 mm)
400
tLutz ( Is , h8)
tpettersson( Is , h8) 300
tLutz ( Is , h12)
tpettersson( Is , h12) 200
100
49.686
0
20
30
40
15
Is
60
65
Current [kA]
3
4
5
50
Figure 6-37: 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.
10
Based on an experimental approach, G. Bernard proposed the following equation for a coaxial assembly made of
11
aluminum enclosures [Bernard1982]:
12
13
14
15
(, ) =
87.4 1.77
0.67
(6-14)
Where is the r.m.s current in kA, 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.
Page 110
h[mm]
I [kA]
15
32
40
50
63
10
12
14
20
34
22
-3
49
x
9
x
x
x
x
x
x
x
40
x
x
x
x
x
16
-6
x
x
x
x
x
x
x
-3
x
x
x
16
6
x
x
x
x
-18
Table 6-12: Differences between burn-through times calculated with Bernards formula and results from
manufacturers experiments [%].
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
smallest 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 6-38.
For 63 kA tests and a 14 mm enclosure, one can see an important difference between the test results and the
10
Bernard formula where the Bernard formula predicts a much longer time (-18 %= -104 ms). This could lead to an
11
overestimate of the actual performance of the enclosure. The reasons for this discrepancy are not clear. One
12
possible explanation could be that the pressure inside the compartment was so high as to have blown a hole in the
13
enclosure, already weakened by the arc erosion (see the right hand picture of Figure 6-36). Bernards formula for
14
the experimental approach and the analytical model presented in publication [Babusci1998] are only valid if the
15
predominant factor is the erosion of the arc and are not valid if the remaining thickness cannot withstand the
16
pressure inside the enclosure.
17
18
19
Figure 6-38: 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.
Page 111
6.4.3 Aluminium 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.
Density
Steel
7.85 g/cm
Aluminum
2.70 g/cm
Melting Temperature
Heat capacity per unit

volume (density x
specific heat)
1400 C
3.53 J/(cm K)
660 C
2.43 J/(cm K)
Table 6-13: Properties of steel and aluminum.
5
6
Material
Using the equation biven by Babusci et al [Babusci1998]
(6-15)
where is the time to burn-through, 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
10
11
To summarize:
12
The generic formula for burn-through can be written as:
13
14
15
16
17
18
19
(6-16)
(6-17)
where is time to enclosure burn-through, 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 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.
Material
Authors/Reference
Arc
condition
Page 112
[Babusci1998]
Aluminum
Moving,
coaxial
arrangement
173
[Bernard1982]
Aluminum
Moving,
coaxial
arrangement
87.4
1.77
0.67
Diessner [Chu1980]
Aluminum
Moving,
coaxial
arrangement
150
[Trinh1989]
Aluminum
Moving,
coaxial
arrangement
179
0.73
[Petterson1977]
Aluminum
Stagnant,
Rod to plane
540
[Babusci1998]
Steel
Moving,
coaxial
arrangement
750
Table 6-14: Comparing properties of steel vs. aluminum.
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.
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 6-39.
From this qualitative approach, it can be seen that a large number of parameters indicate that shorter burn-through
10
times take place in SF6 rather than in air. It should be noted that this is a theoretical hypothesis (for a more detailed
11
understanding of these influencing factors, refer to ANNEX G:); there is not enough test data or experiment
12
available today to validate this claim.
13
Page 113
Influencing factors
(1) Electrode fall voltage

Input
power
Power
loss
Arc
motion
Quantitative comparison of the Qualitative comparison of resulting burninfluencing

through times, , for SF6 and air
factors for SF6 and air
SF6
Air
(SF6) (Air)
(2) Heat of chemical reaction

(exothermic energy)
SF6 > Air
(3) Radius of arc root

(concentration of input power)
SF6 Air
(4) Heat transfer from arc to gas

(conduction)**
SF6 > Air
(5) Heat transfer from arc to gas

(convection)**
SF6 > Air
(6) Heat transfer within metal

wall(s)
SF6 = Air
(7) Heat transfer from enclosure

to gas (conduction)**
SF6 > Air
(8) Heat transfer from enclosure

to gas (convection)**
SF6 > Air
(SF6) > (Air)**
(9) Heat consumption by metal

evaporation
SF6 = Air
tb (SF6) = tb (Air)
(10) Axial velocity - Contribution

ta *
SF6 < Air
tb (SF6) > tb (Air) *
(11) Azimuthal velocity
SF6 < Air
tb (SF6) < tb (Air)
(12) Drag forces at arc foot
SF6 Air
tb (SF6) tb (Air)
(SF6) < (Air)

(SF6) (Air)
(SF6) > (Air)**

(SF6) > (Air)**
(SF6) = (Air)
tb (SF6) > tb (Air)**
(13) Pressure exerted on the

SF6 > Air
tb (SF6) < tb (Air)
tank wall after disk opening
(14) Pressure exerted on the
SF6 < Air
tb (SF6) > tb (Air) ***
tank wall ***
(peak pressure)
*: IEC standard 62271-203 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).
Other
Table 6-15: Factors influencing the burn-trough comparatively for SF6 and air.
2
3
4
5
6
Page 114
Figure 6-39: Phenomena taking place at the arc foot root.
2
3
6.5 Summary
In this section three effects of the internal arc on structures have been analysed:
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 (von Mises stresses) of the switchgear enclosure can be reasonably
10
accurately calculated with off-the-shelf FEA software. With the basic model, the average pressure in the room can
11
be determined (with some limitations that need to be taken into account), and that value can be used to dimension
12
the room walls in Air-Insulated Substations. But for Gas Insulated Substations, the simplified method may lead to
13
underestimation of the pressure. With CFD calculations the localised time-dependent pressure is obtained, which
14
can be used to determine the size and placement of pressure relief openings in the building.
15
In addition it has been shown how the burn-through time can be evaluated, using different empirical formulae.
16
Furthermore, test results and calculations are compared. It should be noted that a large scatter in the test results is
17
possible since the energy released in one test may be different in another test depending on the erratic arc
18
movement for example. This scatter should be considered in order to allow a sufficient safety margin in relation to
19
the calculation results.
20
Lastly these tools can be used to assess the different pressure rise and mechanical stress on switch enclosures
21
when SF6 gas is replaced with air. With respect to the burn-through behaviour, the equivalence when replacing SF6
22
with air is not that obvious. From the qualitative approach developed in 6.4.4, the burn-through time in SF6 is likely
23
to be shorter than in air. However, there is not enough test data and experiments available today to validate this
24
statement.
25
Page 115
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 gnrale de
l'lectricit (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 254261,1992.
[Chabrerie1993]: Chabrerie J.P., Devautour J., Teste Ph.,
A Numerical Model for Thermal Processes in an
10
Electrode Submitted to an Arc in Air and Its Experimental Verification, IEEE Transactions on Components,
11
Hybrids, and Manufacturing Technology, Vol 16, No 4, 1993.
12
[Chu1980]: Chu, F.Y., Law, C.K., Boggs, S.A. "Dynamics of Power Arcs in CO-Axial Electrode Geometry",
13
GaseousDielectrics 11, March 1980.
14
[Chu1982]: Chu, F.; Ford, G. and Law, C. Estimation of Burn-Through Probability in SF6 Insulated Substations,
15
Power Apparatus and Systems, IEEE Transactions, PAS-101, 1391-1399, 1982.
16
[Chu1985]: Chu, F.Y and Lutz, F. and Braun, J.M and Stuckless, H.A, Effects of power arc faults in gas-insulated
17
substations, CIGRESymposium, Brussels, Contribution 340-04, 1985.
18
[Douchin2013]: Douchin J., Gentils F., Pressure rise in switchgear rooms in case of internal arc in AIS MV
19
switchboards; importance of room design and simplified calculation method, CIRED, 22
20
on Electricity Distribution, Stockholm, 2013.
21
[Dullni2013]: Dullni, E., Feeling the pressure, ABB Review 3/13, pp. 54-59, 2013.
22
[Graf1987]: D. Graf, Druckentwicklung beim Strlichtbogenfall in Schaltanlagen und in Schaltanlagengebuden,
23
Referat auf Informationstagung der Betonbau GmbH, 1987.
24
[Hollmann1999]: Hollmann and Driescher (Strlichtbogenbeanspruchung und -begrenzung in Mittelspannungs-
25
Schaltanlagen, Handbuch zur VDE-Seminarveranstaltung, VDE Bezirk Kurpfalz, 1999.
26
[Kuwahara1982]: Kuwahara, H and Yoshinaga, K. and Sakuma, S and Yamauchi, T. and Miyamoto, T.,
27
Fundamental investigation on internal arcs in SF6 gas filled enclosure, Mitsubishi Electric Corporation, 1982
28
[Kolbe1975]: Durchschmelzen von ebenen Metallplattenelektroden durch Hochstromlichtbgen in SF6, ETZ-A
29
Bd.96, 1975.
30
[Lutz1983]: Lutz, F. and Chu, F. Y., Burn-through of GIS Enclosure Due To Power Fault Arcs, EEE Transactions
31
on Power Apparatus and Systems, 1983.
32
[Petterson1977]: K.G. Pettersson, E.V. Granstroem, Design of Gas-Insulated Substations with Respect to Internal
33
Arcing, Report IEE, Publ. 157, 1977.
34
[Pigler1976]: F. Pigler, Overpressure in Switchrooms or Substations resulting from an internal arc in electrical
35
equipement, Energiewirtschaftliche Tagesfragen (EWT), journal 3, 1976.
Page 116
nd
International Conference
[Primus1999]: I.-F. Primus, Strlichtbogenfeste Gebudekonzepte geprfte Konstruktionsprinzipien Nachweise
durch Strlichtbogenprgungen, Handbuch zur VDE-Seminarveranstaltung, VDE Bezirk Kurpfalz, Mrz 1999.
[Pusch2007]: P. Pusch, Schaltberechtigung fr Elektrofachkrfte und befhigte Personen, 5. Aufl. 2007.
[Schumacher et al.]: M. Schumacher and G. Pietch and E. Dullni Zum Druckanstieg in Gebuden bei
Strlichtbogen in Innenraum-Schaltanlagen, Elektrizittswirtschaft, 1994.
[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.
10
Page 117
7 INTERNAL ARC SIMULATION REVIEW
7.1 Introduction
Current standards such as the IEC 62271-203 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 and 200 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
10
tests. However, they do warn that if tests are carried out with air instead of SF6, the pressure rise will be different
11
and care should be exercised in the interpretation of the test results. Some utilities require internal arc withstand
12
tests for all HV switchgear designs where SF6 is replaced by air considering such tests to be more severe.
13
These international standards leave the subject open to the agreement between manufacturer and user. However
14
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
15
accept any argument from the supplier or none at all. Therefore there is a need to provide some guidance in how
16
to interpret and validate a justification of the internal arc withstand capability of switchgear based on calculations.
17
A similar problem occured in the case of replacing short-circuit testing of Power Transformers by calculations. It is
18
very expensive to perform such a type test and it is accepted that the test may be replaced by a theoretical
19
evaluation of the ability of the transformer to withstand the dynamic effects of a short circuit. Annex A of IEC 60076-
20
5 provides guidelines to perform such an evaluation. Even if the dynamic effects of short-circuit currents in
21
transformers and internal arcs in switchgear are different physical phenomena, it is possible to simulate both
22
despite a statistical spread of test results. A similar structure in the theoretical evaluation can therefore be
23
established for the switchgear internal arc tests.
24
Similarly to the short-circuit testing of Power Transformers, the switchgear user might accept replacement of the
25
internal arc withstand test of the specific switchgear by a design analysis carried out by the manufacturer based on
26
tests of a similar design.
27
In this review, the calculated withstand capability of switchgear against the stresses caused by internal arc can be
28
evaluated by following two methods.
29
1. By comparison with reference switchgear which has already been tested (Section 7.3)
30
2. By check against the manufacturer design rules for internal arc withstand. These rules must be based on
31
32
previous test results. (Section 7.4)

Section 7.2 will first explain the information to be provided before the review takes place.
33
34
7.2 Information for the design review
35
Before performing the review, a complete list of the calculation assumptions, input data and results should be
36
provided. This information should be sufficient to be able to reproduce the calculation results, following similar
Page 118
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.

Electric
Geometric
Media
Mechanical
characteristics
Operating pressure of the
Fault current
Volume of arcing and exhaust
(rms,peak)
compartments if applicable
relief device
Type of gas (Air, SF6, Mixture)
Bursting pressure of the
Filling pressure
Rated voltage
Relief opening area
Circuit (3/1)
Exhaust compartment openings
Arc voltage
Phase to phase distance
Arc duration
Phase to ground distance
Type of fault
Inner diameter of GIS enclosure
-factor
arcing enclosure
Material of enclosure
Wall thickness of the
Ambient temperature
enclosure
Material of conductor
Manufacturing type of the

enclosure (casting, plate
welding, etc.)
Table 7-1: Design input for internal arc withstand simulation review.
7.3 Comparison of actual and reference switchgear using design

parameters
The results of a test made in similar switchgear can be used with certain margins to forecast the behaviour 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
10
under more onerous conditions and this other functional unit can be considered as similar to the tested one in the
11
following aspects:
12
structure and strength of the enclosure.
13
architecture of the partition (IEC 62271-200).
14
performance of the pressure relief device, if any.
15
insulation system.
16
physical influences (pressure rise, gas flow and and/or burn-through).
17
18
Table 7-2 and Table 7-3 can be used to determine the similar condition between two switchgears. Due to the
19
differences in pressure rise between typical MV metal-enclosed switchgear and internal arc withstand of HV GIS,
20
two different tables are shown. Table 7-2 provides a comparison criteria between internal arc withstand in two MV-
21
compartments within a family of MV (rated below 52 kV) switchgear. Table 7-3 shows the same criteria for a single
22
High Voltage compartment within a family of HV (Rated above 52 kV) gas insulated switchgear. In these tables the
23
validation criterion is based on the comparison of the design parameter with the reference switchgear. For
Page 119
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
8
9
There are cases where the switchgear has been tested, but the gas exhaust has a special design due to client
10
needs or request (gas ducts, pressure relief). As an example, consider that a particular switchgear passed the
11
test with a given gas exhaust system. Due to client requirements, the special gas exhaust conditions will be more
12
onerous than the one used in the test. A relieving condition is that the short-circuit current is lower than the one
13
used in the test. In these cases the comparison criteria which could be used are as follows:
14
First, a simulation of the tested arrangement shall be made. Results will be used as reference.
15
Then, the actual arrangement is calculated with the same parameters, changing only the geometry and the
16
17
current value.
-
18
19
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
20
the peak value of the structural stress, for the relevant part of the switchgear being different from
21
the reference switchgear. It shall be:
22
Lower for the actual arrangement than the tested one, or
23
Lower than rupture stress. This can be considered only in case the pressure given in the
24
reference simulation matches the tests measurement, within 10 % for example.
25
Page 120
Item
Design parameter
Validation
criterion
Condition
(1)
(2)
(3)
(4)
Phase to phase clearance
Phase to earth distance
Net enclosure/compartment volume
Rated pressure of insulating gas, if

applicable; see note 1
Cross-section of conductors
This concerns the region where the arc is

initiated.
Raw material of conductors (Al or Cu or

their alloys)
same

initiated.
Location of the point of arc initiation
same
Applying the rules of IEC 62271-200.
Insulating material exposed to the arc
same
same

initiated.
The position of the exhaust area in the

compartment has to be the same.
Exhaust area
10
Exhaust opening pressure
11
Strength of fixing elements of relief

device (flap)
12
Strength of the enclosure/ compartment
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
17
Arc duration
Larger areas are only acceptable if an exhaust

duct is used.
Applicable to fluid-tight compartments.
Applicable to non-tight compartments.
The relief device has the same design.
This also includes the strength of partitions and
bushings. Note 2 and 3.
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 121
item
Design parameter
Validation
criterion
Condition
(1)
(2)
(3)
(4)
Phase to phase clearance
Phase to earth distance
Enclosure/compartment volume
Pressure of insulating gas
Cross-section of conductors
Raw material of conductors (Al or Cu)
Same
Location of the point of arc initiation
Same
Distance from the arc initiation point to

the expected nearest obstacle
Insulating material exposed to the arc
Same
10
Exhaust/ventilation opening area
11
Exhaust opening pressure
12
Pressure relief design
13
Strength of the enclosure/

compartment
14
Thickness of the enclosure walls
15
Raw material of enclosure (Al or Steel)
16
Manufacturing type of enclosure

(casting, plate welding, etc.)
Applying the rules of IEC 62271-203
The position of the relief area in the

compartment has to be similar
Same
This also includes the strength of partitions
and bushings.
Note 1 and 2
Same
Same
17
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.
3
4
Page 122
7.4 Design Evaluation
In cases where no valid reference switchgear has been tested, the user may accept results of calculations which
satisfy the manufacturers 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 posses a solid experimental basis.
The rules should be derived from representative tests. The users should first validate the soundness of the
manufacturers 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
10
Evidence of internal arc failures in service: reports of actual internal failures on similar equipment if
11
available
12
The results of the calculations such as pressure rise, mechanical stresses in the enclosure and time to burn-
13
through will be compared with the limits of the design. In order to allow for the absence of test evidence, a
14
minimum safety margin of 20 % should be taken into account.
15
7.5 Summary
16
The switchgear user might accept the replacement of the internal arc withstand test of the specific switchgear by a
17
design analysis carried out by the manufacturer based on tests of a similar design. In order to be able to perform
18
this review the user should receive the necessary data to be able to reproduce at least an estimate of the analysis
19
presented by the manufacturer.
20
If the analysis is based on test results of similar switchgear, it is important to define the boundaries of this similarity
21
by comparison of the design parameters affecting the internal arc withstand capability of the switchgear. Table 7-2
22
and Table 7-3 provide an example of such comparison.
23
For High Voltage Switchgear rated over 52 kV tests are much less frequent and it might be very probable that no
24
test results of similar switchgear are available. In such cases the user can accept results of calculations which
25
satisfy the manufacturers design rules for internal arc withstand capability. These design rules have to be validated
26
by the previous design experience and test evidence provided by the manufacturer.
27
Page 123
8 CONCLUSION
2
3
The goal of this working group 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
10
d. To replace SF6 with air during internal arc testing
11
12
The authors agree that simulations cannot replace type tests, but they could be used for interpolation between the
13
known tests and make good predictions.
14
The working group reviewed existing software tools for calculating the effects of an internal arc fault, focusing on 3
15
main effects of an internal arc:
16
Pressure rise
17
Mechanical stress on enclosure and buildings
18
Burn-through
19
20
a) Providing methods for pressure calculation and allow benchmarking with performed tests
21
This review included various home made software tools, ranging from simple spreadsheets in Excel that most
22
engineers could use with a little effort, to a complex 3-D Computational Fluid Dynamic (CFD) software
23
package whose application remain limited to small number of experts due to the complexity and cost of the
24
software.
25
Three different models for measuring pressure rise are proposed, with each one having different levels of
26
complexity and its own limitations, as shown in the table below.

Approach /model
Appropriate Application
Limitations
Assumes uniform gas parameters
Not applicable for large relief
To quickly calculate uniform pressure

1)
openings
Basic
rise inside an arc compartment and the
(low complexity)
exhaust volume in typical MV switchgear
gas temperature exceeds approx.
and HV GIS applications.
2000 K for SF6 and 6000 K for air.
Calculations are not reliable, when
Doesnt consider gas mixtures in the

2)
Enhanced
(medium
To calculate uniform pressure rise
Page 124
Assumes unifrom gas parameters
as under 1) adding further
complexity)
approximations to better match test
Limitations and applications depend

on the implemented approximations.
results and calculation.
3)
CFD
(High complexity)
High effort for the modeling and
For calculating spatial pressure
meshing of the rooms and
distribution and gas flow in odd shapes
switchgear
geometry and large rooms.
Requires large computing power

and time.
Table 8-1: Models for calculating pressure rise during internal arc fault
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
5
6
% for the arc compartment after adjusting the k p factor and coefficient . The comparison also indicates that
0.5 for air and 0.7 for SF6, the discharge coefficients 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 taken (not calculated) from the
most arrangements can be successfully simulated by applying common input parameters: the coefficient kp of
10
previous internal arc test on the similar switchgear design. Due to stochastic nature of the arc, accurate
11
calculation of the arc voltage in complex switchgear designs is still to be desired; hence recorded values should
12
be used for more accurate predictions of pressure rise and mechanical stress.
13
The basic model also helps to understand which parameters are contributing more to pressure rise then others.
14
For example, arc voltage has much more influence on maximum pressure then level of asymmetry of the fault
15
current.
16
Pressure rise calculations are less accurate for the exhaust compartment. Enhanced models can be used for
17
more accurate results in the exhaust compartment.
18
For large installation rooms and arc/exhaust compartments with complex geometry where pressure isnt
19
uniform, CFD calculation should be used.
20
21
22
b) Providing methods for calculating other effects of the internal arc.

Namely, three additional effects of the internal arc on structures have been analysed in this technical brochure:
23
Mechanical Stress on switchgear due to overpressure
24
Mechanical stress on building walls due to overpressure
25
Burn-through
26
Mechanical deformation (von Mises stresses) of the switchgear enclosure can be reasonably accurately
27
calculated with off-the-shelf FEA software once the pressure curve is known. Also, the walls in Air-Insulated
28
Substations can be dimensioned by calculating the pressure rise using basic model (with some limitations that
Page 125
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 localised 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 test on 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
10
to the calculation results.
11
12
c) Verifying design modification by simulation methods and reducting number of tests
13
In order to reduce number of internal arc tests, this technical brochure provides guideline for internal arc
14
simulation review.
15
The switchgear user might accept the replacement of the internal arc withstand test of the specific switchgear
16
by a design analysis carried out by the manufacturer based on tests of a similar design. In order to be able to
17
perform this review the user should receive the necessary data to be able to reproduce at least an estimate of
18
the analysis presented by the manufacturer.
19
If the analysis is based on test results of similar switchgear, it is important to define the boundaries of this
20
similarity by comparison of the design parameters affecting the internal arc withstand capability of the
21
switchgear.
22
For High Voltage Switchgear rated over 52 kV tests are much less frequent and it might be very probable that
23
no test results of similar switchgear are available. In such cases the user can accept results of calculations
24
which satisfy the manufacturers design rules for internal arc withstand capability. These design rules have to
25
be validated by the previous design experience and test evidence provided by the manufacturer.
26
27
d) Replacing SF6 with air during internal arc testing
28
29
After thorough examination of existing test data, the authors agreed that replacement of SF6 with air during
30
internal arc testing provides mixed results. Its been concluded that there is no silver bullet recommendation
31
and that each case must be evaluated separately. Some observations are listed below:
32
higher than in SF6. Burn-through might happen faster with SF6 then with air under the same conditions.
33
34
Arc compartment: Pressure development and resulting mechanical stresses in air are in most cases
Exhaust compartment: Pressure development and the resulting mechanical stresses in SF6 are in most
35
cases higher than in air. Burn-through is not applicable because there is no arc in the exhaust
36
compartment.
Page 126
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.
5
6
In conclusion, software tools for simulations of the internal arc in MV and HV switchgear can be used in number
applications, including:
8
9
Helping designer to design an arc-resistant switchgear.
10
Helping civil engineer to determine wall dimensions and size of the openings of the installation room.
11
Extending the validity of the type tests on the similar design and thus reduce nubmer of internal arc tests.
Note that Internal arc test must be done on the similar design to get the correct energy input data - k p . It is
12
13
14
15
imporatant to measure the pressure rise during the internal arc test.
Validating various installatio tions, including different exhaust channels/ducts and different size
rooms / pressure relief openings.
16
17
18
19
20
Page 127
ANNEX A:
A.1 Introduction
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 one can use this information and create 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
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 an 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
10
EQUATIONS FOR PRESSURE RISE CALCULATION
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 high pressure

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

pressure region
mass flow through

heat source
Figure A-1: Control volume.

11
Figure A-1 presents an enclosed gas container called control volume with two openings, which allow gas flowing
12
into and out of the volume. The gas can be heated up by a heat source.
13
A.2.1 Definition of specific gas parameters
14
For an ideal gas the specific heat capacity at constant volume is:
15
16
17
(A-1)
where is specific heat capacity at constant volume, is specific internal energy, and is temperature.
And the specific heat capacity at constant pressure is:
where is specific heat capacity at constant pressure, is specific enthalpy, and is temperature.
Page 128
(A-2)
3
4
5
6
The adiabatic index, , is given by the heat capacity ratio:
The specific enthalpy is defined as:
= +
where is pressure and is specific volume.
The ideal gas law is:
13
(A-5)
A relationship between the specific gas constant and the specific heat constants is:
A.2.2 Ideal gas law
12
(A-4)
where is specific gas constant, is universal gas constant, and is molar mass.
11
(A-3)
The specific gas constant is given by the universal gas constant divided by the molar mass:
10
were is volume and is mass.
(A-6)
(A-7)
With equation (A-5) follows
(A-8)
(A-9)
= ( 1)
(A-10)
With equation (A-6) follows
This can be written with equation (A-3) as
14
This equation can be used to calculate the density based on the pressure, temperature, specific heat capacity and
15
adiabatic index:
16
were is the density.
=
( 1)
Page 129
(A-11)
A.2.3 Conservation of mass
The mass change in the volume is given by the difference of the in flowing and out flowing gas mass.
(A-12)
where is mass in the control volume, is mass flow into the volume, and is mass flow out of the
volume.
The mass inside the control volume (CV) may consist of different gas components. In the following two components
are considered but it can be extended to more components.
7
8
=
+
where
is the mass of type A in CV and
is the mass of type B in CV.
The mass fractions can be calculated as:
(A-13)
(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.
10
It is assumed that only gas component A is flowing into the control volume. Then the mass changes of the different
11
gas components in the control volume are described by:
12
13
(A-16)
(A-17)
The sum of gas component changes gives the total mass change:
=
+

, =
+

, =
+
1 =
,1
,1
Page 130
(A-18)
(A-19)
(A-20)
(A-21)
Where 1 is the adiabatic coefficient 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 species into account.
A.2.4 Mass flow
6
7
The following equations describe the mass flow through the opening . The mass flow through the opening
9
10
11
can be calculated with the same equations but adjusted indices. The mass flow is given by:
(A-22)
Where is the area of relief opening, is the gas density in the opening, is the velocity in the opening,
is the discharge coefficient of relief opening, and is the mass flow rate out of the control volume.
The gas density in the opening is given by:
1/
(A-23)
where is the gas density in the control volume, is the gas density in the opening, is the pressure in the
13
control volume, is the pressure in the opening, and is the adiabatic index of the gas in the control volume.
14
The gas velocity in the opening is:
12
The gas density in the control volume can be calculated with equation (A-11).
=
15
16
17
18
(A-24)
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 1
=
+1
(A-26)
19
A.2.5 Conservation of energy
20
The first law of thermodynamics for an open system says that the change of energy within the system boundaries is
21
equal to the difference of energy entering and leaving the system.
Page 131
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 specific internal energy, specific kinetic energy and specific potential energy times
the mass of the gas.
The right side of the equation is the sum of 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.
The left side of the equation is valid for a resting 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
+ 2 +
= + +
2
+ + +
10
The terms are as follows.
11
Left side: mass in the system, specific internal energy,
12
potential energy of the system.
2
2
(A-27)
specific kinetic energy of the system, specific

2
14
Right side: heat transfer, mechanical work,

expansion or compression work, mass flow, specific
15
flow.
16
The following terms can be simplified when applying equation (A-27) to the control volume.
13
17
18
19
kinetic energy of the mass flow,
= 0
specific enthalpy of the mass flow, and specific potential energy of the mass
potential energy is zero for all volumes and regions
=0
constant control volume
21
22
Equation (A-27) can then be written as:
20
23
24
25
no mechanical (technical) work
=0
one heat source
=0
inside the control volume the gas is at rest
( )
2
2
= + + +
2
2
(A-28)
where is the mass in control volume, is the specific internal energy of the gas in the control volume, is
the mass flow, and is the specific enthalpy.
The inflow and outflow is an adiabatic process. Therefore it is:
Page 132
= +
1
2
4
5
6
7
8
9
10
11
= +
(A-29)
(A-30)
where is the specific enthalpy of the gas in the high pressure region and is the specific enthalpy of the
gas in the control volume. Equation (A-28) can be simplified as:
( )
= +
(A-31)
Equation (A-31) can be used to calculate the temperature.
+
= +

(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:
+ ( )
=
specific enthalpy of the gas in the high pressure region
specific internal energy of the gas in the control volume
specific heat capacity at constant volume in CV
(A-33)
specific enthalpy of the gas in the control volume
mass in the control volume (CV)
heat source in CV
mass flow out of the CV
13
14
The pressure follows with the ideal gas law:
12
mass flow into the CV
( 1)
(A-34)
15
A.3 Basic model
16
17
For constant specific heat capacities at constant pressure (classical ideal gas, = constant) the specific enthalpy
18
19
can be expressed as:
(A-35)
For constant specific heat capacities at constant volume (classical ideal gas, = constant) the specific internal
energy can be expressed as:
Page 133
(A-36)
If the specific heat capacities are assumed to be constant, these equations can be used to write equation (A-33)
as:
3
4
5
6
7
8
9
+
=
gas temperature in the control volume (CV)
specific heat capacity at constant pressure in the high pressure region
specific heat capacity at constant volume in the high pressure area
gas temperature in the high pressure region
specific heat capacity at constant pressure in CV
specific heat capacity at constant volume in CV
10
The Euler method is used to solve the differential equations (A-12) and (A-37).
11
Step 1: Initial conditions
12
( = 0)
13
14
15
16
17
18
19
20
21
( = 0)
(A-37)
Gas pressure in control volume

Gas temperature in control volume
The volume , specific heat capacity and adiabatic index are assumed to be constant. The environment
has constant ambient pressure and temperature. The initial mass in the volume follows from equation (A-11):
where is the gas mass in CV.
( = 0) =
( = 0)
( 1) ( = 0)
(A-38)
Step 2: Changes of mass and temperature
where mass flow into the CV and mass flow out of the CV.
(A-39)
(A-40)
Step 3: Numerical integration
( + ) = () +
Page 134
(A-41)
1
2
is the mass change,
TCV
( + ) = () + TCV
(A-42)
The pressure follows from the ideal gas law:
pCV (t + t) =
mCV (t + t)cvCV { 1}TCV (t + t)

VCV
(A-43)
A.4 Enhancement of the basic model
The enhanced model uses the same set of equations as the basic model with the following modifications:
density dependent -factor (Section A.4.1)
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)
10
arc absorbers in the exhaust flow (Section A.4.6)
11
speed of relief opening device (Section A.4.7)
12
temperature dependent gas data (Section A.4.8)
exothermic reaction energy (Section A.4.2)
13
These modifications do not increase the accuracy in predicting the peak pressure in the arc compartment, but do
14
increase the accuracy in simulating the time dependence of the gas outflow and pressure development in the
15
16
A.4.1 Density dependent k p factor
17
An important enhancement of the basic model is the introduction of a density dependent -factor after opening of
19
the pressure relief device. The -factor determines pressure rise until rupture of the relief disc. In the basic model,
20
gas temperatures inside the arc compartment, because the same arc power heats up an ever decreasing gas
21
mass. Since gas temperatures above 10000 K have never been measured in free burning arcs and indeed are not
22
realistic, other effects have to be considered, which keep the gas temperature lower.
23
Tests with arc exhaust into a closed room [Dullni1994] have shown that the pressure rise in the exhaust
24
compartment is not as high as anticipated when assuming a constant factor over the whole calculation duration.
18
is assumed constant after relief opening even after a large exhaust of gas. This assumption leads to rather high
25
Since a cooling effect due to the walls of the room can be neglected, one can conclude that the energy transferred
26
from the internal arc to the surrounding gas is decreasing during the arc duration. This effect can be simulated by a
27
28
-factor diminishing with gas density in the arc compartment [Friberg1999]. It has to be noted that the pressure
29
Unless cool gas from outside enters the compartment through unintentional gaps in the enclosure or metal
30
evaporation from contacts occurs, the internal gas density will become extremely low.
inside the arc compartment finally drops to ambient pressure after exhaust of most of the contained gas mass.
Page 135
1
2
3
4
5
6
A good approximation for a density depending -factor was obtained in an experiment using SF6 as well as air as
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 on 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.
(t)
k p (t) = k p0 c0
0
7
8
9
10
for
k p (t) = k p0
for
(t) < c
(t) > c
(A-44)
(A-45)
where () is the gas density at time , 0 is the normal gas density at ambient pressure and temperature, 0 is the
-factor before transition, 0 adapted pre-factor, 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
12
the case in the publication [Dullni1994], but could also be adapted to different transition densities . The
13
exhaust compartment or installation room. It keeps the gas temperature lower at later simulation times and
14
removes numerical instabilities as a beneficial side effect. The course of pressure inside the arc compartment, in
15
particular up to the pressure peak, is hardly changed.
16
A.4.2 Exothermic reaction energy
17
The material of conductors and walls, where the arc has its roots, influences the value of the -factor. It is known
11
18
19
20
introduction of a density dependent -factor diminishes the exhaust of gas and hence reduces the pressure in the
that SF6 and to some extent also air reacts with aluminium under exothermic energy release. In this case, > 1
may give appropriate results [Zhang2002]. Another approach is to add the exothermic energy from evaporated
metal to the arc energy avoiding an artificially augmented -factor. From the reaction schemes, oxygen and SF6
21
release a very similar exothermic energy when reacting with aluminium [Bjrtuft2005]. However, since air contains
22
only a fraction of 20 % oxygen, the total amount of released exothermic energy is a factor of 5 smaller compared
23
with SF6.
24
A.4.3 Pressure dependent arc voltage
25
The voltage of an internal arc is seldom constant during a half cycle of current, and in addition varies in time over
26
arc duration. It was observed that arc voltage is higher during the high pressure phase in the arc compartment than
27
later after exhaust of gas. An empirical formula has been deduced for a single-phase arc burning between two
28
parallel copper electrodes in an aluminum container [Dullni1994]. The formula given below has been proposed
29
from arc initiation up to the highest pressures for air as well as for SF6 with an adapted initial value 0 of 500 V for
30
air and 950 V for SF6. After the exhaust of gas i.e. after reaching ambient pressure, the arc voltage varied between
31
400 and 600 V for air and between 400 and 800 V for SF6.
Page 136
() = 0 2 2kPa
(A-46)
where 0 is the adapted initial arc voltage and 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 to use the actual
course of measured arc power, i.e. product of momentary current and phase-to-earth voltage, as 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
10
SF6 is added to the mass of air in the compartment, it is assumed that the mixed gas behaves like pure air.
11
Because of the different adiabatic indexes of SF6 and air considering the term ( 1) in equation (2-16) - the
12
pressure rise in the exhaust compartment will be overestimated. In particular for small compartments, where the
13
fraction of SF6 becomes essential, the gas temperature might easily be overestimated by a factor of 2. For
14
improvement, the gas properties , and have to be calculated according to the fraction of gases on the total
15
mass. This requires separate gas equations for all species.
16
Mixing of different gases can be implemented in an enhanced model by applying the equations given in Section
17
A.2.3. In particular this is recommended for the mixing of SF6 gas flowing out of the arc compartment into the
18
exhaust compartment filled with air. Gas mixing is also important for the correct consideration of the effect of
19
evaporated material on the pressure rise in the arc compartment as described in Section A.4.5. Again, the
20
equations given in Section A.2.3 can be applied.
21
22
A.4.5 Metal evaporation and ablation of insulators
23
From tests, it is obvious that not only hot gas is ejected from switchgear, but also e.g. metal vapor and carbon
24
black. Because of the severe absorption of light by metal vapor and dust, the cloud coming out of the relief opening
25
often looks black or brownish, whereas the direct exhaust jet is bright. Evaporation of metal vapor from hot arc
26
roots and ablation of insulating material by the hot arc column can add a considerable amount of gas to the arc
27
compartment. This effect counteracts the ever increasing gas temperature in the simulation, when the gas density
28
inside the arc compartment becomes smaller and smaller due to the exhaust of gas mass. Metal evaporation
29
therefore is highly expected to play a role during the exhaust phase, when the pressure in the arc compartment
30
equals ambient pressure.
31
The evaporation effect can be introduced as an enhancement into the basic model by properly considering the part
32
of the arc energy deposited in the arc roots, . The arising mass of metal vapor is given by the inverse of the
Page 137
1
2
3
4
5
6
7
8
specific evaporation energy of the metal multiplied by the fraction of the arc energy. Alternatively measured
values of evaporated material as function of arc energy can be used (Table A-1).
mevap =
k evap Qel
Wevap
evaporated mass from metal electrodes
fraction of arc energy used for evaporation
specific evaporation energy of metal
electric arc energy
(A-47)
The amount of arc energy deposited in the arc roots is between 10 and 15 % [Zhang2002]. is smaller than
this fraction, since the arc energy is also consumed for heating and melting of the electrode material. Measured
10
material loss of electrodes as function of the integrated arc current provide similar values as the specific
11
evaporation energies, when a cathode and anode drop voltage of approximately 50 V is used. Table A-1
12
[Zhang2002] contains measured electrode mass loss for different materials.
13
14
Electrode loss in mg/As
Al
Cu
Fe
SF6
5.8
Air
5.8
2.9
2.5
Table A-1: Electrode mass loss [Zhang2002].

The mass of metal atoms, , can be simply added to the mass 1 of gas in the arc compartment.
m1 = mevap m12
(A-48)
16
Where 12 is the mass flow of gas out of the volume.
17
gas of the compartment in accordance with equation (A-13) in Section A.2.3. Then also the different mass and
18
specific heat capacities of metal atoms can be correctly considered.
19
The amount of ablated insulator material, , can be estimated from the inverse of the specific dissociation
15
For a more advanced approach, the evaporated metal atoms can be treated as separate gas adding to the filling
20
energy of the insulation material multiplied by some other fraction of the arc energy. The specific dissociation
21
energy can be determined from the total of all binding energies of the chain molecules of the insulation material
22
subtracting all binding energies of recombined gaseous molecules, or alternatively by measuring the released
23
volume of gaseous molecules [Hochhaus1985]. Most often epoxy with glass fillers is involved, which requires
24
subtracting the non-solvable amount of the filler material. Here, it is assumed that most of the ablated material is
25
released as CH4 gas. Here again, the mass of ablated material, , can be simply added to the mass of gas in
26
the arc compartment or implemented with more effort through separate mass equations.
Page 138
m1 = mevap + mabl m12

mabl =
1
2
3
4
5
6
k abl Qel
Wabl
1
= Vabl ins CH4 (1 iller )
Wabl
(A-49)
(A-50)
(A-51)
ablation mass from insulators
specific dissociation energy of insulation material
volume of dissociated solid material per energy
fraction of gas (CH4) on ablated material
fraction of arc energy spent on ablation of solid material
specific density of solid insulation material
It is anticipated that the fraction of arc energy used for ablation or evaporation does not depend on time i.e. is
10
constant for the whole duration of the arc. It is not a volume process depending on the filling gas density, but a
11
surface process. Also, it has to be considered that the evaporated or ablated gas has to get into thermodynamic
12
equilibrium with the surrounding gas in the compartment. Since the evaporated atoms leave the electrode surfaces
13
already with the evaporation temperature, this either constitutes a heat sink or a heat source depending on the
14
surrounding gas temperature being larger or smaller. This energy has to be considered in the temperature balance
15
of the gas mass in the arc compartment assuming instantaneous mixing. For energy exchange the heat capacities
16
of the gaseous materials at constant volume are relevant.
17
18
19
20
T1 =
fraction of filler material in insulator
Q1 mcp1 cv1 T1 + mabl cv,abl (Tabl T1 ) + mevap cv,evap Tevap T1

m1 cv1
specific heat capacity of gaseous ablated material
specific heat capacity of metal vapor
temperature of ablated material
(A-52)
specific heat capacity of the mixed gas
22
23
Calculations confirm that the peak pressure in the arc compartment is hardly changed by evaporated material
24
except in special cases, where exothermic reactions prevail. On the other hand, the course of the declining
25
pressure in the arc compartment and the pressure in the exhaust compartment may be indeed modified by the
21
evaporation temperature of electrode material
Page 139
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 the correct modelling, it is required to take the
composition and mixing of the gas 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
Evaporation energy of aluminium
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 aluminium (Al) at constant volume
470
J/kg K
Ablation temperature of epoxy
unit
Evaporation temperature of Cu
2870
Evaporation temperature of Al
2720
Table A-2: Input values for advanced calculation of pressure rise including material evaporation.
A.4.6 Arc absorbers in the exhaust flow
In some switchgear designs the exhaust of gas is cooled down by arc absorbers. These are composed e.g. of
several layers of fine-meshed grids or other obstacles either metallic or insulating in the path of the exhausted gas.
These devices absorb some fraction of the energy of the out flowing gas, however, may also reduce the effective
10
cross-section of the pressure relief area. In this case the gas remains longer in the compartment and might carry
11
more arc energy into the installation room than without arc absorber counteracting the cooling effect. The effect of
12
such arc absorbers can be implemented in the equations of the models by a reduced area of the relief opening and
13
a concurrent cooling factor reducing the energy transported from the arc compartment into the exhaust
14
compartment. The cooling effect per mesh is however limited to several percent of the energy flow and does not
15
linearly increase with the number of consecutive meshes [AiF2011].
16
A.4.7 Speed of relief opening device
17
Heavy relief flaps as used sometimes for air insulated switchgear, do not open instantaneously. The time
18
dependent increase of the opening area can be calculated from the acceleration of the plate having a certain mass
19
under the force determined by the scalar product of gas pressure and area of the relief opening. These flaps are
20
often hinged on one side so that the force on the plate is reduced in time during the opening. The simulation of its
21
motion together with the determination of the momentary opening area is most often not considered. In principle,
22
the speed of the plate can be calculated by integrating the acceleration of the plate over time. For the light burst
23
plates in SF6 insulated switchgear, this effect is of minor importance, since the full opening area is achieved almost
24
instantaneously.
Page 140
A.4.8 Temperature dependent gas data
The basic pressure calculation method is based on a gas model called 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 is
considered by the number of particles changing with temperature. Thus, the gas properties (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
10
more complex.
11
Some enhanced pressure calculation methods are based on real gas data. In this case the gas particles are no
12
longer regarded as concentrated points but have a certain volume so that interactions between particles exist like
13
dipole forces. The properties of real gases depend not only on their composition, but also on chemical reactions,
14
temperature and pressure. The generation of these data is time consuming. That is why they are collected once in
15
multidimensional tables and retrieved during the pressure calculation method.
16
The specific heat capacity of SF6 increases by more than a factor of 10 around a temperature of 2000 K caused by
17
collision induced dissociation of molecules. A similar anomaly is observed for N2 at a temperature of 6000 K. The
18
temperature and pressure dependence of the heat capacity and the corresponding adiabatic index can be
19
introduced into the equations by analytic approximations of the heat capacity at constant volume during the time
20
steps of calculation. This approach is permitted as long as the temperature in the volume is uniform allowing for a
21
thermodynamic equilibrium of all molecule fractions. This might no longer be true for rapidly flowing gas in the
22
exhaust.
23
The inclusion of temperature dependent heat capacities involve a lot of modifications of the equations, which are
24
not easy to implement into the basic model and therefore are not discussed in detail here.
25
A.5 Summary
26
A detailed description of the equations, which are used for the basic model, is given. The equations are based on
27
the ideal gas law, conservation of mass and energy. The reader can use these equations to develop his own
28
software. It is described how to calculate the initial conditions as well as temperature and pressure change for each
29
time step. The code may be used to recalculate the examples given in Chapter 2.
30
Further on equations are given to enhance the basic model. Enhancements consider density dependent -factor,
31
exothermic reaction energy, pressure dependent arc voltage, mixing of gas in compartments, metal evaporation
32
and ablation of insulators, arc absorbers in the exhaust flow, speed of relief opening device, and temperature
33
dependent gas data.
34
35
Page 141
REFERENCES:
[AiF2011]: Jan Christoph Kahlen, G. Pietsch: Reduzierung der Druckbeanspruchung elektrischer Anlagen im
Strlichtbogenfall, 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 Beitrge zur Hochspannungstechnik, Band
14, ISBN 3861306778, Dissertation, RWTH Aachen University, 2010.
[Bjrtuft2005]: T. Bjrtuft, 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
10
switchboard, 6th Int. Symposium on Short Circuit Currents in Power Systems, Liege, 1994.
11
[Friberg1999]: G. Friberg und G. Pietsch: Calculation of pressure rise due to arcing faults, IEEE Transactions on
12
Power Delivery, Vol. 14(2): S. 365-370, 1999.
13
[Hochhaus1985]: H. Hochhaus, Untersuchung der Wechselwirkungen zwischen Schaltlichtbgen und Isolierstoff-
14
wnden (Investigation of the interaction between switching arcs and insulating walls), PhD Thesis in German,
15
Technical University Braunschweig, 1985.
16
[Zhang2002]: Xiang Zhang: Modellierung der Auswirkungen von Strlichtbgen in elektrischen Anlagen, Ph.D
17
Dissertation, RWTH Aachen, 2002.
18
Page 142
ANNEX B:
PRESSURE SENSORS: TYPES AND ACCURACIES
B.1 Introduction
An important point in the evaluation of internal arc tests and for later comparison with simulations is a correct
measurement of the pressure in compartments. Here not only the type of pressure sensor is decisive, but also the
recording equipment, the post processing of the signal and the data format of storage. Also, the position of sensors
during the test has to be considered.
B.2 Pressure sensors
Most of the pressure sensors are based on the piezo-electric or piezo-resistive effect which generate a voltage or
modifies a resistance when a crystal or a semiconductor is mechanically deformed e.g. by the overpressure in a
10
switchgear compartment caused by an internal arc.
11
On the other hand, only a minor fraction of the market is taken up by capacitive-type sensors. The reasons for the
12
lack in breakthrough can be found in the design complexity and the requirements for a matched sensing circuit
13
[Puers].
14
Piezoelectric pressure sensors are primarily recommended for dynamic pressure measurements, however some
15
quartz pressure sensors have long discharge time constants (DTC) that extend low-frequency capability to permit
16
static calibration and measurement of quasi-static pressures over a period of a few seconds. DTC is defined as the
17
time required for a sensor or measuring system to discharge its signal to 37 % of the original value from a step
18
change of measure (see Figure B-1).
19
20
Figure B-1: Piezoelectric Signal Exponential Decay.
21
The dynamic pressure, originated by an internal arc, is translated into static pressure when it flows towards a wall
22
and stops [Bjrtuft]. Therefore, what is measured depends on the location of the sensor with respect to the gas flow
23
direction. Precision mounting of pressure sensors is essential for good pressure measurements. Always check the
24
installation drawings supplied in the manual with the sensor, or contact the pressure sensor supplier to request
25
detailed mounting instructions.
26
On the other hand, sensors have to be selected according to the expected pressure and frequency range. The
27
sensors need an auxiliary d.c. voltage supply in the range of 10 to 30 V. The output signal can be directly
28
connected to the input impedance of a transient recorder or via some resistive load. The temporal response should
Page 143
be in the millisecond range in order to provide sufficient time resolution. Following the NyquistShannon theorem,
the sampling rate should be at least twice the bandwidth [Shannon].
Besides, 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 hose in order to provide galvanic insulation to the measuring system. As
the connecting tube is exposed to external EMC, it should be designed and installed conforming to Electromagnetic
Compatibility practical installation guidelines ["EMC"], e.g. reducing its length.
Furthermore, the sensor signals might show some 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
10
regular frequency around 1000 Hz is observed. This oscillation may originate from standing waves in the short
11
connecting tube between compartment and sensor with a dependence on the length of the connection. A proper
12
filtering of this frequency e.g. by digital means removes these disturbances and provides a smooth pressure curve,
13
which can be compared with calculations (see Figure B-2). Furthermore, the influence of solid-borne sound should
14
be minimized e.g. by appropriate sensor adapters.
15
16
Figure B-2: Measured pressure curves before filtering (black) and after digital filtering (red, blue).
17
B.3 Accuracy of the measuring system
18
Accurate measurement of the static overpressure created by an internal arc is extremely challenging. When
19
acquiring pressure measurements from transducers mounted on test structures, it is often desired to quantify the
20
rate (rise time) of this overpressure, the relative timing between overpressure and/or structural response events, or
21
both. Regardless of which, it is important to select transducers with adequate rise times to acquire these
22
measurements with fidelity - pressure rise times demand extremely high-frequency response from the measuring
23
pressure transducers and their associated signal conditioning.
24
Therefore, it is necessary to select measurement system components (amplifiers, filters, displays, etc ) capable
25
of maintaining the measurement acquisition fidelity/reliability. However, measurement system components are
26
typically specified in terms of the upper frequency at which they provide -3 dB signal attenuation. A challenge then
Page 144
exists to infer the rise time capability of an entire measurement system based on the -3 dB specifications of its
individual components [Walter].
In addition, concurrent transient temperatures, pressure waves, intense light radiation, fragment impact, ionized
gases, and other undesired environments all attempt to couple into the transducer and its mount, the
instrumentation cabling, and other measurement system components.
Nevertheless, every manufacturers transducers will respond to these undesired environments, and thermal
transient responses (such as intense radiation) could be mitigated by application of ceramic or RTV (room
temperature vulcanizing) coatings (silicone elastomer) on the face of the transducer diaphragm [Walter 2]. These
provide a thermal delay, hopefully until the blast event is over. [Hilten] provides one such quantitative study of time
10
delays that are achievable.
11
If the influence of these undesired environments is not compensated for or isolated, the signal output from the
12
measurement system can be severely corrupted [Walter 3].
13
For example, strain, acceleration and temperature can all interact with the piezoelectric crystal within the
14
transducer to result in an erroneous pressure indication [Walter 2]. Furthermore, thermoelectric, photoelectric,
15
electromagnetic, triboelectric, and other energy induced effects can result in additive electrical signals that create
16
errors in the transducer output. All of these extraneous signals can be viewed as noise, which contaminates the
17
desired pressure measurement. To validate that the transducer output signal is not contaminated (i.e., it is solely
18
attributable to pressure), a combination of placebo [IEST] and check channels [Stein] could be used.
19
The placebo transducer can be applied in the test in the same manner as any of the operational transducers, but it
20
will not respond to mechanical inputs (pressure, acceleration, strain). Any electrical output from it identifies signal
21
contamination due to thermoelectric, photoelectric, electromagnetic, and/or triboelectric effects. Besides, to identify
22
the combined effect of acceleration and strain on the piezoelectric element, an operational transducer can be taken
23
and isolated it from the desired pressure environment. It becomes a check channel. Any signal output from the
24
check channel in excess of that produced by the placebo transducer would be noise induced by strain and/or
25
acceleration.
26
The combination of the placebo transducer and check channels allows to document almost all of the
27
aforementioned undesired responses with the noted exception of thermal effects due to transient temperature.
28
B.4 Summary
29
Summarizing, accuracy in the measurements is crucial because the typical value of acceptable tolerance of the
30
results to be obtained in the overpressure simulations if compared to the real laboratory test results are 5% to 10%
31
[Feitoza].
32
Besides, every pressure measuring system shall be required to undergo type and routine tests followed by
33
consecutive performance tests and checks throughout its service life, as IEC advises e.g. IEC 61298-2. These
34
performance tests and checks shall prove that the measuring system can measure the intended test overpressures
35
within the uncertainties given in the International Standards, and that the measurements are traceable to national
36
and/or international standards of measurement. The measuring system distributor must guarantee the
Page 145
reliability/exactness of his products, and the customer should verify and guarantee the accuracy of the pressure
system device along time.
3
4
REFERENCES:
5
6
[Bjrtuft2005]: T. Bjrtuft, 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
10
[Feitoza2010]: S. Feitoza, "Guidelines for the use of simulations and calculations to replace some tests specified in
11
international standards". COGNITOR Guide 2010.
12
http://www.cognitor.com.br/GUIDE_Simulations_v0_October2010.pdf
13
[Hilten1978]: Hilten, John, Vezzetti, Carol, Mayo-Wells, J. Franklin, Lederer, Paul, Experimental Investigation of
14
Means for Reducing the Response of Pressure Transducers to Thermal Transients, NBS Tech Note 961, January
15
1978.
16
[IEST]: Shock and Vibration Transducer Selection, Institute of Environmental Sciences and Technology, IEST RP-
17
DTE011.1, Sec. 7.9, Oct. 2002.
18
[Puers1993]: Puers, R. "Capacitive sensors, When and how to use them". Sensors and Actuators A, 37-38 (1993)
19
93-105 93-105.
20
[Shannon1998]: C. E. Shannon, "Communication in the presence of noise", Proc. Institute of Radio Engineers,
21
vol.37, no.1, pp.1021, Jan.1949. Reprint as classic paper in: Proc.IEEE, vol.86, no.2, (Feb.1998).
22
[Stein1992]: Stein, P. K., The Unified Approach to the Engineering of Measurement Systems, Stein Engineering
23
Services, Phoenix, AZ, April 1992.
24
[Walter2004]: P.L. Walter, Shock and Blast Measurement - Rise Time Capability of Measurement Systems?,
25
Technical Note, PBC Group (www.pcb.com), 2004.
26
[Walter2005]: P.L. Walter, Introduction to Air Blast Measurements. Part III: Guaranteeing that Validated Pressure
27
Measurements are Acquired". Technical Note, PBC Group (www.pcb.com), 2005.
28
[Walter2010]: P.L. Walter, Measuring Static Overpressures in Air Blast Environments, Technical Note, PBC
29
Piezotronics (www.pcb.com), 2010.
30
Page 146
ANNEX C:
COTTON INDICATORS: ENERGY ABSORPTION AND

FLAMMABILITY
2
3
C.1 Introduction
Two primary reasons for arc fault testing are safety and environmental concerns with SF6 (if appropriate).
Calculations and simulations are 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
10
personnel clothing. They are positioned in a relative proximity of the switchgear equipment (for example at 30 cm in
11
case of testing for authorized personnel access only and at 10 cm in case of general public access equipment). If
12
the cotton indicators catch on fire during the simulated arc test the test fails. Withstand testing of the switchgear to
13
internal arcing is typically accomplished by following one or more of the following standards:
14
15
16
IEC 62271-200 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
17
Type A arc-resistant construction at the front only
18
Type B arc-resistant construction at the front, back, and sides
19
Type C arc-resistant construction at the front, back, and sides, and between compartments
20
21
IEEE C37.20.7-2007 IEEE Guide for Testing Medium-Voltage Metal-Enclosed Switchgear for Internal
Arcing Faults includes
22
Type 1 similar to EEMAC Type A above
23
Type 2 similar to EEMAC Type B above
24
Annex A addresses suffixes B and C
25
26
Type 1C Type 1, but also with arc-resistance designs or features between adjacent
compartments
27
28
Type 2B Type 2 with LV instrument compartment door open relay and maintenance
personnel survive
29
30
Type 2C Type 2 with arc-resistance features between adjacent compartments

switchgear survives with minimum damage
31
Type 2BC The ultimate in protection combines types 2B and 2C
32
33
34
35
36
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.
Following Pass/Fail Criteria are used:
Door, covers, etc. do not open. Bowing or other distortion is permitted except on those which are to
be used to mount relays, meters, etc.
Page 147
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

2
Two types of cotton indicators are used: thick (so called Baumville, 150 gr/m ) and thin (so called Filtex, 40 gr/m ).
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 of the two kinds of
cotton samples.
C.3 Apparatus overview
10
Flamability testing of the cotton indicators was performed at WPI (Worchester Polytechic Institute in
11
Massachusetts) in the controlled environment of the laboratory setting. A cone calorimeter was used. It is a fire
12
instrument based on the principle of oxygen consumption calorimetry. The core of the instrument is the cone
13
shaped radiant electrical heater, which irradiates a horizontal sample. The heat flux level is set and controlled via a
14
temperature controller, which uses three thermocouples attached to the heating element (see Figure C-1).
15
An intermittent spark igniter, powered by a 10,000 V transformer and a 3 mm gap, located 13 mm above the
16
sample, provides the piloted ignition. The following properties are determined: rate of heat release rate, total heat
17
released, effective heat of combustion, time to ignition, mass loss rate, total mass loss, and smoke obscuration.
18
C.4 Overview of test procedure
19
The heating element is turned on and allowed to heat to the proper heat flux. The heat flux is measured using a
20
calibrated Heat Flux Gauge. The sample is then placed on the load cell (Figure C-2). The spark is placed and the
21
shutter opened. Observations for ignition and flameout are made.
22
C.5 Results of the testing
23
Although the WG has not performed exhaustive testing several different samples were tested for time to ignition at
24
different levels of the heat energy applied in the calorimeter. Since the reason for placing cotton samples in the
25
actual internal arc testing is to simulate the human clothing and the degree of heat exposure that could be harmful
26
to personnel, the samples should respond consistently to heat exposure and ignite faster at higher heat flux levels.
27
Page 148
Figure C-1: Overall test apparatus.
Figure C-2: Test specimen mounted on the edge frame and specimen under test.
4
5
6
7
The degree of harm that heat exposure could cause to humans is often calculated by the so called Stoll limit. Stoll
limit defines energy density causing second degree burn on humans. It is expressed as:
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 149
1
2
0.291
=
(C-1)
2
where in kJ/m is the energy density, is Stoll constant = 50.204 kW/m , and is the time to reach max
temperature .Time to reach depends on many factors such as the arc, arc configuration, arc electrodes,
gas, flammable/ combustible materials, calorimeters, distances, dimensions, etc. Typical result for reaching the
Stoll limit is shown in Figure C-3 as Temp rise () versus time [Schau2011]. When the Stoll limit crosses over the
test result for a given case the second degree burn to a human occurs.
6
7
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 high incident energy applied for a short time.
10
Figure C-4, Figure C-5 and Figure C-6 summarize results of the testing for the two types of the cotton samples.
11
They show the time to ignition (in seconds) as a function of the heat flux density in kW/m . Individual points show
12
the specific sample tests whereas the line is a linear regression function showing the trend. As expected the higher
13
the heat flux density applied the shorter the time to ignition. However, it is also clear from the testing that there is
14
visible scatter of results reaching up to 30-40 % (even if one rejects the two data points in Figure C-6 as bad).
15
Considering that the safety is the primary objective of the testing always the lowest data point should be taken into
16
account to account for the worst case scenario. This scatter sheds some doubts on the repeatability of the tests
17
both in the laboratory environment as described above where the test conditions are highly controllable as well as
18
with the actual switchgear subjected to internal arcing tests where the test conditions, particularly the arc behavior
19
and the exact amount of arc energy are impossible to control.
Page 150
1
2
Figure C-4: Time to ignition for thin fabric (actual test points and regression line are shown).
3
4
Figure C-5: Time to ignition for thick fabric (actual test points and regression line are shown).
Page 151
Time to ignition
3.5
3
Time (sec)
2.5
Thin
Thick
Regression1
1.5
Regression 2
1
0.5
0
50
70
90
110
Heat Flux (kW/m2)
1
2
3
Figure C-6: The same time to ignition for both thin and thick fabric tested repeated at different date 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, NFPA
70E, IEEE 1584 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 standards is the cotton samples used
in switchgear testing that supposed to simulate the human clothing.
10
Before comparing the test results we have to recall the heat energy density and heat power density defined for
11
PPE in the IEC/EN 61482-1-2, NFPA 70E standards.
12
13
Table C-1: Summary of NFPA 70E hazard/risk categories and IEC/EN 61482-1-2 classes.
14
Table C-2 and Table C-3 below represents the results of the cotton sample testing and compare with the arc flash
15
risk categories prescribed in NFPA 70E/IEEE1584 standards.
Page 152
Applied
Heat
Flux
Thin - test 1
kW/m2 J/cm2 cal/cm2

60
8,1
1,94
70
4,83
1,15
80
5,28
1,26
90
3,87
0,92
Thick - test 1
NFPA 70 E
Hazard
Category
1
1
1
1
J/cm2 cal/cm2
13,32
3,18
12,67
3,03
11,12
2,66
9,81
2,34
NFPA 70 E
Hazard
Category
1
1
1
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 standard.

Thin
J/cm2 cal/cm2
6,12
1,46
6,93
1,66
8,24
1,97
11,25
2,69
9,4
2,25
Thick
NFPA 70 E
NFPA 70 E
Hazard
Hazard
Category J/cm2 cal/cm2 Category
1
17,64
4,22
2
1
14,35
3,43
1
1
7,52
1,80
1
1
17,55
4,19
2
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 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 that with the hazard category 1 . 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 causes 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 the
Hazard category 1 without igniting. This means, that the switchgear internal arc test with such a cotton sample can
10
pass the flammability and yet produce the heat energy levels unsafe for Hazard category 1.
11
There could be several reasons for scatter of the flammability tests and the results presented, inconsistency of the
12
cotton samples (textile) and or the fibers used for producing the samples, humidity levels in the air where samples
13
were stored prior to calorimeter testing, pollution of samples, or other chemical reaction (oxidation, exposure to
14
different atmosphere with different gases, etc.) that the samples might have been exposed to. Regardless of the
15
reasons cotton samples used for internal arc testing are exactly the same as used in the laboratory testing and are
16
not controlled for these possible factors. Caution has to be taken to interpret results of the internal arc testing with
17
cotton samples as indicators and translate to the arc flash protection requirements described in safety standards.
In NFPA 70E Hazard category 1 defines Arc-Rated Clothing with minimum arc rating of 4 cal/cm2.
Page 153
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 in major part on the cotton indicators placed in front of the
equipment during testing. If the cotton samples catch on 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 bear some statistical uncertainty. On the other hand, the protective clothing designed for
personnel is classified by their resistance to the energy density that switchgear internal arcing could produce
10
(Joules/m or cal/cm ). Although there was very limited data available the scatter of the results may cause the
11
cotton samples withstand the energy densities above what would be considered safe (hazard category 1) without
12
igniting.
13
14
REFERENCES:
15
[Schau2011]: Holger Schau, Herbert Bessei, The Influence of Fuses on Arcing Fault Energy And Perrsonal
16
Protective Clothing Required, 9th INTERNATIONAL CONFERENCE on ELECTRICAL FUSES AND THEIR
17
APPLICATIONS, ICEFA 2011.
18
19
20
21
Page 154
ANNEX D:
ROOM PRESSURE CALCULATION USING CFD EXAMPLES
D.1 Introduction
In the experience of room pressure calculations are some main issues how to handle pressure relief openings or
how to position them. Further there are some topics about the first pressure peak. The first pressure peak is mostly
the main reason for a damage in the room. With knowledge of these issues the switchgear room can be made
more safe in the case of an arc fault. It is not always possible to follow all these guidelines, but for planning these
topics can be helpful to prevent higher damage in the case of an arc fault.
D.2 Pressure development and pressure relief
As a result of the internal arc, the pressure spreads out more or less spherically coming from the ignition location
10
and will be reflected from walls, the ceiling, the ground and other objects in the room. The first pressure wave, due
11
to spherically spreading of pressure causes a first high pressure peak at the near walls. This can only be influenced
12
by a pressure relief opening located immediately there. Otherwise a pressure relief opening has no influence on the
13
first pressure peak (Figure D-1 and Figure D-2).
First pressure peak

14
15
Figure D-1: First pressure peak 36 mbar at time 20 ms in a closed switchgear room.
16
First pressure peak
17
18
19
Figure D-2: First pressure peak 36 mbar at time 20 ms in the same switchgear room with an opening
(green).
Page 155
The first pressure peak can be decreased by increasing the distance of switchgears to 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
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 rather affects 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 for getting 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.
10
11
Average pressure t = 25 ms by both simulations about 5 mbar at the wall behind the switchgear.
12
The pressure relief in the room cannot be effective, until the first pressure wave has reached any opening. In
13
special cases an increasing of pressure relief opening cross section does not decrease the pressure amplitude, but
14
the rate of decrease. (For instance if the pressure peak amplitude is higher than the later existing room filling
15
pressure).
16
In long rectangular rooms the fluid tends to longitudinal pressure waves. Pressure relief openings are less effective,
17
if they are located on the long sides. In this case the pressure wave is passing over the opening. In long
18
rectangular rooms it is recommended to define pressure relief openings at the small sides. In that case the velocity
19
of pressure waves spreads out the gases perpendicular to the opening area and so the pressure decreases faster.
Page 156
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 spread out. These waves are swinging in different directions
depending on the room geometry and are passing pressure relief openings only temporary.
Pressure relief
5
6
7
Figure D-5: There is a longitudinal pressure wave with 6 openings at the long side of the switchgear room,
the first pressure peak is 3 mbar at 15 ms.
Pressure relief
9
10
11
Figure D-6: There is a longitudinal pressure wave with 4 openings at the long side and 2 openings at the
small side of the switchgear room, the first pressure peak is also 3 mbar at 15 ms.
12
With only 4 openings (2 openings at the small sides and 2 openings at the long sides) there is a longitudinal
13
pressure wave, where the first pressure peak is 3 mbar at 15 ms. The resulting pressure with 66% of opening cross
14
section here leads to nearly the same pressure relief over time as 6 openings at the long side (Figure D-5 and
15
Figure D-6).
16
D.3 How to determine worst case scenarios for simulations.
17
The selection of the damaging panel (position of arc fault) should be the worst case, i.e. the case with the highest
18
pressure impact in the room. Therefore the following statements are to help in selecting the worst position of the
19
panel under internal arc test:
Page 157
In general it should be selected a panel, having the longest distance to a pressure relief opening, so that the first
pressure relief time is late. Further the first pressure peak is not decreased by the pressure relief opening.
4
5
6
Figure D-7: Panel with the greatest distance to the opening has an arc fault, resulting in the worst case of
pressure increasing.
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
10
11
Figure D-8: Pressure in the left front ceiling corner (MP5) is 6.6 mbar, more than the first pressure peak.
12
There is a longitudinal pressure wave and 6 openings at the long side of the switchgear room, the first pressure
13
peak is 3 mbar at 15 ms, but in the ceiling corner front left (MP5) is a greater pressure peak of 6.5 mbar after a time
14
delay of 70 ms, the wave needs to reach there (Figure D-8).
Page 158
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 very fast from the duct trough many openings into the room. To get
a realistic first pressure peak it should be damaged a panel, 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).
Realistic first pressure peak
6
7
8
9
10
Realistic first pressure peak of a panel:

Panel directly located under a channel opening
Figure D-9: Switchgear with a channel at the top and absorber openings in this channel (green).
11
12
Figure D-10: Example of a power curve for a switchgear with 25 kA and a peak value of 62 MW at 10 ms.
13
Determining the arc fault power it is assumed a three phase arc fault, having the highest power peak and so the
14
highest energy exchange. The simulation calculates (unless otherwise agreed) a three phase circuit breaker arc
Page 159
fault. Further a generator remote short circuit is assumed. This current becomes constant after decay of direct
current terms after the first milliseconds (like in the arc fault test) (Figure D-10).
D.4 Summary
The issues discussed in this section are based on many arc fault simulations for customer switchgear rooms and
different switchgears. In an arc fault accident the pressure must out into the environment to prevent a damage of
the switchgear building. It is necessary to define pressure relief openings to let the pressure get out of the housing.
In these projects the cross section of the pressure relief opening and its position are the main issue for the
customer. The arc fault pressure simulations give the answer how to handle the pressure relief openings. Using
this information above the customers may also influence the maximum of the first pressure peak after the arc fault
10
beginning.
11
12
Page 160
ANNEX E:
EFFECTS OF NEUTRAL EARTHING ON THE INTERNAL

ARC
E.1 Introduction
For internal arc testing two different approaches exist, regarding the earthing (USA: grounding) 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".
10
Alternatively, in IEEE C.37.20.7 cl.5.2.6 it is stated that:
11
"the neutral of the supply system must be grounded or connected to the ground bus of the switchgear
12
assembly by a separate bus. If the neutral is grounded, the ground bus of the switchgear assembly must
13
be grounded. If the neutral of the supply system is connected to the switchgear assembly by a separate
14
bus, the switchgear assembly may be isolated from ground as required by the laboratory" followed by a
15
note stating that: "this requirement recognizes that test laboratories may not allow the flow of intentional
16
ground fault current due to safety and/or instrumentation concerns".
17
In other words, where IEC prescribes a floating situation (without significant neutral current), IEEE prescribes an
18
"extended neutral", with a low impedance, where significant neutral current may develop. It is unclear, how the
19
earthing regime affects the impact of an internal arc.
20
The difference between both test-circuits is shown in Figure E-1: switch NS closed is the "IEEE case" whereas NS
21
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.
22
23
E.2 Test Comparison
24
In order to investigate the consequences of both test-circuits, tests were carried out under identical conditions, the
25
only difference being the closed/open position of the switch NS in Figure E-1.
26
The test object (TO) was an arcing volume of 0.36 m , filled with air at ambient pressure with an open exhaust (225
27
cm diameter) to the environment, refer to Figure E-2. The electrodes ( 25 mm, length 50 mm three in line at 110
28
mm distance), were replaced after each test. A steel burning plate (connected to the earthing point via an 120 mm
29
Litze) was placed around the conductors at a distance of 110 mm from the electrodes.
Page 161
1
2
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 on 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 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, so less than 3 % of the
10
phase current value. The current measurement system has been checked by verifying that the sum of the
11
three phase currents is within 5 % of the neutral current.
12
Arcing voltages are about 300 Vrms to earth
13
The measured pressures for both situations (floating and neutral extension) are compared in Figure E-4.
14
15
16
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 162
50
0
-50
0
-50
0
-50
0
current in extended neutral (50 kA/div)
-50
0
-2
250
current in extended neutral (2 kA/div)

300
350
400
450
500
550
600
1
2
3
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).
4
1.8
ungrounded
pressure (b), current (100 kA)
1.6
1.4
extended neutral
1.2
1
0.8
0.6
current
0.4
0.2
0
260
280
300
340
320
time (ms)
360
380
Figure E-4: Pressure rise with and without extended neutral present.
Later on also an internal arc test was performed in the same test lab at 63 kA -0.5 s. test object was a large air-
filled switchgear cubicle, with an extended neutral (current herein limited to 20 kA because of test-plant limitations).
This test confirmed also that only a small fraction of the phase current returns through the extended neutral (in the
10
mentioned 63 kA case the neutral current was below 0.4 kA, so less than 1 % of the phase value).
11
E.3 Discussion
12
Earlier observations from several internal arc tests in various laboratories suggest that the neutral return current (in
13
case of low-impedance extended neutral) can have values up to about 20 % of the three-phase current. This may
14
then lead to currents across the enclosure from arc foot points to the neutral bus connection, causing dynamic /
15
thermal stresses to screwed connections, hinges of doors and other parts possibly not designed to withstand high
16
current flow. This may result in failure to pass the criteria.
Page 163
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 very severe way, eg. by initiating
the arc between two phases only and leave the non-used phase de-energized, a very large neutral current arises
(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
initiation
neutral
phase current unbalance (in %)

(rel. to average RMS value of 3 ph)
current in
neutral
arc energy
max
pressure
kA
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
63
0.5
3 phase
extended (20 kA max)
1.0
-0.1
-1.0
0.4
1.33
44.70
Table E-1: Test parameters and results.
So, 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 very good conductor in the presence of
10
the arc(s), it is obvious that in an unbalanced situation a neutral current will arise.
11
Unbalanced test-circuits are very common. Some test stations have the possibility to tune the impedance per
12
phase, some cannot. If the latter is the case, supply circuits will be unbalanced, as mutual inductances create
13
unbalance in the three-phase circuits that otherwise would have identical positive sequence impedance. In the test-
14
cases of Table E-1, the phase impedances were tuned individually to have an unbalance below 1 %.
15
Discussion should be started on the degree of unbalance that circuits in service might have. Tests with well-
16
balanced test circuits may be less severe for the switchgear having an extended neutral, because of the absence
17
of considerable current across the tank surface. On the other hand, every Amp that is conducted away through the
18
neutral, will not contribute to the arc-energy anymore.
19
E.4 Summary
20
It can be concluded that in case of an extended neutral, the severity of internal arc tests depend heavily on the
21
balance of the supply circuit.
22
In low-voltage (LV) testing significant neutral current is a well-known [Dunki1972] phenomenon. In LV testing, or in
23
MV switchgear testing with a supply voltage far lower than rated switchgear voltage, the arc voltage may be in the
24
same order as the supply voltage. Due to differences in (momentary) arc voltages, a considerable neutral current
25
may flow. This unbalance in the arc voltages can even occur in perfectly balanced supply circuits.
26
27
28
REFERENCES:
29
30
[Dunki1972]: J.R. Dunki-Jacobs, "The Effect of Arcing Ground Faults on Low-Voltage System Design", IEEE Trans.
31
On Ind. Appl., Vol. IA-8, Nr.3, 1972, pp. 223-230.
32
Page 164
ANNEX F:
F.1 Introduction
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.
FIGURES FROM SENSIBILITY ANALYSIS
F.2 Sensitivity analysis for selective cases
10
In order to illustrate the impact of different parameters on the pressure curve, several simulations are performed
11
with a chosen parameter varying within a range from half to double of the reference value for both air and SF6. The
12
reference value is the original parameter value defined by a test case (Table F-1). The results, depicted in the
13
figures below, are the families of pressure curves which include the simulation of reference case (black bold line),
14
the pressure curve in the arc compartment (continuous lines) and the pressure curves in the exhaust compartment
15
16
(dashed lines). In addition, the ratios of variation of , , and relative to the reference value are
17
corresponds to
18
ordinate is the ratio of the calculated result to the reference value.
shown by the lines with markers: each marked point corresponds to one simulation and the crossing point
the reference value. The abscissa is the relative variation of the studied parameter and the
19
2.00 1.00 SF6 0.390 0.0060 65.0 450
0.49
1.00
[Hz] [s]
50
0.34
0.90 1.00 SF6 1.217 0.0620 25.0 1800 3
0.17
1.50
60
1.34
10
0.40 0.70 Air
0.284 0.0445 24.0 364
0.12
0.12
1.400 0.100 50
1.18
11
0.55 1.00 Air
0.416 0.0490 38.0 320
0.12
0.20
0.730 0.400 50
1.18
28
0.80 0.80 SF6 0.509 0.0046 14.0 366
0.14
0.28
1.275 0.012 60
1.05
46
0.40 1.00 Air
1.190 0.1200 45.7 430
0.10
0.08
1.880 0.409 50
0.10
56
0.60 0.80 SF6 0.100 0.0020 20.0 500
0.50
1.00
1.000 0.010 50
0.30
57
0.35 0.90 SF6 0.300 0.0250 19.6 1200 3
0.10
0.33
50
1.00
59
0.65 0.75 SF6 0.300 0.0250 19.9 900
0.10
0.05
50
1.00
case
20
gas 1
3
[m ]
1
2
[m ]

[A] [V]
,1 ,1 2
[MPa] [MPa] [m ]
2
2
[m ]
Table F-1: Reference parameters used for the simulations.
21
Page 165
0.9
0.8
pressure [MPa]
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1.5
time [s]
0.8
0.8 Ex
0.4
0.4 Ex
0.52
0.52 Ex
0.64
0.64 Ex
0.76
0.76 Ex
0.88
0.88 Ex
1
1 Ex
1.12
1.12 Ex
1.24
1.24 Ex
1.36
1.36 Ex
1.48
1.48 Ex
1.6
1.6 Ex
SF6
0.9
0.8
0.7
pressure [MPa]
AIR
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1.5
time [s]
0.8
0.8 Ex
0.4
0.4 Ex
0.52
0.52 Ex
0.64
0.64 Ex
0.76
0.76 Ex
0.88
0.88 Ex
1
1 Ex
1.12
1.12 Ex
1.24
1.24 Ex
1.36
1.36 Ex
1.48
1.48 Ex
1.6
1.6 Ex
2.50
3.00
2.50
2.00
2.00
t burst
s burst
1.50
t max
1.00
p max
t burst
1.50
s burst
t 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
Figure F-1: Influence of factor (case 28).
Page 166
1.00
1.50
2.00
2.50
0.9
0.8
pressure [MPa]
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1.5
time [s]
14000
14000 Ex
7000
7000 Ex
9100
9100 Ex
11200
11200 Ex
13300
13300 Ex
15400
15400 Ex
17500
17500 Ex
19600
19600 Ex
21700
21700 Ex
23800
23800 Ex
25900
25900 Ex
28000
28000 Ex
SF6
0.9
0.8
0.7
pressure [MPa]
AIR
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1.5
time [s]
14000
14000 Ex
7000
7000 Ex
9100
9100 Ex
11200
11200 Ex
13300
13300 Ex
15400
15400 Ex
17500
17500 Ex
19600
19600 Ex
21700
21700 Ex
23800
23800 Ex
25900
25900 Ex
28000
28000 Ex
2.50
3.00
2.50
2.00
2.00
t burst
s burst
1.50
t max
1.00
p max
t burst
1.50
s burst
t 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
Figure F-2: Influence of fault current [A] (case 28).
Page 167
1.00
1.50
2.00
2.50
0.9
0.8
pressure [MPa]
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1.5
time [s]
366
366 Ex
183
183 Ex
237.9
237.9 Ex
292.8
292.8 Ex
347.7
347.7 Ex
402.6
402.6 Ex
457.5
457.5 Ex
512.4
512.4 Ex
567.3
567.3 Ex
622.2
622.2 Ex
677.1
677.1 Ex
732
732 Ex
SF6
0.9
0.8
0.7
pressure [MPa]
AIR
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1.5
time [s]
366
366 Ex
183
183 Ex
237.9
237.9 Ex
292.8
292.8 Ex
347.7
347.7 Ex
402.6
402.6 Ex
457.5
457.5 Ex
512.4
512.4 Ex
567.3
567.3 Ex
622.2
622.2 Ex
677.1
677.1 Ex
732
732 Ex
2.50
3.00
2.50
2.00
2.00
t burst
s burst
1.50
t max
1.00
p max
t burst
1.50
s burst
t 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
Figure F-3: Influence of arc voltage [V] (case 28).
Page 168
1.00
1.50
2.00
2.50
AIR
SF6
1.8
1.8
1.6
pressure [MPa]
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0.05
0.15
0.25
0.35
0.45
25000
12500
16250
20000
23750
27500
31250
35000
38750
42500
46250
50000
1.6
1.4
pressure [MPa]
25000
12500
16250
20000
23750
27500
31250
35000
38750
42500
46250
50000
1.2
1
0.8
0.6
0.4
0.2
0
0.05
0.55
0.15
0.25
0.35
0.45
0.55
time [s]
time [s]
2.50
2.50
2.00
2.00
t burst
1.50
s burst
t max
1.00
p max
0.50
t burst
1.50
s burst
t max
1.00
p max
0.50
0.00
0.00
0.00
0.50
1.00
1.50
2.00
2.50
0.00
0.50
1.00
Figure F-4: Influence of fault current [A], without overshoot (case 3).
Page 169
1.50
2.00
2.50
AIR
SF6
pressure [MPa]
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
0.1
0.2
0.3
0.4
0.5
0
0 Ex
0.52
0.52 Ex
1.05
1.05 Ex
1.57
1.57 Ex
2.09
2.09 Ex
2.62
2.62 Ex
3.14
3.14 Ex
1.8
1.6
pressure [MPa]
1.8
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
0.1
time [s]
0.2
0.3
0.4
0.5
time [s]
[s]
[MPa/s]
[s]
[MPa]
1.85
[s]
[MPa/s]
[s]
[MPa]
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
0.000
0.021
29.18
0.113
0.000
0.108
5.56
Figure F-5: Influence of fault initiation angle [radian] for slow process (case 56).
Page 170
0.300
1.44
0
0 Ex
0.52
0.52 Ex
1.05
1.05 Ex
1.57
1.57 Ex
2.09
2.09 Ex
2.62
2.62 Ex
3.14
3.14 Ex
AIR
SF6
0.8
0.8
0.7
pressure [MPa]
0.6
0.5
0.4
0.3
0.2
0.1
0.045
0.0225
0.02925
0.036
0.04275
0.0495
0.05625
0.063
0.06975
0.0765
0.08325
0.09
0.7
0.6
pressure [MPa]
0.045
0.0225
0.02925
0.036
0.04275
0.0495
0.05625
0.063
0.06975
0.0765
0.08325
0.09
0.5
0.4
0.3
0.2
0.1
0
0
0.05
0.1
0.15
0.05
0.1
time [s]
0.15
time [s]
1.10
2.50
2.00
t burst
1.50
s burst
t max
1.00
p max
0.50
0.00
1.05
t burst
s burst
1.00
t max
p max
0.95
0.90
0.00
0.50
1.00
1.50
2.00
2.50
0.00
0.50
Figure F-6: Influence of time constant [s] (case 59).
Page 171
1.00
1.50
2.00
2.50
AIR
SF6
3.5
3.5
pressure [MPa]
3
2.5
2
1.5
1
0.5
0.34
0.17
0.221
0.272
0.323
0.374
0.425
0.476
0.527
0.578
0.629
0.68
pressure [MPa]
0.34
0.17
0.221
0.272
0.323
0.374
0.425
0.476
0.527
0.578
0.629
0.68
2.5
2
1.5
1
0.5
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.1
0.2
0.3
time [s]
0.4
0.5
0.6
0.7
time [s]
1.10
1.00
1.05
t burst
s burst
1.00
t max
p max
0.95
0.80
t burst
0.60
s burst
t max
0.40
p max
0.20
0.90
0.00
0.00
0.50
1.00
1.50
2.00
2.50
0.00
0.50
1.00
Figure F-7: Influence of fault current duration [s] (case 1).
Page 172
1.50
2.00
2.50
0.284
AIR
0.284
SF6
0.284 Ex
0.284 Ex
0.142
0.35
0.142
0.35
0.142 Ex
0.142 Ex
0.1846
0.1846
0.1846 Ex
0.3
0.1846 Ex
0.3
0.2272
0.2272
0.2698
0.25
0.2698 Ex
0.3124
0.3124 Ex
0.2
0.355
0.355 Ex
0.15
0.3976
0.2272 Ex
pressure [MPa]
pressure [MPa]
0.2272 Ex
0.2698
0.25
0.2698 Ex
0.3124
0.3124 Ex
0.2
0.355
0.355 Ex
0.15
0.3976
0.3976 Ex
0.3976 Ex
0.4402
0.1
0.4402
0.1
0.4402 Ex
0.4402 Ex
0.4828
0.4828
0.4828 Ex
0.05
0.05
0.5254
0.15
0.25
0.35
0.45
0.4828 Ex
0.05
0.05
0.5254 Ex
0.5254
0.15
0.25
0.568
time [s]
0.45
0.5254 Ex
0.568
time [s]
0.568 Ex
2.50
0.35
0.568 Ex
2.50
2.00
2.00
t burst
1.50
s burst
t max
1.00
p max
0.50
t burst
1.50
s burst
t max
1.00
p max
0.50
0.00
0.00
0.00
0.50
1.00
1.50
2.00
2.50
0.00
0.50
1.00
Figure F-8: Influence of arc compartment volume [m3] (case 10).
Page 173
1.50
2.00
2.50
AIR
SF6
0.8
0.8
0.7
pressure [MPa]
0.6
0.5
0.4
0.3
0.2
0.1
0.025
0.0125
0.01625
0.02
0.02375
0.0275
0.03125
0.035
0.03875
0.0425
0.04625
0.05
0.7
0.6
pressure [MPa]
0.025
0.0125
0.01625
0.02
0.02375
0.0275
0.03125
0.035
0.03875
0.0425
0.04625
0.05
0.5
0.4
0.3
0.2
0.1
0
0
0.05
0.1
0.15
0.2
0.25
0.05
0.1
time [s]
0.15
0.2
0.25
time [s]
2.50
3.00
2.50
2.00
t burst
1.50
s burst
t max
1.00
p max
0.50
2.00
t burst
s burst
1.50
t max
1.00
p max
0.50
0.00
0.00
0.00
0.50
1.00
1.50
2.00
2.50
0.00
0.50
1.00
Figure F-9: Influence of arc compartment opening area [m2] (case 57).
Page 174
1.50
2.00
2.50
AIR
SF6
0.8
0.8
0.7
pressure [MPa]
0.6
0.5
0.4
0.3
0.2
0.1
0.33
0.165
0.2145
0.264
0.3135
0.363
0.4125
0.462
0.5115
0.561
0.6105
0.66
0.7
0.6
pressure [MPa]
0.33
0.165
0.2145
0.264
0.3135
0.363
0.4125
0.462
0.5115
0.561
0.6105
0.66
0.5
0.4
0.3
0.2
0.1
0
0
0.05
0.1
0.15
0.2
0.05
0.1
0.15
0.2
time [s]
time [s]
2.00
2.50
2.00
1.50
t burst
s burst
1.00
t max
p max
0.50
t burst
1.50
s burst
t max
1.00
p max
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
Figure F-10: Influence of arc compartment bursting pressure [MPa] (case 57).
Page 175
2.00
2.50
0.049
AIR
0.049
SF6
0.049 Ex
0.049 Ex
0.0245
0.0245
0.0245 Ex
0.0245 Ex
0.03185
0.45
0.03185
0.45
0.03185 Ex
0.03185 Ex
0.0392
0.04655
0.35
0.04655 Ex
0.0539
0.3
0.0539 Ex
0.06125
0.25
0.06125 Ex
0.0686
0.0392 Ex
0.04655
0.35
0.04655 Ex
0.0539
0.3
0.0539 Ex
0.06125
0.25
0.06125 Ex
0.0686
0.0686 Ex
0.2
0.0392
0.4
0.0392 Ex
pressure [MPa]
pressure [MPa]
0.4
0.0686 Ex
0.2
0.07595
0.07595
0.07595 Ex
0.15
0.1
0.035
0.0833 Ex
0.1
0.035
0.09065
0.055
0.075
0.095
0.115
0.135
0.07595 Ex
0.15
0.0833
0.09065 Ex
0.0833
0.0833 Ex
0.09065
0.055
0.075
0.098
time [s]
0.095
0.115
0.135
0.098 Ex
0.09065 Ex
0.098
time [s]
0.098 Ex
2.50
2.50
2.30
2.00
t burst
1.50
s burst
t max
1.00
p max
0.50
2.10
t burst
1.90
s burst
1.70
t max
1.50
p max
1.30
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
Figure F-11: Influence of arc compartment opening area [m2] (case 11).
Page 176
2.00
2.50
0.409
AIR
0.409
SF6
0.409 Ex
0.409 Ex
0.2045
0.2045
0.2045 Ex
0.2
0.2
0.2045 Ex
0.26585
0.19
0.3272
0.18
0.17
0.38855
0.38855 Ex
0.16
0.4499
0.4499 Ex
0.15
0.51125
0.14
0.51125 Ex
0.5726
0.13
0.3272
0.3272 Ex
0.17
0.38855
0.38855 Ex
0.16
0.4499
0.15
0.4499 Ex
0.14
0.51125 Ex
0.51125
0.5726
0.13
0.5726 Ex
0.63395
0.12
0.26585 Ex
0.18
0.3272 Ex
pressure [MPa]
pressure [MPa]
0.26585
0.19
0.26585 Ex
0.5726 Ex
0.63395
0.12
0.63395 Ex
0.11
0.63395 Ex
0.11
0.6953
0.6953 Ex
0.1
0.02
0.75665
0.04
0.06
0.08
0.1
0.6953
0.6953 Ex
0.1
0.02
0.75665 Ex
0.75665
0.04
0.06
0.818
time [s]
0.08
0.1
0.818 Ex
1.10
0.75665 Ex
0.818
time [s]
0.818 Ex
1.10
1.05
t burst
s burst
1.00
t max
p max
0.95
0.90
1.05
t burst
s burst
1.00
t max
p max
0.95
0.90
0.00
0.50
1.00
1.50
2.00
2.50
0.00
0.50
1.00
1.50
Figure F-12: Influence of exhaust compartment opening [m2] (case 46).
Page 177
2.00
2.50
F.3 Pressure overshoot equation
The pressure curve in the arc compartment after the burst 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 comparatively 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
7
8
considering that the power input rate is constant () = . The overshoot happens if the pressure at the instant
of burst increases i.e. the derivative (slope) is positive. This slope can be calculated using ideal gas law (A-34).
( 1)
then
> 0
( 1)
( + )
>
(F-2)
After several transformations and using mentioned above equations this condition can be expressed as follows:
+1
10
(F-1)
2 2(1)
3
+ 1
>

1
,1
(F-3)
This is equivalent to (3-5) with the constant
1/2
(F-4)
11
The dimension of is
12
F.4 Summary
13
The impact of different parameters on the pressure curve in arc and exhaust compartment is depicted in a visual
14
way. Those include factor , fault current, arc voltage, fault initiation angle, time constant, arc duration, arc
Kkg
+1
2 2(1)
+ 1
=
15
compartment volume, arc and exhaust compartment opening area, and arc compartment bursting pressure. The
16
derivation of the overshoot condition for slow process from equations of the basic method is described.
17
18
19
ANNEX G:
EFFECT ON REPLACING SF 6 WITH AIR ON BURNTHROUGH
20
G.1 Introduction
21
Section 6.4.4 presented the qualitative approach to compare the burn-through times in SF6 gas and in Air. From
22
Table 6-15 it can be seen that a large number of parameters and most influencing one are indicating that shorter
Page 178
burnt-through times are in SF6 than in air. This Annex takes a closer look at some of those parameters, most
important one, and their dependency on type of gas.
G.2 Arc voltage
As for , it is known that there exists an electrode fall voltage region in an arc, which is specific quite thin area
close to the electrode and provides significant voltage drop as schematically shown in Figure G-1. Joule heat
provided in an electrode fall voltage region can be considered to be the most dominant energy source for melting
through an enclosure, because it releases significant power in a quite close area to enclosure. Figure G-2 indicates
electrode fall voltages of SF6, argon, and air for various electrode materials measured by a laboratory test
[Yokomizu1996]. As seen in Figure G-2, electrode fall voltages depend on electrode material more dominantly than
10
type of gas, and they differ among the gases by only a few percentage for the same electrode material.
Arc Column
11
12
13
Figure G-1: Voltage drop along an arc column. Electrode region appears as a dominant power source for
melting the enclosure.
Page 179
1
2
3
Figure G-2: Electrode fall voltages of SF6, argon and air for various electrode materials at atmospheric
pressure [Yokomizu1996].
G.3 Heat of chemical reactions
Chemical reaction between vaporized metal and gas should be also investigated for both SF6 and air, because it
may cause significant difference in temperature that the enclosure would be exposed to, and pressure rise in the
enclosure. It is known that aluminum reacts with both SF6 and air (oxygen) and these reactions provide 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
10
reacts with SF6 and O2. However, O2 content in air is only 21 %, therefore more exothermic energy could be
11
released in SF6 than in air under the same filling pressure.
12
One mol of Aluminum + SF6:
13
One mol of Aluminum + Air (oxygen):
3
3
Al + SF6 AlF3 + SF4 + 850 kJ
2
2
3
1
Al + 02 Al2 O3 + 837 kJ
4
2
(G-1)
(G-2)
14
G.4 Radius of Arc root
15
Radius of arc root is one of the influencing factors for burn-through time, because it corresponds to concentration of
16
input power at the erosion spot on the enclosure. Although some papers have reported comparison of arc radiuses
17
of wall-stabilized arcs with forced gas flow for different gases [Yokomizu JRED2000], none was found for a free-
18
burning arc in an enclosure like an internal fault arc. Figure G-3 shows analytical temperature distribution of wall-
19
stabilized gas arcs with forced gas flow as in breaker arcs (wall radius 5 mm, current 50 A, gas flow rate
Page 180
5 litter/min). As seen in Figure G-3, it is known that arc radius in SF6 is thinner than that in air in such small current
mainly due to dissociation of SF6 molecule around 2,000 K. On the other hand, Figure G-4 shows that the
difference of arc radius (that is here defined as the zone carrying 99% current of total) becomes closer in higher
current condition even with forced gas flow. Consequently, arc radiuses could be considered probably comparable
between SF6 and air for high current free-burning internal arcs.
6
7
8
Figure G-3: Analytical temperature distribution of wall-stabilized arcs with forced gas flow [Yokomizu,
JRED2000]
9
10
Figure G-4: Conductive radius dependence on arc current [Yokomizu, JRED2000]
11
12
G.5 Velocity of Arc Motion
13
First stage after the arc ignition is its displacement along the conductor until it meets a spacer which could act as a
14
barrier against its movement. This time belongs to the total time to burn-through.
2 0
=

[Lutz1983]
Page 181
(G-3)
0.28

[Chu1985]
(G-4)
According to equation (G-3) and (G-4), replacing SF6 by air at same pressure will lead to accelerate the axial
velocity because of the ratio of their gas density. The arc will then spend less time axially blocked against the
partition barrier which has for effect to decrease the theoretical time to burn-through.
Refering to standard requirements, the arc should be initiated in the vicinity of a partition furthest from the point of
injection (IEC 62271-203). Due to this recommendation, it appears that time where arc is moving axially is usually
lower 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
10
then the moving character of the arc, this could affect highly the time of burn-through.
11
G.6 Pressure exerted on the tank wall
12
Subsection 6.2 deals with the evaluation of the pressure development for air or SF6. Burn-through are phenomena
13
that will normally happened lately after bursting disk opening. It is clear that the pressure exerted by the gas on the
14
weakened structure plays an important role because the ablation process decreased the thickness of the wall and
15
this is combined to a loss of mechanical properties of the material.
16
Even if the reader must be careful and conduct it own pressure/temperature calculation for the considered case,
17
one can generally expect higher pressure after some hundreds of ms after disk opening inside an SF6 filled
18
compartment than in an air compartment.
19
G.7 Summary
20
This annex presents discussions on some influencing factors for burn-through comparison between SF6 and air, as
21
a complement to Table 6-15 of Section 6.4.4. Even if the consequence of SF6 replacement by air is not obvious, we
22
theoretically could expect shorter burn-through times in SF6 than in air. This conclusion in mainly drawn from the
23
change of behaviour regarding the heat of chemical reactions, the difference between azimuthal velocities, and the
24
change of pressure exerted on the tank walls after the bursting disk opening.
25
26
REFERENCES:
27
[Chu1985]: Chu, F.Y and Lutz, F. and Braun, J.M and Stuckless, H.A, Effects of power arc faults in gas-insulated
28
substations, CIGRESymposium, Brussels, Contribution 340-04, 1985.
29
[Lutz1983]: Lutz, F. and Chu, F. Y., Burn-through of GIS Enclosure Due To Power Fault Arcs, EEE Transactions
30
on Power Apparatus and Systems, 1983.
31
[Yokomizu1996]: Y Yokomizu and T Matsumura and R Henmi and Y Kito, Total voltage drops in electrode fall
32
regions of SF6, argon and air arcs in current range from 10 to 20 000 A, Journal of Physics D: Applied Physics, Vol
33
29.
Page 182
[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 183

A3 14 (SC) 01 (B)

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

A3 14 (SC) 01 (B)

Uploaded by

Copyright:

Available Formats

TOOLS FOR THE SIMULATION OF

EFFECTS OF THE INTERNAL ARC IN

ISBN : (To be completed by CIGRE)

ISBN : (To be completed by CIGRE)

TOOLS FOR THE

EXECUTIVE SUMMARY ................................................................................................................. 5

CALCULATION OF PRESSURE USING A BASIC MODEL ................................................ 17

SENSITIVITY ANALYSIS OF THE BASIC MODEL ............................................................. 38

CALCULATION OF PRESSURE WITH CFD SOFTWARE .................................................. 52

INTERNAL ARC TESTING ................................................................................................... 60

EFFECT OF INTERNAL ARC ON STRUCTURES ............................................................... 74

INTERNAL ARC SIMULATION REVIEW ........................................................................... 118

CONCLUSION ..................................................................................................................... 124

ANNEX A: EQUATIONS FOR PRESSURE RISE CALCULATION ........................................ 128

ANNEX B: PRESSURE SENSORS: TYPES AND ACCURACIES ......................................... 143

ANNEX C: COTTON INDICATORS: ENERGY ABSORPTION AND FLAMMABILITY ......... 147

ANNEX D: ROOM PRESSURE CALCULATION USING CFD EXAMPLES .......................... 155

ANNEX E: EFFECTS OF NEUTRAL EARTHING ON THE INTERNAL ARC ........................ 161

ANNEX F: FIGURES FROM SENSIBILITY ANALYSIS.......................................................... 165

ANNEX G: EFFECT ON REPLACING SF6 WITH AIR ON BURN-THROUGH....................... 178

prediction (utilizing extrapolation) remains limited.

Figure 0-1: Arc resistance test per IEC 62271-200.

including both SF6 and air-insulated switchgear (Figure 0-2).

effects of an internal arc:

b. Mechanical stress on enclosure and buildings

performing a design analysis based on tests of a similar design

enclosure, resulting in a pressure rise.

1-2), and endanger personnel. Failures typically occur due to:

material or mechanical defects

Figure 1-2: Internal arc fault test.

Arc temperature can exceed 10,000 C.

released by exploding 2 kg of dynamite.

Debris may travel at speeds up to 1000 km/h.

and by several energy exchanges.

manner, it is difficult to describe them quantitatively.

properties are assumed to be independent of temperature and pressure.

effects such as exothermic reactions, ablation of material and mixing of gases.

equations is solved three-dimensionally with a computational fluid dynamics (CFD) solver.

Doesnt consider spatial nonuniformity of gas parameters

To quickly calculate uniform pressure

Not applicable if the relief opening

rise inside an arc compartment and the

area is too large in relation to the

exhaust volume in typical MV switchgear

and HV GIS applications.

Calculations are not reliable, when

Doesnt consider gas mixtures in the

Doesnt consider spatial non-

To calculate uniform pressure rise

uniformity of gas parameters

as under 1) adding further

(pressure, temperature, density) in

approximations to better match test

each volume part.

results and calculation.

Limitations and applications depend

High effort for the modeling and

For calculating spatial pressure

meshing of the rooms and

distribution and gas flow in odd shapes

geometry and large rooms.

Requires large computing power

Table 1-1: Model selection table.