Frictional Ignition OF Powders: A Review by Geoff Lunn Health and Safety Laboratory, Harpur Hill, BUXTON, Derbyshire SK17 9JN, UK

Download as pdf or txt
Download as pdf or txt
You are on page 1of 34

FRICTIONAL IGNITION

OF
POWDERS:
A REVIEW

by

GEOFF LUNN

Health and Safety Laboratory, Harpur Hill,


BUXTON, Derbyshire SK17 9JN, UK
1. INTRODUCTION

Surveys of industrial incidents in powder and dust handling plant show that in a
substantial percentage, friction and mechanical failure and flames and flaming material
are known ignition sources. Surveys for the UK1,2 covering 1979-1988, and reviewing
303 events, showed friction and mechanical failure to be the reason for ignition in 18% of
these incidents, and flames and flaming material to be responsible in another 15%.
Overheating and spontaneous heating featured in a further 17%. Similarly, a survey by
the Berufsgenossenschaftliches Institut für Arbeitssicherheit (BIA)3 showed mechanical
causes to be the most frequent source of ignition, with smouldering nests the second most
frequent. The relevant percentages from the BIA survey were 26% for mechanical
sparks, 11% for smouldering nests and 9% for mechanical heating.

A review by Billinge4 of industrial friction ignition incidents divided potential frictional


sources into three groups:

Low energy : approximately 10J, eg 500g falling 2m


Medium energy: approximately 1kJ, eg 25kg falling 4m
High energy : in excess of 1MJ, eg road tanker crash

The highest percentage by far of frictional ignitions is in the medium energy range, with
some 50% of these due to impact.

When surfaces contact, impact and/or friction and grinding can produce sparks and hot
spots; both are potential ignition sources for dust clouds or dust accumulations. If dust
becomes trapped at the point of contact, some dusts may ignite at much lower friction-
generated temperatures than those required to directly ignite a dust cloud. Clearly, plant
operations such as grinding and screw-feeding have the potential to produce mechanical
sparks and hot surfaces, as have accidental circumstances such as the presence of tramp-
metal.

Impact is a short duration event in which local hot spots may occur and heated fragments
of material are torn away to produce sparks. Friction and grinding occurs over a longer
duration and may produce hot surfaces and showers of sparks.

The ATEX Directive5 has introduced the hazard of mechanical ignition into legislation
for the first time and there is currently underway an extensive programme of standards-
making on techniques for prevention of ignitions due to mechanical effects. Although
there is some published work on ignition by mechanical sparks6 and hot surfaces7, there
is little in the way of practical guidance except in limited areas. Gibson has recently
reviewed the likely scenarios for mechanical ignitions and the availability of published
data, and has discussed the information required to complete an overall picture of the
risks in real plant8.

This paper is a review of the literature on the effects that frictional heating and sparking
can have on combustible dusts, on the ways in which frictionally ignited dusts can burn,
and on the methods by which combustion in burning dusts can ignite an explosive dust
cloud or propagate into a more extensive dust accumulation.

2. REVIEW
The energy dissipated when two surfaces slide over each other produces heat and thus hot
surfaces. Generally, temperatures not exceeding the lower of the melting points of the
materials concerned develop, but, if wear and transfer of materials takes place,
temperatures can approach the higher of the two melting points9. Mechanical sparks are
hot particles of surface material torn off by impact, friction and grinding. If the spark
material can oxidize in air, the spark temperature increases during flight.

2.1 Mechanically Generated Sparks

The ignitability of a spark depends on its temperature, its size and probably its velocity.
Powell9 has reviewed published data on the temperature of sparks and their ability to
ignite gas-air and vapour-air mixtures. Particles of 100 micron diameter need to be
greater than 2000oC if they are to ignite methane-air; but carbon steel sparks in air were
measured at 1850oC only, 1750oC in methane-air and 1500oC in fuel-air mixtures where
towns gas, hydrogen and acetylene were the fuels. The addition of fuel to air decreases
the oxygen concentration and so slows the reaction between the steel and the air. Light
metals burn at well in excess of 2000oC.

The surface materials are of crucial importance in determining whether an ignition of a


specific explosive atmosphere will occur. Powell9, in his review of gas and vapour
ignitions, has produced tables that give a ranking of the incendivity of ignition sources
produced by a range of impact and rubbing situations. These tables are reproduced as
Tables 1 and 2. Available evidence in the literature suggests a similar ranking for
ignition of dust clouds.
Mechanically generated sparks take three forms:
Grinding sparks - a quick contact (20-50 milliseconds) of two surfaces in relative
motion
Friction sparks - rubbing together over an extended time (0.5 - 2.0 seconds)
of two surfaces in relative motion
Impact sparks - single contact of two surfaces in relative motion
Impact sparks are the result of the application of high forces - sufficient to cause
permanent deformation - for the order of a millisecond9, and they account for the largest
proportion of frictional ignitions in non-mining industry - 65% of dust and powder
ignitions is the figure from a published review of incidents due to mechanical ignition4.
During impact, energy dissipation occurs at a high rate eg. a hammer blow of 1 J for 1 ms
gives a power of 1 kJ if all the available energy is dissipated9, although only about a third
goes into heating the impact surfaces. Measurements by Pedersen and Eckhoff10 show
that ignition of dust clouds by sparks from single impacts is very difficult. They
concluded that up to a net impact energy of 20J, single, tangential impacts between steels,
steel and rusty steel or concrete were unable to ignite clouds of grain and feed dust, or
flour even when dry. Titanium impacting on rusty steel was able to ignite dust clouds,
the probability of ignition increasing as the MIE of the dust decreased.

Similarly, in experiments by Reimer on the ignition of methane-air mixtures by steel-on-


steel impact sparks11, ignition by a flying spark was never observed. Mixtures were
either ignited either at the point of impact of the drop weight on the test plate or by a
glowing spark lying on the floor of the explosion vessel.

Reimer’s experiments identified the important factors influencing the incendivity of the
impact sparks. The hardest material pairs gave the most intensive sparks; the more the
available kinetic energy is converted into shearing energy and frictional heat, the larger
the number of sparks produced. As the available energy increased, so did the incendivity
of the sparks. The roughness of the test plate, however, had an effect on the type of
sparks produced. If the potential energy was not sufficient to break through the grooves
making up the roughness, small, individual, highly incandescent pieces of steel were
produced from the grooves. At higher potential energies a larger, cohesive splinter was
removed from the plate material below the grooves, resulting in lower incendivity. If,
however, the roughness can be broken off, rather than sheared, (eg. lateral grooves rather
than longitudinal ones) the incendivity of the sparks is likely to be similar to those from a
smooth plate. The effect of surface roughness on spark incendivity depends on the
hardness of the steel; generally the harder the steel the higher the increase in spark
incendivity.

Between 35o and 65o, the angle of impact of the drop weight on the test plate had little
effect on spark incendivity. Rust had an inhibitory effect on the formation of incendive
sparks, especially on rough plates, and the addition of a 1 mm thick coal dust layer to the
test plate decreased the incendivity due to its lubricating effect.
A very incendive type of spark is produced if the impacting materials can produce the
thermite reaction. Gibson et al12 used a stainless steel hammer striking a rusty mild steel
target with an aluminium smear to test for ignition of dusts. Of the 95 powders used 46
produced flame following a thermite flash, of which 27 produced flames that propagated
beyond the ignition zone. Impacts of standard quality aluminium on rusty steel produced
no sparks in some experiments by Pedersen and Eckhoff10, only a smear of aluminium.
Similarly, when Gibson et al used several metals and alloys to test for the production of
the thermite reaction during impact, soft metals such as zinc and aluminium rarely
produced the reaction in the first strike against rusty steel smeared with aluminium. Only
if the soft metal became impregnated with rust and aluminium after repeated impacts did
the reaction occur. Hard metals such as steels and brass readily produced the thermite
reaction.

Friction and grinding are examples of continuous or intermittent contact giving a rubbing
action - intermittent contact between rotating components, for example, is more likely to
be a rubbing action than impact9 - and they can produce sparks that are capable of
igniting dust clouds13. The likelihood of ignition depends on the dust and the materials of
contact. Dahn and Reyes14 showed that contact between a rotating grinding wheel and
stainless steel rods produced ignition, but when the rods were aluminium, ignition did not
occur. The peripheral speed of the wheel was 20 m/s. Dahn and Reyes’ data also show
that the relative speed between components required for ignition depended on the dust13.
The shower of sparks from a high-carbon steel in contact with a grinding wheel can
contain many large burning particles (> 100 microns)15; ignitions of 10% moisture grain
elevator dust were consistently obtained; ignitions were obtained with as few as five
sparks.

The first real attempt at quantifying the probability of igniting explosive dust
atmospheres by specific types of spark was made by Ritter6 and Muller16, who introduced
the idea of the equivalent electric spark energy. This equivalent ignition energy is
defined17 as the amount of energy in the spark from the discharge of a capacitor that has
the same incendivity as the ignition source under characterisation. The electrical circuit
contains an inductance to ensure the discharge is over an extended time period.

The equivalent ignition energy is measured by comparing the range of ignitable


concentrations for a given mechanically generated ignition source to the spark discharge
Minimum Ignition Energy (MIE) over the same range. Figure 1 demonstrates the
procedure for a flintstone friction spark and lycopodium dust18. The ignitable
concentration for the friction sparks ends at concentrations for which the MIE is 100 mJ.
This, then, is the value of the equivalent ignition energy for this particular type of friction
spark.
Figure 1. Determination of the Equivalent Ignition Energy
(from Reference 18)

Figure 1 shows that it is easier to ignite a dust cloud when the turbulence is low.

The characteristics of the sparks depend on the nature of the contacting surfaces.
Guidance has been published on the igniting ability of different types of mechanical
spark19, with dusts classified either as ignitable or non-ignitable based on the measured
values of Minimum Ignition Energy (MIE) and Minimum Ignition Temperature (MIT).
These data are reproduced in Figure 2, and show that the type of spark, whether impact,
friction or grinding, influences the likelihood of ignition. The higher effect of steel
friction sparks as compared to steel grinding sparks can be explained as due to a high
temperature of the torn away material because of longer exposure to frictional heating.
Figure 2. Conditions for Ignition of Dusts by some Mechanically
Generated Sparks
(Reference 19)

The guidance given in Figure 2 is applicable only to mechanically generated sparks, and
should not be applied when the ignition source is the hot surface produced by friction.
Neither should it be used indiscriminately even for sparks; the characteristics of
mechanical sparks can alter when the frictional behaviour creating them changes, for
example, if the load between the surfaces varies11. Furthermore, there is no evidence in
the literature for the implication in Figure 2 that dust clouds react to different ignition
sources with the same relative sensitivities. In fact, the very opposite appears to be true.
Gibson et al12 demonstrate a lack of correlation between ignition of dusts by the thermite
reaction and measurements of Minimum Ignition Temperature from the Godbert-
Greenwald furnace. They state that MIT measurements cannot be used to assess
sensitivity of a powder to ignition by the thermite reaction. If measurements of dust
MIEs and MITs are displayed on a graph, as in Figure 3 20, the lack of correlation
between these ignitability characteristics is immediately obvious.

Figure 3. Measurements of Minimum Ignition Energy


and Minimum Ignition Temperature of dusts.
(Reference 20)

It is also well known that the minimum energy required for ignition of a dust cloud by an
electric spark varies with the characteristics of the spark. The spatial and temporal
distributions of energy in electric sparks are important factors. Boyle and Llewellyn21
showed that the minimum capacitor energy capable of igniting a given powder decreased
substantially when a series resistance was included in the circuit. Because the series
resistance absorbs a large fraction of the capacitor energy during discharge, the energy
going into the spark gap is only a small part of the theoretical stored energy, generally
about 5 to 10%13. Thus, in Boyle and Llewellyn’s experiments, at a series resistance of
104 - 105 ohms, for some dusts the ignition energy was reduced to approximately 1% of
the energy necessary without the resistance. Similar changes were measured by Line et
al22. Smielkow and Rutkowski23 increased the spark duration either by adding a large
inductance or by adding a large resistance to the circuit, and again the theoretical ignition
energy from the capacitor decreased by a factor of ten, approximately.

The optimum spark discharge duration measured by Matsuda and Naito24 and estimated
in reference 13 from the data in references 21 and 22 were 0.1 - 1.0 milliseconds,
decreasing with the net spark ignition energy.
Measurements by Parker25 showed that for some dusts there was a fairly distinct region
where the discharge duration produced the lowest ignition energies. With other dusts,
however, no such region occurred. Parker used four dusts, and those with the higher
ignition energies showed the optimum spark duration effect.

The physical length of the spark gap also has an effect on the ignition energy. Tests by
Ballal26, using metal dusts and carbon, demonstrated that there was an optimum spark
gap length for each dust, and that it increased as the ignition energy at the optimum gap
length increased, although the result for carbon did not fit in with the results for the metal
dust. The optimum spark gaps fell in the range 2 - 7 mm. Measurements by Norberg et
al27 showed the optimum spark gap length to be in the range 6 - 8 mm, for short duration
capacitative sparks.

Likewise, complex characteristics of mechanical sparks and showers of sparks will affect
the ignition process. It is difficult to see the justification for suggesting that the graphs in
Figure 2 linking MITs and the equivalent ignition energy give guidance that is
universally applicable to dusts. Dusts can have very different mechanisms of ignition and
combustion depending on the means of ignition and the characteristics of the ignition
source, and there is no universal ranking of dust ignitability. An important practical
requirement is that these graphs should be checked for their true meaning, and, if
possible, extended to different ignition sources.

The likelihood of frictional ignition at rubbing surfaces depends on the relative velocity
and the pressure at the contact point. A relation between the igniting ability of rotating
parts and their relative speed, V, has been published19.

V < 1 m/s no additional hazard from friction contact


V = 1 - 10 m/s each case must be considered on its own merits taking into
account data for the specific product and material
V > 10 m/s an ignition hazard exists

The test results from which these velocities are derived apply to grinding and friction
sparks and hot surfaces. The tests were done with a 4 mm diameter steel pin contacting a
grinding wheel. The contact pressure required for ignition at a given relative velocity
rises rapidly as the relative velocity drops to 2-3 m/s, as is shown in Figure 4.
Sparks falling onto dust deposits could produce smouldering combustion and eventually
produce a sufficiently energetic burning volume that could act as an ignition source for a
dust cloud. The amount of energy necessary to produce such an event from a small, high
temperature source such as a single spark is not known, but Eckhoff describes a test for
measuring the energy of an electric spark necessary to ignite a 2mm thick dust layer13.
With the specific example that Eckhoff gives - a pyrotechnic powder - the minimum net
spark energy is approximately 1-2 mJ with an optimum spark discharge time of 2-3x10-4
seconds. The ignition energy rises to 8-10 mJ when the spark duration is 10
milliseconds.

Figure 4. Disc
Limit curves for materials:
the formation of St 37, V2A, St 79 -
grinding and Pin diameter:
friction sparks as 4 mm
well as hot
surfaces
(Reference 19)
2.2 Hot Surfaces produced by mechanical means

Although friction-generated sparks ignite dust clouds, this is, experimentally, a rare event
compared to the frequency of ignition by the hot surface simultaneously produced. The
likelihood of ignition depends on the dust characteristics and the area and temperature of
the hot surface. Temperatures in the range 500-1000oC can be developed in local areas
by friction between metal surfaces28. Bartknecht reports data on dust cloud ignitions
produced by steel pins rubbing against steel discs18. Temperatures of at least 1200oC are
required when the pins are 4 mm and 6 mm in diameter. This temperature was reached in
a rubbing time of 1.5-3.5 seconds. A graph relating the measured Minimum Ignition
Temperature of the dust to a parameter involving the optimum surface temperature of the
pin, the diameter of the rubbed spot, the length of the temperature-discoloured zone and
the temperature difference at the rubbed spot is shown in Figure 5. It is difficult from
Bartknecht’s text to fully understand what these definitions precisely mean, but the
parameter has the unit of area modified by a temperature ratio, and as this parameter
increases the more likely are dusts of decreasing ignitability to ignite. Measurements
reported by Pinkwasser29, also show that dusts with Minimum Ignition Temperatures of
approximately 400oC require temperatures of 1000oC-1200oC to ignite if the area of the
hot surface is in the range 12-20 mm2 (Figure 6).

Figure 5. Ignition of Dusts by Hot Pins (from Bartknecht, reference 18)


(Minimum Ignition Temperature vs Properties of the Hot Surface)

Topt - Optimum Temperature oC


<T - Temperature Difference oC
DR - Diameter of hot surface mm
S - Length of temperature – discoloured zone
The configuration of the hot-surface is also a factor. The MIT measurements from the
BAM oven are lower than expected from an extrapolation of Pinkwasser’s results
because the hot surface is a relatively large one and surrounds the dust cloud as a cylinder
- Pinkwasser’s areas were made up of hot wires and coils.

Figure 6. Effect of Surface Area on Minimum Ignition Temperature


of Dust Clouds
(Reference 29)
The most important hazard as regards a hot surface is the potential ignition of a dust layer
that has settled on it. The layer ignition temperature is measured in a standard test for a
depth of 5 mm30, but because of the insulating effect of dusts, thicker layers can ignite at
lower temperatures. The practical dangers are that a smouldering or burning layer can act
either directly as an ignition source for a dust cloud or nests of burning material can break
away from deposits and ignite a dust cloud in another part of the plant.

Harper, Plain and Gibson31 have discussed the burning behaviour of powder
accumulations on hot surfaces. The stages of ignition and the form of the combustion
zone can be complex. Some powders burn directly in the solid phase either with a flame
or by smouldering, others melt and burn as a liquid, whilst some burn with a large
amount of flame. Some dusts can evolve large amounts of flammable gas when
subjected to heat. A change from solid to liquid or agglomeration/expansion of dust
particles to give an extended mass of material can block burning if diffusion of oxygen to
the seat of burning is prevented.

A measure of the ignitability of a dust layer and intensity of burning of a dust layer is the
Combustion Class (CC)30. This classification is based on the behaviour of a defined dust
heap when subjected to a gas flame or hot platinum wire:
CC1 No ignition; no self-sustained combustion
CC2 Short ignition and quick extinguishing; local combustion of short duration
CC3 Local burning or glowing without spreading; local sustained combustion but no
propagation
CC4 Spreading of a glowing fire; propagation of smouldering combustion
CC5 Spreading of an open fire; propagating open flame
CC6 Explosible burning; explosive combustion

The train firing test assesses flammability with reference to different ignition sources, and
allots the Combustion Class. This test is shown in Figure 7.

Figure 7. Testing the flammability and burning


behaviour of dust deposits with different
ignition sources
(Reference 30)

At temperature above ambient, the Combustion Class of a particular dust will increase.
Dextrin, for instance, has a CC of 2 at 22oC, but a CC of 5 at 100oC.

The train firing test does not, however, at first sight, give an indication of the likelihood
that a burning nest will form. The powder or dust needs to coagulate and any
accumulations that detach need to travel for some distance as a unit. Furthermore,
combustion behaviour in plant with flowing air will be different to when the air is
stationary, as Zockoll32 has shown.
The ignition hazard of deposits close to or at points of friction has been studied, for
powders capable of fast decomposition, by Gibson and Harper33. With three different
methods for producing mechanical friction, Gibson and Harper found that small
differences in test method produced very different behaviour with regard to both
initiation and progress of combustion. The mass of metal at the site of frictional heating
is an important factor in determining the likelihood of ignition. They concluded that
laboratory attempts to simulate industrial mechanical friction to characterise powders,
and make the tests relevant to industrial situations would be generally unsuccessful. For
unconfined powders Gibson and Harper decided that the Train Firing Test was
appropriate, using a small flame as ignition source, and with a (2:1 w/w) mixture of
silicon/lead dioxide mixture intermixed with the first 1-2 cms of the train. A temperature
of 1000oC was considered to represent the likely temperature in a situation of high
mechanical friction.

Once a smouldering or burning deposit has developed, nests, ranging in size from several
millimetres to several centimetres, may break off and, carried along by an air stream until
they reach an extensive dust cloud, then act as an ignition source.

Work at Syngenta34 on the ignition of dust clouds has shown that clouds can be ignited by
various burning or smouldering ignition sources. Three ignition sources -
paraformaldehyde, which burns with a flame, Fe3+(H2), which smoulders and
incandescent particles of saw dust - were used. Sulphur and lycopodium dust clouds of
various concentrations were blown over the first two of the ignition sources, and both
dusts ignited. The incandescent particles were introduced into the dust clouds soon after
the clouds had been produced. The sulphur clouds ignited, but the lycopodium did not.

These tests were repeated with dusts of various MIT values, from 270oC to above 1000oC
as measured in the Godbert Greenwald furnace. With the burning layer, dusts with MITs
above 600o to 800oC did not ignite; with the smouldering layer, dusts with MITs above
340oC, approximately, did not ignite; with the incandescent particles, dusts with MITs
above about 330oC did not ignite.
Some tests using layer ignition sources of various areas and temperatures showed that as
the area decreased, for a given temperature, the dust MIT above which a dust cloud did
not ignite increased.

At present, however, it is unclear what properties of a dust nest make it an effective


ignition source in practice. Despite all the reports of ignition incidents in industrial plant,
experimental studies have in the main indicated that ignition of dust clouds by hot nests is
not easy. Pinkwasser29 showed that smouldering material entering a pneumatic
conveying line was soon extinguished - the distance to extinguishment depending on the
dust concentration. Pinkwasser used an 80m length of 100/110 mm i.d. pipe, with six 90o
elbows, which ended in a cyclone tested to 10 bar. Nests of smouldering material were
introduced through an air-lock at the end of the pipe remote from the cyclone. The
powder conveying rate was measured from the weight of powder introduced into a
known volume of conveying air over a given time, and the mean powder concentration
calculated. The temperature of the smouldering material was measured by
thermocouples, and the distance of travel of the smouldering material by spark detectors
and flame detectors. The powders examined were three grades of flour, with Kst - values
below 100 bar m s-1, and CC ratings of 5.

Only one of the powders was capable of producing smoulder nests. The powder
properties were: moisture content 8.9%, bulk density 290 kg/m3, median particle size 120
microns and Minimum Ignition Energy approximately 1000 mJ. A much coarser-grained
flour with a bulk density of 510 kg/m3 failed to produce nests, as did a finer grained flour
with the properties: moisture content 13.1%, bulk density 440 kg/m3, median particle size
55 microns and Minimum Ignition Energy approximately 500 mJ.

Glowing clumps up to 1.5 cm diameter, with temperatures of 500oC to 550oC, were fed
into the line. In dust-free air, glowing particles were transported, in conveying velocities
of 10 and 20 m/s, as far as 68m. But as the powder loading in the airstream increased the
distance to extinction of nests of approximately 10g decreased substantially, as shown in
Figure 8. Extinction was promoted by breaking up of a smouldering nest into individual
glowing particles. No dust explosions were detected when powder-loading was within
the explosive range.

Figure 8. Hazard range for pneumatic transport of patent flour


and screenings: from Pinkwasser’s experiments.
(Reference 29)
Pinkwasser concluded that it was impossible, with the dusts tested, for smouldering nests
of approximately 10g weight to be conveyed in a powder loading of greater than 1 kg/m3
of air, but that in conditions where powder loading is low - exhaust systems and under
startup - smouldering material could be conveyed over relatively long distances.

Alfert et al35 used a pneumatic transport system ending in a 5.8 m3 filter unit. At an air
speed of 35 m/s, only very strong nests were transported. Charcoal nests of 50 cm3
volume entered the filter only as small agglomerations (<1 cm3) even after a relatively
short distance (11 m), and with maize starch as the explosive atmosphere in the filter
even large nests (0.5 litre) produced no ignitions in the system. When fine wood dust
was used to give the explosive atmosphere, ignition occurred in the filter, but not in the
pipeline.

Pinkwasser36 showed those smouldering nests with a temperature of 700oC, free-falling


into dust clouds, did not produce an ignition in explosive atmospheres of wheat flour or
wheat starch. Only when nests of at least 25 mm diameter and weight of at least 15g
landed on the bottom of the 1m tall test column did some ignitions occur. Jaeger37 found
that smouldering nests could be produced only with dusts having a Combustion Class
greater than 3. A minimum nest area of 75 cm2 and surface temperature of 900oC were
required for igniting clouds of dust with Minimum Igniting Temperatures less than
600oC. Alfert et al35 noted that nests of low mechanical strength disintegrated during a
fall and generated a large fire ball that acted as an ignition source. Mechanically stable
nests were capable of igniting the cloud only when they reached the silo floor, but could
be covered with dust before an explosion had time to start. In these tests, nests of known
size were dropped through dust free air in a silo with a height of 22m. The nests were
made of charcoal, cork dust and wood dust. Charcoal has a strong nesting structure; no
burning of the charcoal particles occurred, and when the nest reached the floor of the silo,
a shower of glowing particles was produced. Cork dust forms stable nests; flaring up of
the nests was noted at approximately half the height of the silo, and on the silo floor.
Wood dust produces unstable nests; these could break up in the very top part of the silo
creating a fireball, remain intact and flare up in the upper half of the silo or reach the
floor and flare up on impact or not at all.

When nests were dropped into an explosive atmosphere of maize starch, nests of wood
dust with a size of 0.5 to 1 litre produced no ignition in 40% of the tests. Nests with
volumes of 1 to 1.5 litres ignited equally in the upper part of the silo or on the floor.
Nests of cork dust produced no ignitions, but could start fires in settled maize starch
powder. Charcoal nests could produce ignitions if they were mechanically broken up,
and also a short time after reaching the floor.
Zokoll32 has reported some tests using milk powders as both the nest material and the
explosive dust cloud. Initial tests in which both dust and smouldering nest were dropped
into the test chamber simultaneously showed that fist-sized nests at temperatures
approximately 100oC or more above bulk powder Minimum Ignition Temperature as
measured in the BAM furnace test did not ignite ground corn and milk powder clouds.
Even when the nests were broken up by blades while falling ignitions did not occur.
Ignition of dust clouds could not be achieved with nests which did not burn but had
internal temperatures of 700-800oC. Nests of 1200oC did ignite the dust clouds but only
after impact on the floor of the explosion vessel. When dust was dropped over
smouldering nests on the vessel floor, cloud explosions could occur at temperatures of
about 860oC. A flaming nest could, however, be practically extinguished by the dispersal
of milk powder around it in the explosion vessel. Tests on the development of
smouldering in nests under the influence of a 0.5 m sec air stream showed that
smouldering developed differently depending on the type of milk powder. At higher air
speeds open fires occurred in relatively large quantities of skimmed milk powder. At air
speeds of about 10 m/s compact smouldering nests reached temperatures of 1200oC in the
hottest spots. The transition from smouldering into open fire occured around 800 -
850oC, depending on the type of milk.

In summary, the likelihood of ignition of a dust cloud by a hot nest depends on the
temperature of the nest, its residence time in the dust cloud and the availability of oxygen
to the burning area. Greische and Brandt38 have shown that the Minimum Ignition
Temperature of a dust-cloud decreases substantially when the residence-time of a dust in
a Godbert-Greenwald furnace increases as Figure 9 shows. The longer a local part of a
dust cloud remains in contact with a smouldering nest, the more likely it is that an
ignition will occur. It appears, also, that burning nests can be extinguished by dust clouds
and an ignition prevented, if conditions are right.
Figure 9. Variation of Minimum Ignition Temperature of
Brown Coals with residence time in
the Godbert – Greenwald Furnace.
(Reference 38)

The size and temperature that are necessary if a solid hot body is to ignite a dust cloud
have been studied by Alfert et al35. Steel cylinders with a diameter of 20 mm and lengths
of 20 - 30 mm were heated and introduced into an explosive cloud of maize starch in a
silo. Cylinders of 75 g weight and a temperature of 950oC ignited the cloud, but 75 g
cylinders at 900oC, and 50 g cylinders at 950oC did not.

Eisfeld39 has conducted experiments on the heating of coal dust deposits by an


electrically heated wire, and has made recommendations for the prevention of a fire
hazard from high power intrinsically safe circuits. On the basis of his experiments,
Eisfeld calculated the maximum permissible power dissipation at an intermittent contact
of a single stranded copper cable with a cross-section of 0.5 mm2, as well as the
associated maximum current. The values were 0.72W and 4.3A. Safety in respect of a
fire hazard in coal dust is assured if the circuit has been tested with the break flash
apparatus with a maximum rms short circuit current of 2.6A or less. If this current is
greater than 2.6A, and the intrinsic safety with regard to firedamp has been demonstrated,
the fire hazard with regard to coal will be avoided provided the cables are single stranded
and the copper cross-section is 0.2 mm2 or greater.

Another potential hazard from smouldering nests is propagation into a larger


accumulation of burning from a hot nest fallen on to a dust deposit. Work has been done
on the effect of deposits of embedded hot nests and hot bodies, by Krause and Schmidt40,
and by Nelson7. Alfert et al35 have done some work on the effect of surface hot spots. A
smouldering nest or other hot object may fail to ignite a cloud, but could cause extensive
smouldering in a deposit which could then act as a much more effective ignition source.
In the work by Krause and Schmidt40, combustion on the surface did propagate back into
the bulk sample. In Alfert et al’s work, hot steel cylinders were dropped onto a 20 cm
thick layer of maize starch or wood dust34. If any effect occurred, it was either on instant
flaring up of the contact surface or initiation of a slow burning fire. With a 75 g cylinder,
flaring up of wood dust occurred at 650oC, maize starch at 700oC. A slow burning fire
occurred in wood dust at 450oC, but not up to 700oC in maize starch. The glow
temperature of the wood dust was 280oC and the Minimum Ignition Temperature (MIT)
430oC; the MIT of the maize starch was 470oC.

Krause and Schmidt40 have investigated the behaviour of hot nests and hot bodies
embedded in powder accumulations. The materials used were fine and coarse cork,
beech wood and cocoa. The hot bodies were porcelain spheres with diameters of 25, 30,
35 and 40 mm, and the hot nests were held in cylindrical mesh baskets 11, 15, 25, 34, 40
and 45 mm in diameter and a height to diameter ratio of unity. The accumulations of
dusts were held in cylindrical mesh baskets with volumes of 6, 12, 50 and 200 litres with
height to diameter ratios of unity. The hot bodies and nests were heated in a furnace,
positioned at the centre of the cylinder and the bulk dust then introduced. A smouldering
front propagated from hot nests, if ignition occurred, moving through the dust
accumulation at an approximately constant smouldering temperature. The larger the size
of the glowing nest, the more likely the dust accumulation was to ignite. The minimum
size of hot nest required for ignition depended on the dust. The smouldering temperature
produced by hot bodies was independent of the size of the hot body, but decreased as the
size of the dust accumulation increased. The temperature of the hot body necessary to
produce ignition of a given size of an accumulation of a given dust fell as the size of the
hot body increased. The onset of smouldering was caused by temperatures above 400oC.
The rate of smouldering propagation was connected with the availability of oxygen. Fine
cork dust with a bulk density of 59 g/m3 smouldered more quickly than coarse cork with
a bulk density of 171 g/m3. The propagation velocity was 0.33 mm/min.

Leisch, Kauffman and Sichel41 have studied the smouldering behaviour of grain dust. A
layer of 10.2 cm was used in all tests, with the dust held in a box 56.2 cm by 61.0 cm.
The upper surface of the layer was level with the floor in grain dust varied from 0.21
mm/s to 0.48 mm/s depending on the depth in the layer and the ignition source depth.
With an airflow of 4 m/s over the layer, the smouldering velocity increased by a factor of
2 to 2.5. In some tests with the airflow, a transition from smouldering to flaming
combustion occurred after a substantial cavity had formed due to propagation of the
combustion wave. The temperature of the combustion zone increased by about 180oC
once flaming combustion occurred. The availability of oxygen influenced the velocity of
propagation.

Clearly, hot surface ignition is not only related to surface temperature. Geometry, contact
time, air flow, contamination and the chemistry and history of the material can have an
effect. If the air above the dust deposit is at a temperature higher than normal room
temperature, the requirements for ignition of a deposit by embedded hot objects may fall.
Layer ignition temperatures typically decrease by 40oC-60oC at an air temperature of
100oC. Various tests have been developed to measure the ignition behaviour of dust
deposits in streams of hot air.

The IChemE Guide, Prevention of fires and explosions in dryers described tests
developed to simulate various conditions and obtain measurements of the temperature at
which exothermic reaction begins42.

The aerated powder test simulates conditions where a hot air stream passes through
material. The powder is held in an 80 mm long, 50 mm diameter glass cylinder closed at
each end by sintered glass. Air is passed downwards through the powder at the same
temperature as the surroundings inside the fan-assisted oven in which the test is
performed. A screening test may be performed, but a more thorough study is made with
isothermal tests at different temperature and periods. The number of isothermal tests will
depend on the precision required in the result.

Impurities, slight changes in composition and autocatalytic reactions can have a marked
effect on the temperature at which exothermic activity begins. Performing the test with a
temperature cycle akin to that likely to occur in practice is a useful addition to the
isothermal tests. In order to minimise the hazard, a material temperature 30-50oC below
the measured temperature is generally recommended, but this safety factor should not be
the only basis for safety. A layer test where hot air passes over deposits or layers uses a
layer 75 mm by 40 mm and 15 mm deep. A screening test can be used, but isothermal
tests, each lasting perhaps for several hours, are the main tests by which an ignition
temperature is obtained. If the layer depth properly simulates practical conditions, the
temperature at which an exotherm can progress to red heat can be used as a basis for safe
procedures rather than the somewhat lower temperature at which exothermic activity
begins. An adequate safety margin is usually 20oC.

A bulk powder test utilises the same apparatus as for the aerated powder test but the hot
air flows around the sample and not through it. Screening tests, isothermal tests, low heat
loss tests and simulation of process cycles can be performed. If the operating
temperature is 50oC less than the measured exotherm onset temperature from the
screening tests, dangerous decomposition is unlikely to occur at least up to 1 tonne
capacity. However, if this temperature difference is less than 50oC, or the operating cycle
is longer than the test duration, or the measured temperature is less than 200oC,
isothermal tests should be performed at 50oC above the process temperature and with a
duration longer than the operational time, followed by a low heat loss test using a Dewar
flask.

Nelson7 has reviewed the literature on smouldering combustion, and concluded that the
rate of smouldering is controlled by the diffusion of oxygen to the combustion front.
Thus, the more porous the sample the higher the rate of smouldering. Thermal
conductivity is also important - the higher the thermal conductivity, the more heat
required to initiate smouldering, all other things being equal.

In an extensive series of tests Nelson used small electric heaters to simulate a hot spot
formed by frictional heating. The hot spots were controlled at a constant power. Nelson
found that localised combustion took place around the hot spot at sub-critical power, and
self-sustaining combustion above. The temperature at which wheat flour began to
combust fully was 500oC. Nelson’s results showed that the size of the bed had negligible
influence on the critical power if the bed was large relative to the size of the hot spot, and
that for iron oxide there was little change in the critical power with size of hot spot. As
the ambient temperature of the deposit increased the critical power fell, but as the
ambient temperature approached the isothermal ignition temperature, the critical power
did not fall as rapidly as expected, because of a decreased power output from the deposit
material due to some oxidation at these elevated temperatures. The critical power
increased with thermal conductivity, all other things been equal, decreased as the heat of
reaction increased and increased as the activation energy increased.
2.3 Industrial Plant

Prevention of ignitions is an unavoidable part of the basis of safety in the design and
operation of powder handling plant, and taking heed of published data leads to some
simple precautions. In Scholl’s opinion43, screw feeders, rotary valves and elevators
should have external bearings and be operated with a relative speed of < 1 m/s, middle
bearings in screw feeders should be temperature-monitored, and off-track running, slip
and temperature should be monitored at the bottom bearing of some elevators. Overload
protection may also be appropriate. It is impossible to identify all the places in specific
items of equipment where mechanical impact and friction difficulties may arise. Dust
handling plant comes in a variety of shapes and sizes and design features may alleviate
the risk from mechanical ignition. For instance, if large masses of metal surround the
contact surfaces, the temperature will not rise as rapidly or as high as in locations which
are more insulated against heat loss. The size of the contact area may vary depending on
the scale of the equipment.

A new concept of ‘constructional safety’ has been introduced for non-electrical


equipment for use in explosive atmospheres; a draft standard has been prepared by
Working Group 2 of CEN Technical Committee 30544. Types of equipment that contain
no ignition source in either normal operation or cases of expected malfunction have good
design and engineering principles applied to them so that the probability of mechanical
failures leading to incendive impact or friction sparks and hot surfaces is reduced to a
very low level. Requirements for bearings, power transmission systems, clutches and
couplings, brakes and braking systems and springs are included in the draft document.

In this section of the paper some items of industrial plant prone to mechanical ignition
problems are discussed.

Publications which contain descriptions of powder handling plant include a booklet


published by the ‘Machine Safety’ section of the International Social Security Agency
(ISSA), Working Group 6: Dust Explosions: Collection of Examples45 and ‘Bulk Solids
Handling’ by Woodcock and Mason46.

2.3.1 Grinders and Pulverizing Machines

Blockage, high-speed internal fitments and tramp metal are all capable of producing
mechanical ignition sources in grinders and pulverizing machines. There have been
occasional incidents in the US where explosions in coal pulverizers have been ignited by
hot surfaces or sparking due to tramp metal, but most ignitions have been due to
spontaneous combustion of coal deposits47.

a) High-speed mills

Pinned disk mills, beater mills and hammer mills are invariably sources of ignition.
Hammer mills46 employ swing hammers mounted on rotating shafts to impact raw
material and pulverize it against an adjustable block. They have relatively high power
requirements. Avoidance of ignition sources is only applicable as an explosion
prevention method in exceptional circumstances, when the powder or dust has a very
high Minimum Ignition Energy and Minimum Ignition Temperature45.

b) Tumbling mills

Ball Mills, tube mills and rod mills pose an ignition hazard due to a substantial heating
effect. Ball mills comprise a rotating horizontal cylinder with steel or alloy balls inside.
As the cylinder rotates the material is crushed by the impact and grinding action of the
balls. The particles are carried to classifiers by hot air. Reference 47 gives an example
that in mills of 2-3 m diameter, the temperature can rise from 20oC to 100oC within the
space of 2 hours. Lumps of smouldering material are a hazard from some powders, if the
deposits are thick enough, and the residence time in the mill is long enough. These lumps
can act as ignition sources for dust explosions.

A document published by the Expert Commission for Safety in the Swiss Chemical
Industry48 allots products, with regard to milling, into safety classes on the bases of
Minimum Ignition Energy (MIE), Minimum Ignition Temperature (MIT), Burning Class,
Thermal Stability, presence of flammable liquids and the result of the Impact test (see
Figure 10). Products presenting the least hazard (class SCM 0) have MIEs above 1 J, and
MITs above 500oC, which, judging by the data in Figure 2, means they are difficult to
ignite by mechanical sparks. Reference 48 recommends, however, that effective ignition
sources must be avoided by separation of tramp metal, regular inspection of bearings for
hot running and other rotating parts for friction and caking and other precautions.

c) Roll mills

Roll mills are vertical cylinders containing a rotating bowl in contact with two or more
spring-loaded rollers47. Raw material is crushed between the rollers and a grinding ring
on the bowl. Roll mills are usually not an ignition hazard if frictional heating and tramp
metal can be avoided45.
d) Cutting mills
If large quantities of fine dust are produced in cutting mills, then the mills pose the same
threat as high-speed mills45.

2.3.2 Crushers

The hazard from high-speed crushers is the same as from high-speed mills. Low speed
crushers, with a circumferential speed of around 1 m s-1, have been shown by experience
not to be ignition sources45.

Figure 10. Safety Classes for Milling Operations (Switzerland. Reference 48)
2.3.3 Mixers

When mixers have moving internal components frictional ignition hazards are present45.
If the internal components are slow moving (a circumferential speed of less than 1 m s-1)
and the mixer has low power requirements (no higher than 4 kW) then experience
suggests that no additional ignition hazards are introduced. When components move
more rapidly and the power requirements are moderately high, then the ignition hazard
can be avoided, if:

- there is a high degree of filling (ƒ 70%) to limit occurrence of explosive


dust atmospheres

- the mixer is operated at reduced speeds ( 1 m s-1) during charging and


re-charging

- sufficient clearance is present between moving parts to avoid contact

- tramp material is avoided

- products with a tendency to spontaneously ignite under the operating


conditions are not used

These rules for limiting speeds and limiting power requirements generally apply to other
dust handling plant with moving internal components such as screens, dust collectors and
classifiers45.

The conditions inside plant, and the dusts ignition properties, are also important. In jet
mills with classifiers, frictional ignition sources may occur, but because of the high
turbulence ignitions of dusts with a Minimum Ignition Energy > 10 mJ and a Minimum
Ignition Temperature >3000C is not expected45. When the ignitability of the dust is
greater, ignitions must be expected.
2.3.4 Feeders

a) Rotary Table Feeders

This device is a horizontal circular table below and concentric with the hopper opening.
A fixed blade acts as a plough to remove the material. Most of the shearing resistance to
the rotation of the table comes from the mass of material in the centre. The table rotates
at, typically, 2-10 revs/min. There is, probably, little likelihood of frictional ignition.

b) Screw Feeders

The enclosed screw or auger conveyor is designed to run at relatively high speeds (200
revs/min to 2000 rev/min). There is always the risk that dust can be trapped or heated at
contact points between the flight and the casing, and some dusts may ignite and begin to
smoulder.

2.3.5 Conveyors

Conveyors are items of equipment that transport powders from one place to another.
They have moving parts, power transmission systems and bearings.

An example of an Ignition Hazard Assessment for a belt conveyor is given in the


documentation for a one-day seminar on the ATEX Directive given under the TREX
European Network programme49. The effects of temperature rises in moving parts, loss
of lubricant, misalignment of components, excessive vibration in worn parts, slippage in
the belt, friction between the belt and moving parts and filling by dust of gaps between
moving parts which are close together are some of the situations which should be
considered.

2.3.6 Elevators

Centrifugal discharge elevators generally have speeds of travel of the buckets in the range
1.3 to 2 m/s, but with free flowing granular material the speed can be greater than 3.5 m/s
without difficulty. Positive discharge elevators, generally used for sticky materials or
those that tend to pack, have a speed of 0.7 m/s. Continuous bucket elevators are
typically operated at around 1 to 1.3 m/s.

There is always the danger that moving buckets may strike the elevator casing and
produce hot spots and frictional sparks among other mechanical friction effects. The
same ignition hazards as arise with conveyors should be considered.
3. CONCLUSIONS

Although there is much experimental data in the published literature it is not sufficient to
produce a unified and wide-ranging set of guidance. There are too many gaps in the
information that is necessary for both linking the general behaviour and characteristics of
mechanical ignition sources to the ignitability and combustibility properties of dust
clouds and dust layers and relating the ignition sources produced and studied in the
laboratory to those likely to be created in full-scale powder handling plant.

3.1 Ignition by mechanical sparks

There is the possibility that some potentially explosive dust atmospheres can be ignited
by single impact sparks, though the evidence suggests that this is true only for very easily
ignitable dusts and with impacting materials producing high temperature sparks. Single
sparks have failed to ignite methane-air mixtures during flight11, and ignition of maize
dust clouds by a single spark was never witnessed in other tests10. A more likely danger
is that a spark falling on a dust deposit may initiate smouldering combustion leading
eventually to a sufficiently energetic volume of burning that could act as an ignition
source for an explosive dust atmosphere. Sparks falling onto a surface may smoulder for
a time - the same single sparks that failed to ignite a methane-air mixture when in flight
did produce ignitions if they landed on the floor of the test chamber and continued to
glow there11. An example of the level of energy required to ignite a dust layer comes
from tests with an electrical spark ignition apparatus, in which, admittedly for a
powdered pyrotechnic, energy of 1-2 mJ was sufficient13.

Dusty environments are more likely to be ignited by showers of sparks, even when the
mechanical event is an impact; although showers consisting of only five particles have
been shown to be capable of igniting dust clouds15. The thermite reaction readily ignites
some dust clouds12.

Some guidance has been produced linking the igniting ability of certain types of sparks to
the ignitability characteristics of dusts19, but any temptation to apply it to the hot surfaces
that are simultaneously produced should be resisted. Furthermore, the sparks produced
from two materials vary depending on the way the two surfaces rub together11. At low
loads, sparks may be at a relatively low temperature and do not oxidise rapidly while in
flight. At high loads, hotter sparks are produced and combustion during flight will be
faster, leading to higher temperatures. In addition, the relative sensitivity of dust clouds
to ignition is not the same for all ignition sources12, 20. Care must be taken in applying to
practical situations information that relates ignition sensitivity to correlations between
ignition measurements from widely different ignition sources. Real mechanical sources
may be different to laboratory created ones and the different modes of combustion that
dusts can have may render simple classifications void.

The ignition risk from mechanically produced sparks requires systematic investigation.

3.2 Ignition by hot surfaces

Conditions inside dust handling plant are so varied that any attempt to describe a typical
mechanically produced hot surface is practically impossible. There is an infinite number
of permutations of temperature, area and configurations. The power inputs necessary to
produce a given temperature and the temperature necessary to ignite a deposit of a given
dust can vary widely. For instance, a power input of less than five watts can produce
temperatures in a hot spot of 10mm diameter sufficient to ignite deposits of wheat flour
(500oC)7; tens of watts are required to ignite conical dust deposits over a heated box with
dimensions 100mmx50mmx25mm (280oC for saw dust)50; hundreds of watts are required
for two steel wheels rotating against each other (270oC for wood flour)51. Experiments
show that the ignition temperature of dust clouds depends on the size and configuration
of the hot surface. Clearly, potential ignitions in practice have to be anticipated by
determining the type of frictional event likely to occur and applying data and information
from a test that is close as practicable to the expected event.

In practice, only a limited set of tests to determine the risk of ignition will be available
and they will be expected to cover all likely situations; they will be chosen not only on
technical grounds but on cost and time considerations also. What is first required is an
accepted test for measuring the ignition temperatures of dust deposits by hot surfaces,
followed by accepted tests for generating the hot surfaces produced by rubbing surfaces.
There are opportunities in current research work to go some distance towards this goal.
The European funded project SMT4-CT98-2273: Electrical Ignitions in Dusty
Environments with a Potential Risk of Explosion is part concerned with ignitions of thick
layers and large accumulations. The outcomes from these parts of the project may lead to
tests not only applicable to electrical apparatus but also to the hazards from mechanically
generated hot surfaces. Ignitions of dust accumulations by rotating surfaces submerged
in the dust is being studied at the Health and Safety Laboratory. Comparisons between
the results from these experiments and other methods of igniting dusts at hot surfaces are
being made to see whether ignition temperatures from relatively easy tests can be used to
quantify the risks from mechanically produced hot surfaces. The risks of an explosion
produced by hot nests encountering a dust cloud are also being studied in this project.
4. REFERENCES

1. Abbott, J.A. BMHB Survey of dust fire and explosions in the UK 1979-84
(British Materials Handling Board, ISBM 0 85624 4554).
2. Porter, B. Industrial Incidents. Paper presented at Dust Explosions: Assessment,
Prevention and Protection, 24th November, London (1989).
3. Jeske, A and Beck, H. Evaluation of dust explosions in the Federal Republic of
Germany, EUROPEX Newsletter No 9 p2 (July, 1989).
4. Billinge, K. The frictional ignition hazard in industry - a survey of reported incidents
from 1958 to 1978. Fire Prevention Science and Technology. 24, 13-19, (1979).
5. Directive 94/9/EC of the European Parliament and of the Council of 23 March 1994
on the approximation of the laws of the Member States concerning equipment and
protective systems intended for use in potentially explosive atmospheres.
6. Ritter, K. Mechanics erzeugte funken als zundquellen. VDI - Berichte 494, VDI -
Verlag GmbH, Dusseldorf, 129 (1984).
7. Nelson, M. Detection and extinction of fire and smouldering in bulk powder. Paper
to ‘Dust Explosions. Protecting People, Equipment, Buildings and Environment. IBC
Technical Services, London 11/12 October (1995).
8. Gibson, N. Friction ignition hazards in powder handling plant. Review of existing
assessment procedures and need for further work, Health and Safety Laboratory
Report TD/00/02 (2000).
9. Powell, F. Can non-sparking tools and materials prevent gas explosions? Gaz-Eaux -
Eaux Usees. 66(6), 419-428, (1986).
10. Pedersen, G.H. and Eckhoff, R.K. Initiation of grain dust explosions by heat
generated during simple impact between solid bodies. Fire Safety Journal. 12, 153
(1987).
11. Reimer, H. Investigation of the capacity of a methane-air mixture to be ignited by
sparks struck from steel. Dissertation from TU Stuttgart (1956). HSE Translation No.
8860 (1980).
12. Gibson, N. et al. Fire hazards in chemical plant from friction sparks involving the
thermite reaction. I. Chem. E. Symposium Series No. 25 (1968: Instn Chem Engrs,
London).
13. Eckhoff, R.K. Dust explosions in the process industries. Second Edition. Butterworth
Heinemann. (1997).
14. Dahn, C.J. and Reyes, B.N. Determination of metal sparking characteristics and the
effects on explosive dust clouds. Industrial Dust Explosions, ed. Cashdollar, K.L. and
Hertzberg, M. Special Technical Publication 958, ASTM, Philadelphia, USA. pp324-
332 (1987).
15. Allen, J. and Calcote,H.F. Grain dust ignition by friction sparks. A preliminary
investigation. Final Report, Aerochem TP-403a. Aerochem Research Laboratories,
Inc Princeton, New Jersey, U.S.A (1981).
16. Muller, R. Zundfahigkeit von mechanisch erzeugten funken und heissen oberflachen
in staub-/luftgemischen. VDI-Berichte 701, VDI-Verlag GmbH, Dusseldorf, 421
(1989).
17. Siwek, R. and Cesana, C. Ignition behaviour of dusts : measuring and interpretation.
Process Safety Progress 14 (2), 107, (1995).
18. Bartknecht, W. Dust Explosions. Course, Prevention, Protection. Springer - Verlag.
(1989).
19. ISSA. Working Group 6: Rules for dust explosion protection for machines and
equipment - preventive and constructional measures. International Social Security
Agency, Mannheim (1987). Latest Edition (German only, 1998).
20. Bailey, M. Private Communication. Syngenta, Manchester UK (2000).
21. Boyle, A.R. and Llewellyn, F.J. The electrostatic ignitability of dust clouds and
powders. J. Soc. Chem. Ind. Trans. 69, 73 - 181, (1950).
22. Line, L.E., Rhodes, H.A., and Gilmer, T.E. The spark ignition of dust clouds. Journal
of Physical Chemistry. 63, 290, (1959).
23. Smielkow, G.I., and Rutkowski, J.D. Badania zjawiska zaplonu mieszanin
pylowopowietrznych wywolanego wyladonaniami iskrowymi. Chemia Stosowana
XV3, 283, (1971).
24. Matsuda, T., and Naito, M. Effects of spark discharge duration on ignition energy for
dust/air suspensions. Particulate Systems, Technol. and Fundam. (ed. J.K. Beddow)
Hemisphere Publ Corp./McGraw Hill Int. Book Co., 189, (1983).
25. Parker, S.J. Electric spark ignition of gases and dust. Ph.D. Thesis, (August) City
University, London. Dust part also published as Report No. 8 53351-4, (December)
Chr. Michelsen Institute, Bergen, Norway.
26. Ballal, B.R. Ignition and flame quenching of quiescent dust clouds of solid fuels.
Proc. Roy. Soc., Series A, London (July) (1979).
Norberg, A., Xu, D., and Zhang, D. Powder ignition energy measured utilizing a new
fluidized bed ignition chamber. Report ISSN 0349-83 52 (September), Institute for
High Voltage Research, Uppsala, Sweden, (1988).
27. Bowden, F.L. and Tabor, D. Friction and Lubrication. Methuen, London (1967).
Pinkwasser, Th. On the extinction of smouldering fires in pneumatic conveyors.
Symposium on Control of the Risks in Handling and Storage of Granular Foods,
Paris. (1985).
28. ISSA. Determination of the combustion and explosion characteristics of dusts.
International Social Security Agency, Mannheim (1998).
29. Harper, D.J. et al. Use of intrinsically safe circuits and enclosures to control ignition
risk from equipment in powder handling plant. I.Chem.E. Symposium Series No. 141
463. Hazards XIII. Process Safety – the Future (1997).
30. Zockoll, C. Ignition effect of smouldering pockets in dust-air mixtures. VDI-Berichte.
701, p295. VDI-Verlag GmbH, Dusseldorf (1989).
31. Gibson, N and Harper, D.J. Friction and localised heat initiation of powders capable
of fast decomposition – An exploratory study. I.Chem.E.
32. Bailey, M. and Walker, N. Private Communication. Syngenta, Manchester, UK.
33. Alfert, F. et al. The ignition capability of nests of smouldering material and hot
objects in industrial plants. VDI-Berichte No 701. pp 303-319 (1988).
34. Pinkwasser, Th. On the ignition capacity of free-falling smouldering fires. Euromech
Colloquium 208, Explosions in Industry. (1986).
35. Jaeger, N. Zundwirksamkeit von glimmnestern in staub/luft-gemischem. VDI-
Berichte 701, p263. VDI-Verlag, GmbH, Dusseldorf, (1989).
36. Griesche, G. and Brandt, D. Einflussfaktoren auf die zundtemperatur von staub-luft-
gemischen. Die Technik. 31, 504 (1976).
37. Eisfeld, D. Heating of coal dust by intermittent contacts in intrinsically safe circuits.
Gluckauf-Forschungshefe 2. 99-105 (1985).
38. Krause, U. and Schmidt, M. Initiation of smouldering fires in combustible bulk
materials by glowing nests and embedded hot bodies. J.Loss.Prev. Process Ind. 10
(4), 237 (1997).
39. Liesch, S.O., Kauffman, C.W. and Sichel, M. Smouldering combustion in horizontal
dust layers. Twentieth Symposium (International) on Combustion. 1601-1610. The
Combustion Institute. (1984).
40. Abbot, J. Prevention of fires and explosions in dryers – A User Guide. The
Institution of Chemical Engineers. 2nd Edition. Rugby, UK. (1990).
41. Scholl, W. Explosion protection measures in the processing of wood in the chipboard
industry. 2nd. World Seminar on the Explosion Phenomenon and on the Application
of Explosion Protection Techniques in Practice. 4-8 March 1996. Ghent, Belgium.
42. CEN TC 305. Working Group 2. Draft Standard: Non-electrical equipment intended
for use in potentially explosive atmospheres – Part 5: Protection by constructional
safety.
43. ISSA. Collection of examples. Working Group 6 : Dust Explosions – of the Machine
Safety Section of the International Social Security Agency. Mannheim.
44. Woodcock, C.R. and Mason, J.S. Bulk Solids Handling. An Introduction to the
Practice and Technology. Blackie Academic and Professional (Reprinted 1996).
45. Zalosh, R.G. Review of coal pulverzier fire and explosions incidents. Industrial Dust
Explosions. Ed. K.L. Cashdollar and M. Herzberg. ASTM Special Technical
Publication 958, p191 (1987).
46. Expert Commission for Safety in the Swiss Chemical Industry (ESCIS). Series
Safety. Milling of Combustible Solids. Safety Evaluation of the Feed Material,
Protective Measures with Mills. Booklet 5 1994.
47. TREX. A European Network on the transfer of results from standardisation in the
field of protection against explosion to industry. A one-day seminar on the ATEX
Directive. London, 29 September, 1998.
50. Torrent, J. Private Communication. LOM, Madrid, Spain. (2000).
51. Gummer, J. Private Communication. HSL, Buxton, UK. (2000).

Acknowledgements. The author wishes to thank Dr N. Gibson for reviewing the


technical content of this paper and Syngenta, Manchester, UK for permission to use their
unpublished data.
TABLE 1 : SUMMARY OF IMPACT TESTS
(Reference 9)
Group I or N Impact < 200 J Impact > 200 J
Ce, Zr, Hf, Ti on hard materials Very hard steel on very hard steel
Hand hammer blows on light metal smears on rust Mg, Ti, Al Carbon steel on carbon steel
I and Alloys on rusty steel Rotary impact high carbon steel on high
Chrome-steel on sandstone and carborundum carbon steel
Al bronze, CuBe, glass on sandstone Tungsten carbide, steel on sandstone
Group 1 Tool steel, bronze, nickel chrome steel on sandstone Hobnail on sandstone
methane Brass on corroded magnesium anode

Mild steel on sandstone Zu, Cd, Al bronzes on rusty steel


N Tungsten carbide on sandstone C Steel, CuBe. Brass on rusty steel
Mild Steel on sandstone

Carbon Steel on Carbon Steel Rotary impact carbon steel on carbon steel
I Hobnail, tungsten carbide, Al bronze, CuBe Rotary impact carbon steel on mild steel
Tool steel, Lamp glass on sandstone Mild Steel on Mild Steel (600J)
Group 11A
Propane
etc Mild Steel on rusty mild steel (180J) Rotary Impact of steel on steel
N Hard steel on hard steel (180J)
Al bronze on rusty steel (180J)

Bottle glass on sandstone


Group 11B I Tool steel on rusty mild steel
Ethylene Mild Steel on rusty mild steel
etc Al bronze on clean mild steel

N Tool steel, Mild Steel on clean mild steel


Brass, CuBe, Al bronze on rusty steel

CuBe on clean and rusty mild steel Mild steel on rusty mild steel
Al bronze on clean and rusty mild steel Rotary impact bronzes on steel
Group 11C I Silver steel on rusty mild steel CuNi Alloy on rusty steel
Hydrogen CuAl Alloy on rusty steel CuAlNi Alloys on rusty steel and hard steel
etc CuZnAl Alloy on rusty steel

N Brass on rusty mild steel Zinc Alloys on steel


Silver Steel on clean mild steel
TABLE 2 : SUMMARY OF RUBBING TESTS
(Reference 9)

Group I or N Rubbing < 10 m/s Rubbing > 10 m/s


Sandstone on Sandstone (500-750W) Steel of buffing disk (460W)
Rusty steel and light metals and alloys. Mild steel on mild steel
o
I Tungsten carbide tipped machine picks on Thermosetting plastics on materials mpt >723 C (>22 m/s)
sandstone. Sandstone, Al bronze 197, Mild Steel on Al bronze 197 disk
Mild Steel on mild steel (>3700W) AB197. AB2 on mild steel disk
Group 1 – 1%P Cast irons on mild steel CMA2. mild steel on CMA2 disk
Methane Brass on light alloy S19 (90 m/s)
- 1% P Cast irons on mild steel

N 0.5% Cr Cast iron on mild steel Mild steel on buffing disk (460W)
3%P Cast iron on mild steel Drill steel on grindstone (1000W)
Brass on sandstone CMA2 on mild steel disk (46 m/s)
70/30 Brass on mild steel (90 m/s)

Group 11A I Grinding of steels – from hot surface, not sparks


Propane etc Ti, Mg, Steels on runway materials
Ti alloy, St steel on anodised Al alloy disk

0.5% Cr Cast iron on mild steel Al alloy on runway materials (18 m/s)

Steel on steel (400W) Steels on grindstone


Group 11B I Rusty steel on copper disk (75 m/s)
Ethylene etc AB197, AB2, CMA2 on mild steel disk (46 m/s)
70/30 Brasses, Copper on mild steel disk (90 m/s)
0.5% Cr Cast iron on mild steel

N 60/40 Brasses on mild steel disk (90 m/s)


Light alloy on light alloy (140 m/s)
Non-metallic composite brake material on mild steel

Stainless steel on Al alloy and GRP disks


Al on GRP disk
I CuBe, Ni alloy, phosphor bronze on steel wheel
Copper on rusty steel wheel (70 m/s)
Group 11C Steel on copper wheel
Hydrogen Bronzes, CuBe, Copper alloys on grindstone if long enough
etc

Sparks from high alloy and chromium steel on grindstone


N CuBe and Cu alloy sparks from grindstone
Carborundum on Al alloy and GRP disks
Al on Al disk 90 m/s
Al Zn and their alloys on grindstone

You might also like