Autoignition temperatures for coal particles

dispersed in air

Martin Hertzberg
US Department of the Interior Bureau of Mines, Pittsburgh Research Center, PO Box
18070, Pittsburgh, PA, USA
(Received 7 January 1991; revised 20 March 1991)

The apparent discrepancies in the data for the ignition temperature of coal particles measured within
extended isothermal sources can be resolved, in part, by considering the marked difference in the mixture
test volumes used by the various investigators. Data obtained with systems whose mixture test volumes
are too small compared to their isothermal source volumes are not considered here because they are too
strongly overdriven to be of practical use. The consequences of ignition events in such overdriven systems
are of trivial significance for the system as a whole. Data are presented and analysed for the autoignition
temperatures of dispersed carbonaceous dusts in a 1.2 1 furnace whose test volume and source volume are
comparable, and in which an ignition event leads to a dust explosion involving drastic changes in the
composition, temperature and pressure of both the test volume and the source volume. In the limit of fine
particle sizes, the minimum autoignition temperatures for carbonaceous dust clouds in air vary from as
low as 400C for the totally volatilizable polyethylene to as high as 925C for a non-volatile graphite, with
coal values at intermediate temperatures depending on the coals rank or yield of combustible volatiles.
On the basis of the available data presented for the coals, for the oil shale dusts and for artificial mixtures
of polyethylene and graphite, it is concluded that the pyrolysis and devolatilization process is generally
rate controlling.

(Keywords: coal; ignition; temperature)

Ignition sources can be characterized in terms of the kind In addition to the type ofenergy, there are other variables
of energy they introduce into a flammable mixture or that are under the control of the observer--experimental-
substance. The ignition energy initially deposited into a ist. and some of those other variables may be more
test mixture can be electrical, chemical, mechanical, important than the initial energy type, or even the
optical or purely thermal. Regardless of that initial form, absolute magnitude of the amount of energy introduced
the dynamical processes of ignition inevitably result in into the test mixture. Those other variables involve the
complex transformations between those various energy variety of time-dependent profiles for the energy delivery
forms and the effectiveness of an ignition source is rate or the power density being injected into the test
determined by the final result of those transformations. mixture or system, and also the variety of geometric sizes
The ignition source is effective only if it successfully and shapes that are possible for the source volume
initiates a flame, an explosion or some other self- relative to the mixture test volume. The source volume
sustained combustion process within the test mixture. is the volume within which the ignition source energy is
Thus. the result of a successful ignition is the rapid confined, whereas the mixture test volume is the total
consumption of the test mixture and its transformation volume of the system containing the test mixture being
into high temperature combustion products. Ultimately, studied.
the effectiveness of the ignition source is determined by At one end of that spectrum or matrix of possibilities
whether it triggers the release of that chemical combus- are concentrated sources: ignition sources whose energy
tion energy stored in the test mixture. is concentrated spatially and temporally. The energy of
In these preliminary considerations, no distinctions such a source is concentrated within a small fraction of
need be made between test mixtures that consist of the test volume, and the energy delivery rate is almost
homogeneous gases and test mixtures that are hetero- instantaneous. Specific examples of such sources are
geneous, containing dust dispersions or liquid droplet electric sparks, focused laser beam pulses or chemical
suspensions. For purposes of these definitions they can pyrotechnics. Those concentrated sources are small in
be treated identically; however, the data to be presented volume, and are initiated within test volumes that are
and discussed consist entirely of dust dispersion ignition much larger. Since the source volume is small, the energy
data. added to the system by the ignition source is only a trivial
The form or type of energy introduced into the test fraction of the combustion energy that can be released
mixture is initially controlled by the researcher or by the successful ignition of the larger test volume. For
observer; however, once that energy is deposited, the such a source, the effectiveness of ignition appears to
complex transformations that subsequently take place be uniquely determined by a single parameter: the
between the various forms of energy, are beyond the absolute magnitude of the source energy. A minimum
researchers control. Those subsequent energy transfor- amount of energy is required in order to trigger the flame,
mations are determined by the laws of thermodynamics. explosion or other self sustained combustion process.

00162361/91/10111549
fi 1991 Butterworth-Heinemann Ltd. FUEL, 1991, Vol 70, October 1115
Autoignition temperatures for coal particles: M. Hertzberg
That threshold energy is referred to as the minimum
ignition energy. Static conditions for the flow field are
normally required to reach that minimum value, for if
the test mixture is in motion relative to the source, higher
ignition energies are required, especially if the flow is
turbulent.
At the other end of the spectrum or matrix of possible
sizes and energy delivery rates are extended sources:
ignition sources that are both spatially and temporally
extended. The spatial extension can be so large that the
ignition source volume can be comparable to that of the
test volume. Accordingly, in the extended source, the
ignition energy is diffused uniformly throughout the test
mixture. In addition, such sources are extended in time
so that the energy is continuously present and quasi
steady-state conditions are normally attained during an
ignition test. The simplest example of such a spatially
and temporally extended source is a purely thermal one:
an oven that is isothermal so that it is maintained at
some constant, elevated temperature during the test
period. The test mixture or substance completely fills the
oven so that the shape and volume of the source and the
test mixture are essentially identical. In an actual
experiment, the initially cold test mixture is rapidly
injected into the oven or furnace, and as it completely
fills it, the test mixture soon reaches the constant, elevated
temperature of the oven walls before the ignition process
starts. Thus, a quasi steady-state isothermal condition is
initially attained in the test mixture.
For such an extended thermal source, the effectiveness
of ignition appears to be uniquely determined by another
single parameter: its initial, steady-state temperature. A
minimum temperature is needed for the test mixture to
be ignited and to rapidly react to completion; the
threshold temperature is referred to as the minimum
autoignition temperature (AIT). For a spatially extended,
isothermal system, the AIT is much less sensitive to flow
conditions than the minimum ignition energy. However,
the AIT in the spatially extended system is sensitive to
the total test volume, or more precisely, to the surface
to volume ratio of the oven or test enclosure.
All of the data to be reported in this study consist of
such AIT&. They were obtained in a 1.21 furnace
volume, and were measured for dust particles dispersed
in air at a controlled dust concentration. The term cloud
ignition temperature is also used to describe such a
measurement. Although the furnace used for these
measurements was geometrically isothermal and a quasi
steady-state condition was normally reached during the
test period, the test volume and duration were neverthe-
less limited: the former to a furnace volume of 1.21, the
latter to a test period that did not exceed several seconds.
The extent to which the measured AITs may be sensitive
to the size of the test volume or to its surface to volume
ratio or to the test duration remains to be determined.
It should also be made clear that between the two
extremes described, of concentrated sources and extended
sources, there exists a multidimensional matrix of
intermediate possibilities involving various combinations
of ignition energy types, rates of energy deposition,
geometric shapes and relative volumes. As indicated in
the general survey by von Elbe6, within that matrix of
intermediate possibilities, neither the source energy nor
its temperature need be a unique determinant of the
effectiveness of ignition. That is especially the case for
the temperature parameter in a non-isothermal system,
1116 FUEL, 1991, Vol 70, October
where the temperature measured in one region of the test
volume need not be representative of the temperature
throughout. In some instances it may be neither the
source energy nor the source temperature that is
determining but rather some other parameter such as the
energy flux density flowing into the test volume over
some minimal contact area between it and the ignition
source4. Clearly, it is beyond the scope of this paper to
consider all such conceivable cases within that multi-
dimensional matrix of possible configurations and time
histories.
There is one problem that is within the scope of this
study and that is central to the resolution of a most
important issue for coal particle ignition: namely, the
glaring apparent discrepancy in the data for ignition
temperatures reported with extended isothermal sources.
The data for coal particle ignition temperatures reported
by Essenhigh and co-workers-* differ markedly from the
measurements reported by Hertzberg and co-workerszP5.
As indicated in a recent review of the subject, the
Essenhigh data also contradict the traditional body of
literature on the subject of coal dust cloud ignition
temperatures that can be traced to the early development
of the Godbert-Greenwald furnace. The Godbert-
Greenwald furnace was the precursor of the 1.2 1 furnace
more recently developed by Liebman2 and used by
Hertzberg and co-workers in their studies3-5. The
Essenhigh data give much lower ignition temperatures
than those reported by Hertzberg and co-workers. The
particle size dependence reported by the Essenhigh group
is precisely the opposite of that reported by the Hertzberg
group: coarser particles have lower ignition temperatures
according to Essenhigh and co-workers, whereas Hertz-
berg and co-workers report that coarse particles have
much higher ignition temperatures. Finally, the ignition
temperatures reported by Essenhigh and co-workers are
relatively insensitive to coal rank, whereas Hertzberg and
co-workers report a marked sensitivity to coal rank.
These apparent contradictions are not limited to these
two groups. The general tendencies for the particle size
dependence of the Essenhigh measurements are mirrored
in the data reported independently by Karcz and the
magnitude of the ignition temperatures reported by the
Hertzberg group are confirmed by measurements re-
ported by Hattwig using the BAM (Bundesanstalt fur
Materialprufung) furnace .
Recent studies by Wall et ~1.~ strongly support the
data reported by the Hertzberg group. They also show
that the appearance of a visible flash, which was the
ignition criterion used by Essenhigh, is of trivial signifi-
cance for the overall particle combustion process. It is
thus essential, as a matter of definition, to clearly
distinguish between the two physical systems used to
obtain the data, and the two criteria used to define the
ignition event in those two systems. The drop furnace
systems .*.l involve the exposure of single particles (or
small arrays of particles) to elevated temperatures within
a relatively large furnace volume. Although such a drop
furnace in which single particle ignition is studied is also
an extended thermal source, the dimensions and mass of
that ignition source are much larger than the dimensions
or mass of the test mixture. The ignition source
dimensions of the drop furnace involve the large volume
of hot gas contained within the furnace, and also the
large internal, radiant surface area of the furnace walls.
By contrast, the test mixture volume in the drop furnace
 Autoignition temperatures for coal particles: M. Hertzberg

experiment is only the single particle itself plus the small obtained for oven temperature profiles and coal dust
air volume surrounding the particle. Thus, in drop dispersion uniformity have been published elsewhere2. As
furnace experiments, the ignition source dimensions are the dust is dispersed, there is a characteristic time
much larger than the test mixture volume. The single constant for heating the dust-air mixture. That time
particle test mixture volume and mass are trivial in constant is generally much shorter than the time constant
comparison to the ignition source volume and its mass. for ignition, especially when the system is near the AIT
As a result, for the single particle/drop furnace experi- boundary. Ignition delay times vary from several seconds
ment, the energy available from the combustion process near the AIT boundary, to several milliseconds at oven
whose ignition is being studied, is utterly trivial temperatures that markedly exceed the minimum AIT.
compared to the energy of the ignition source that is In the former case, a steady-state isothermal condition
driving the process. The ignition source is no longer the is attained within the test sample volume prior to ignition.
trigger that is controlling the cannon, instead their In the latter case, ignition occurs before the dispersed
roles have been reversed. dust-air mixture can reach the steady-state temperature
As a matter of definition, meaningful ignition studies but such early ignition does not affect the observers
should involve large amounts of combustion energy ability to locate the AIT boundary. For data obtained
available in the test mixture volumes that are to be near the AIT boundary, the system inevitably attains a
triggered by relatively small amounts of driving energy quasi steady-state isothermal condition at the initial oven
in the ignition sources. By contrast, in the drop furnace temperature before the ignition event occurs.
experiments with single particles, the combustion energy Autoignition of the test volume leads to an explosion
available in the test mixture is trivial compared to the which ruptures the fibreglass diaphragm at an over-
huge amount of driving energy in the ignition source. In pressure of 0.1 x lo-0.3 x 10 Pa. The flame that drives
the terminology used for flammability limit measure- the explosion generates a burned gas volume that fills
ments, the single particle/drop furnace experiments are the oven at a temperature that is at least IOOOC above
enormously overdriven. They are overdriven because the initial temperature. The criterion for ignition is the
the consequences of an ignition and subsequent combus- rupture of the diaphragm with the simultaneous emission
tion of the test mixture are of trivial significance for the of flame from the top of the furnace. Data obtained for
system as a whole. In the experiment reported in drop a variety of dusts have been presented elsewhere3~ .
furnaces, ignition of the small, single particle in the In order to evaluate the overall accuracy and reliability
enormous furnace volume leads to a tiny flash of light, of the system, it was used to measure the AITs of
but for the total system as a whole, the average homogeneous gas mixtures that had already been studied
temperature remains essentially the same, and the final extensively by other investigators: hydrogenair and
average composition in the furnace is also essentially methaneeair. The data for those homogeneous mixtures
unchanged. compared favourably with the independent measure-
By contrast, for the autoignition furnace systems2.3,5,. 1 ments of those other investigators5. The agreement for
the source volume and test volume are of comparable homogeneous gases confirms the reliability and accuracy
mass and dimensions, and the consequences of an ignition of the 1.2 1 experimental system.
event are drastic: the ignition leads to an explosion
which raises the temperature of both the source and test
DATA FOR COALS AND OTHER DUSTS
volumes by at least IOOOC, and radically alters the
systems average composition. In all the data to be The experimental results for four dusts are shown in
reported in this study, the criterion for ignition is the
occurrence of such an explosion throughout the system Furnace collar and ,Rupture dlaphrogm
as a whole.
The issue of how to relate the data reported here for
the AITs of dust clouds to the measurements of single
particle ignition in drop furnaces, has been addressed
by Wall et al. and will not be considered here. In
order to avoid confusion, however, and for purposes of
clarity, the two experimental systems must be clearly
distinguished from one another.

EXPERIMENTAL
A schematic of the 1.2 1 ceramic furnace used for AIT
measurements is shown in Figure 1. The furnace volume
is essentially cylindrical in shape with resistance heating
elements wound on its outside surface in a pattern
designed to give a uniform furnace temperature. Dust
particles to be tested are placed in the disperser which
is inserted into the bottom of the furnace only moments
before the ignition test sequence of events is started. A
fibreglass diaphragm covers the top of the furnace and -Supporl rods
isolates the 1.2 1 test volume. A pulse of air that lasts for
30ms disperses the dust into the furnace as it injects
-65 cm3 of air from a pressurized reservoir. Data Figure I Schematic of the I.2 1ceramic furnace for AIT measurements

FUEL, 1991, Vol 70, October 1117

Autoignition temperatures for coal particles: M. Hertzberg

Figure 2, which is a plot of temperature versus concentra- of temperature versus dust concentration (at ambient
tion of measurements which gave ignitions and those that pressure, for the dust particle sizes studied). The AIT
did not give ignitions. The autoignition boundaries are boundary lines separate ignitions that are all above and
drawn as solid lines in the thermodynamic state-space to the right of the line from non-ignitions that are below
and to the left of the line. The anthracite coal dust which
has the lowest proximate volatility (8%) of all the dusts
shown, is the most difficult to ignite with a minimum
AIT of 760C. The Pittsburgh seam bituminous coal with
a 37% proximate volatility has a minimum AIT of
54OG56OC. Polyethylene dust with a 100% volatility has
O t Anthracite Coal a relatively low minimum AIT of 400C. Sulphur dust,
6OOL I I Al I I I
which is also totally volatilizable, is the most easily
I I v, I I I -
ignited of the dusts shown in Fiyure 2, with a minimum
Pittsburgh Bituminous Coal AIT of 280C.
. . - The solid boundary lines or AIT curves all have similar
shapes. At low dust concentrations the ignition tempera-
tures are high, but as the dust concentrations increase,
the AITs all decrease markedly. As concentrations
increase further, the AITs eventually level off to minimum
values that become insensitive to concentration at high
dust concentrations. The concentration at which that
levelling off occurs is also a function of volatility: the
more volatile dusts begin to level off at relatively low
concentrations, whereas for the less volatile dusts, the
AIT curves do not level off until high concentrations are
500 l
Sulfur attained. For a given dust, the concentration at which
H its minimum ATT is attained appears to coincide with
400
. that concentration that will generate a near stoichio-
300 - B B metric concentration of volatiles.
8 B
I I I Ehlr I presents a summary of the minimum AIT
200
0 100 200 300 400 1,200 2,000 measurements with emphasis mainly on the coals. The
DUST CONCENTRATION, g/m3 tabulation is presented in order of decreasing AIT and
includes graphite, the various ranks of coal, polyethylene
Figure 2 Autoignition data from the I.2 1 furnace plotted as tempera- and several other carbonaceous dusts. The data for
turc versus concentration: 0. ignitions; 0, non-ignitions

Table 1 Minimum AITs for various dusts in air as measured in the 1.2 1 furnace

Proximate Minimum
volatility Particle diameter AIT

Dust type (%I Q (pm) D, (pm) (C)

Graphite <1 28 43 925

Anthracite coals
Reading 5 6 9 780
Pennsylvania 8 21 42 760
Bituminous coals
Pocahontas 16 16 59 635
Sewell 29 22 44 560
Pittsburgh 37 28 48 540
Subbituminous coals
Western (10% H,O) 35 25 71 475
Western (dried fully) 39 25 71 450
Lignite coals
North Dakota (26% H,O) 33 43 115 600
North Dakota (dried fully) 43 43 115 555
Beulah, ND (27% H,O) 30 15 29 440
Beulah, ND (dried to 8% H,O) 41 15 29 425

Gilsonite 85 20 50 480
Lycopodium 85 27 28 435
Polyethylene 100 27 37 400
Decane (liquid) 100 NA NA 275

NA, not applicable

1118 FUEL, 1991, Vol 70, October

 Autoignition temperatures for coal particles: M. Hertzberg

decane are also shown as an example of a totally contained in a crucible within an oven whose temperature
vaporizable liquid hydrocarbon that devolatilizes more is maintained at 900C. An inert atmosphere is main-
rapidly than any of the dusts. The second column in tained to avoid oxidation, and the final weight loss of
Table 1 lists the dusts proximate volatility as measured the sample is its proximate volatile yield. That same
in the ASTM D3175 test. All volatilities in Table 1 are procedure when used for gilsonite, lycopodium, poly-
given on a dried basis so that the moisture in the ethylene and decane, gives their volatility yields as listed
as-received dust is not included in the volatility value in Table 2. For oil shale dusts, whose autoignition data
shown. Particle diameters are given as surface-area- will be presented and discussed shortly, a lower tempera-
weighted means, D,, and as mass or volume-weighted ture of 500C must be used in the Fischer assay method
means, 0,. for estimating the volatility or oil content of the shale.
An inspection of the minimum AIT data shown in The lower temperature is needed to avoid the decomposi-
Tcrhle 1 further reinforces the trend seen for data in tion or calcining of the mineral carbonate that comprises
Figuw 2. The major factor controlling the minimum AIT the bulk of the shale. Despite the widespread acceptance
for a fuel dust is its yield of combustible volatiles during of these test methods for the characterization of these
pyrolysis. The coals with a high volatility have a low natural products and the widespread commercial USCof
minimum AIT; the coals with a low volatility have a high those standard test methods. it has long been known that
minimum AIT. That inverse correlation is most simply the yield of combustible volatiles, especially from the
explained by a thermal autoignition process that is under coals, is sensitive to the effective heating flux that drives
pyrolysis and devolatilization rate control. the devolatilization process, and to the temperature
attained by the coal particles, and to their particle size,
and to other boundary conditions and constraintsT I,
ANALYSIS AND INTERPRETATION
The processes involved in coal particle pyrolysis and
The same processes that govern flame propagation in devolatilization are exceedingly complex. involving both
a dust-air mixture3 govern the autoignition process in heat and mass transport limitations, as well as hysteresis
the 1.2 1 furnace. Those processes involve the following effects, and the detailed kinetic mechanisms involved are
sequence: first, the heating and devoiatilization of the the subject of considerable controversy . Neverthe-
dust particles; second. the mixing of the emitted volatiles less, it has been established rather conclusively by
in the air space between particles; and third, the virtually all investigators that the yield of volatiles from
exothermic combustion of the resultant fuel-air mixture. pulverized coal particles under flame heating conditions
For dust flame propagation, the third process generally substantially exceeds that obtained from the standard
occurs homogeneously in the gas phase13. For the ASTM proximate volatility method. The precise compo-
exothermic combustion reactions that lead to auto- sition of these voiatiles from a given coal is also sensitive
ignition in the 1.2 1 furnace, it remains to be determined to heating rate, temperature and transport limitations;
whether they start in the gas phase or whether they start however, there is no substantive evidence that those
heterogeneously on the walls of the furnace or on the composition variables have any significant intlucncc on
surfaces of the char residues. In any case, the data in the combustion behaviour of the given coal dust.~.
Table 2 suggest that it is the first process, the pyrolysis The energy content per gram of combustible volatilcs
and devolatilization of the dust particles, that is rate and their burning velocity is not sensitive to their prccisc
controlling during the thermal autoignition of the dust composition or molecular weight.
clouds. It is that same rate controlling process that is A comparison of the measured AITs for the various
responsible for the levelling off of the AIT curves at high dusts with their volatility yields shows an invcrsc
dust concentrations (Figure 2), and the essential absence correlation. That inverse correlation can be explained in
of the normal. rich limit behaviour for dust clouds. terms of the dusts onset temperature for significant
Neither the AIT curves shown in Figure 2 nor the pyrolysis and devolatilization. Coals with high volatile
pressures measured during dust explosions show the yields have low onset temperatures for pyrolysis and
normal rich limit behaviour that is displayed by homo- devolatilization. whereas coals with low volatile yields
geneous gas mixtures of hydrocarbon fuels in air. For have high onset temperatures. That invcrsc correlation
the dusts, autoignition occurs as soon as a near is even more dramatically illustrated by the data for
stoichiomctric concentration of volatiles is generated by graphite. polyethylene and decane. Graphite has the
the pyrolysis and devolatilization process, and the highest AIT and the lowest volatility yield. Graphites
presence of a large excess of solid fuel does not strongly low volatile yield correlates with its highest onset
influence the autoignition process so long as the excess temperature for devolatilization, and hence the graphites
fuel is not present in the vapour phase. For homogeneous AIT is the highest. Polyethylene has the lowest AIT of
gas mixtures where the excess fuel is present in the vapour all the carbonaceous dusts and the highest volatility yield.
phase, the influence of the excess fuel is strong and the Polyethylenes high volatility correlates with its low onset
result is the normal rich limit behaviour. temperature for devolatilization, and hence its AIT is the
It is beyond the scope of this paper to consider the lowest of the carbonaceous dusts. Decant is a iiquid and
detailed kinetic mechanism of coal particle pyrolysis and its AIT is cvcn lower than that of polycthylcnc. Dccanc
devolatilization, and it is not possible to discuss the is also totally volatilizable but its onscl tempcraturc for
complexities involved in determining the final yield of vaporization is even lower than that of polyethylene.
volatiles and their precise composition. Nevertheless, it Hence it has a lower minimum AIT than polyethylcnc.
is important to appreciate the complexities of those Thus, as indicated earlier, it is the particle pyrolysis
processes before proceeding further. For coals, the and devolatilization process that appears to be rate
standard test method for determining the proximate controlling during the thermal autoignition of the
volatility values that are listed in Tub/e I, involves the dispersed dust cloud. The cxothermic reactions between
pyrolysis and devolatilization of a macroscopic sample the hydrocarbon volatiles and the oxygen of the air in

FUEL, 1991, Vol 70, October 1119

Autoignition temperatures for coal particles: M. Hertzberg
the furnace are clearly not rate controlling. The data for
the minimum AITs of heavy hydrocarbon vapours homo-
geneously mixed in air, show that they autoignite at low
temperatures, namely at 200-260C4. Those low temper-
atures are consistent with the decane data in Table 1,
which autoignites at only 275C. All the carbonaceous
dusts, and especially the coals, have much higher
minimum AITs because much higher temperatures are
needed for them to pyrolyse and devolatilize at a rapid
enough rate to generate the volatiles needed to sustain
the exothermic autoignition process. The exothermic
autoignition reactions themselves are fast enough to
proceed at a runaway rate at much lower temperatures.
Clearly, it is that rate of pyrolysis and devolatilization
that is rate controlling for the dusts in the autoignition
furnace.
A detailed consideration of the mechanism of particle
pyrolysis and devolatilization is beyond the scope of this
discussion. Independent measurements reported else-
where have shown that the process of coal particle
devolatilization is heat transport controlled and that
it proceeds in the form of a non-isothermal pyrolysis
wave front that propagates inward from the particle
surface9,2,24,29. For the furnace exposure of the dis-
persed dust cloud, the heat flux that is driving the devola-
tilization wave front into the dust particles is the sum of
conductive and radiative fluxes26. The conductive flux is
from the high temperature gas in the 1.2 1gas furnace; the
radiative flux is mainly from the furnace walls.
Coal dust particles with relatively high H/C ratios will
pyrolyse and devolatilize rapidly at relatively low
devolatilization wave temperatures and relatively low
heating fluxes. Those low heating fluxes are attained at
relatively low furnace temperatures. At those lower
temperatures, the liquid metaplasts and hydrocarbon
fragments generated by the depolymerization process
occurring within the devolatilization wave front, are able
to vaporize rapidly. The volatiles reach the surface
rapidly without exposure to too high a temperature, and
they are emitted with minimal char formation. On the
other hand, coal dust particles with low H/C ratios will
pyrolyse and devolatilize more slowly, and require higher
heating fluxes. Those high heating fluxes are attained
only at high furnace temperatures. At those higher
temperatures, the vaporization process for the liquid
metaplasts and hydrocarbon fragments must now
compete with parallel condensation reactions that lead
to char formation. The lower the H/C ratio of the coal
(i.e. the higher its rank), the higher is the temperature
required to initiate the pyrolysis and devolatilization
process, the more aromatic is the structure of the
intermediate metaplasts, the greater are the rates of the
competing char-forming, condensation reactions, the less
likely it is that the intermediates can reach the surface
as volatiles, and the higher is the AIT that is required to
generate a given concentration of volatiles in the gas
phase. In any case, as discussed earlier, whatever the H/C
ratio may be, whether it is as high as 2.0 (polyethylene)
or as low as near zero (graphite), temperatures of at least
400C are required for the onset of significant pyrolysis,
and for the coals the required temperatures are con-
siderably higher. Those temperatures markedly exceed
the 20s275C temperatures required for the exothermic
gas-phase combustion reactions to proceed at a runaway
rate. Accordingly, the autoignition process is rate
1120 FUEL, 1991, Vol 70, October
controlled by the particle pyrolysis and devolatilization
process.
OIL SHALE DUST
There is a glaring exception to the inverse correlation
between volatile yield and AIT, and that exception is
shown by the oil shale dust data in Table 2. The exception
actually supports the rule for it further reinforces the
above arguments, The data for the Green River oil
shale dusts with volatilities that vary in the range of
9-22%, show AITs that are all significantly lower than
those measured for coals with comparable volatiles.
Clearly, the kerogen matrix in the oil shale pyrolyses
and devolatilizes at a lower temperature than the various
coal matrices. But the oil shale case differs significantly
from the coals in that the oil shale AITs are essentially
independent of grade or volatility. Their AITs remain
essentially invariant at -500C even as the oil content
of shale dust more than doubles from 83-209 ml kg-.
(corresponding to volatilities that increase from 9 to
22%). A comparable increase in volatility for the coals
would have resulted in a 200C decrease in AIT. The
effect of increasing the oil content (i.e. volatility) for
the shale dust is merely to decrease the dust concentra-
tion at which that minimum AIT is reached but the value
of the minimum temperature itself remains essentially
constant. For example, the richest shale shown in
Table 2 (22% volatility) reaches its minimum AIT
at a dust concentration of = 300gmP3, whereas the
leanest shale (9% volatility) requires a concentration of
- 800 g me3 to reach essentially that same minimum AIT.
The oil shale dusts have a common ATT because they
have a common kerogen matrix, whereas the coals of
differing rank have different AITs because they have
different matrices or structures for each rank. For the
coal dusts, the variations in the yield of combustible
volatiles with varying coal rank are associated with
essential differences in their chemical structure as re-
flected in their changing H/C ratios. For the oil shale
dusts, the variations in volatility yield are associated only
with the fraction of the dust mass that consists of kerogen.
The remainder of the shale dust is the inert mineral
matrix in which the kerogen is imbedded. For the shales,
the essential structure of that kerogen matrix remains
unchanged as one progresses to the richer shales. For
the coals, however, the essential chemical structure
changes markedly as one progresses to the more volatile
coals.
Thus for the shales, the temperature required to obtain
a sufficiently rapid rate of devolatilization in order to
sustained the exothermic autoignition process, remains
Table 2 Minimum AITs for oil shale dusts as measured in the 1.21
furnace
Minimum
Oil shale Volatility Particle diameter AIT
(ml kg-) W) R (Irm) 9 (pm) (C)
209 22 18 51 475
183 20 24 51 500
146 16 22 61 500
100 11 22 88 500
83 9 29 49 500
 Autoignition temperatures for coal particles: M. Hertzberg

essentially constant for the various grades of shale, as finally reflects the high AIT (925C) of its graphite
seen in Tub/e 2. All that does change for the shales is the content. For the entire volatility range between 5% and
dust concentration that is required to generate a lOO%, the polyethyleneegraphite mixture displays an
stoichiometric concentration of volatiles. Naturally, the AIT behaviour that is dominated by polyethylenes low
leaner the shale, the higher the required dust concentra- onset temperature for pyrolysis and devolatilization. The
tion. The small increase in AIT from 475 to 500C reason for the nearly invariant low ATT is the fact that
between the richest shale and the leanest shale dust simply the volatile component has an invariant structure with
reflects the fact that the leaner shale with the smaller a constant HC ratio of 2.0 regardless of the mass of
kerogen fraction must carry along more excess baggage added graphite that dilutes it. In going from pure
in the form of a larger mass of inert, mineral component polyethylene (100% volatility) to a 1:3 polyethylene-
in order for it to generate a stoichiometric concentration graphite mixture (25% volatility). the AIT of the mixture
of volatiles. increases only 50 C, and that relatively small increase
merely reflects the fact that the lower volatility mixture
must carry along much more excess baggage of inert
DATA AND INTERPRETATION FOR
graphite in order to be able to generate a near stoichio-
POLYETHYLENE--GRAPHITE MIXTURES
metric mixture of volatiles. By contrast again, for the
A behaviour quite similar to that of the kerogen of the coals, the decrease in volatile content associated with
shales is displayed by the polyethylene content of artificial their varying rank. is associated with changes in their
mixtures of graphite and polyethylene. Those mixtures intrinsic chemical structure, and to a decrease in the H/C
were initially used for flammability limit measurements. ratio of the pyrolysing system. Thus for the coals, the
As indicated earlier, there is considerable uncertainty in onset temperature for pyrolysis and devolatilization
the true volatility yields of the various ranks of coal under increases markedly as the volatile content decreases,
the high flux heating rates that characterize pulverized which is reflected in a more marked increase in AIT. The
coal fired burners or coal dust explosions. Such un- corresponding AIT for the IvB coal with 25% volatility
certainties in volatile yield can be completely removed is 680 C, which is now an increase of ZXOC over the
by using artificial mixtures of polyethylene and graphite value for pure polyethylene.
in order to simulate coals whose volatile contents would For the coal systems depicted in Figure 3, if attention
be precisely known l3 The graphite, whose volatility is is focused on the mid range of the AIT curve; that is,
< 1%. is essentially inert and non-explosive on the rapid between gilsonite and the anthracites, the behaviour may
time scale required for flame propagation. The graphite be considered to be ideal, namely the AIT curve for the
simulates the char residue of the devolatilized coal various compositions is a straight line between the two
particle. The polyethylene component of the artificial end points. The line starts at 400 C for the high volatilities
mixture is totally volatilizable, and the combustion with H/C ratios near 2.0, and ends on the left near XOOC,
behaviour of its pyrolysis volatiles is virtually identical for the low volatility anthracites with their low H/C
to that of homogeneous gas mixtures of saturated ratios. If one were to correct the coal line slope for the
hydrocarbon fuels in air. The polyethylene component excess baggage effect associated with the inert char (and
thus simulates the volatile fraction of the coal that is ash) contents of the dusts, the slope would diminish but
combustible. For the artificial mixtures of polyethylene there would still be a marked negative slope to the AIT
and graphite, the volatility is precisely known, and is line. The corrected AIT for the coals would have an
essentially identical to its polyethylene content. intercept on the left of 725C, which is still 325 C hotter
The minimum AITs of those mixtures were also studied than its 400 C intercept on the right.
and the data are shown in Figure 3. The minimum AITs By contrast, a similar correction for the excess
for the artificial mixtures are plotted as a function of baggageeffect associated with the inert graphite contents
polyethylene content (volatility) and that data is con- of the polyethylene-graphite mixtures would result in a
trasted with the minimum AITs for the various ranks of perfectly horizontal line with essentially zero slope at a
coal and gilsonite of equivalent volatility. For the coals, constant AIT of 400C for all mixture volatilities that
the volatilities used in the plot are those measured (or are finite. The physical meaning of such a horizontal line
estimated) at high heating fluxes for fine particles. The
coals are indicated by: A, for anthracite; IvB, for low
volatile bituminous (Pocahontas seam); and hvB, for
high volatile bituminous (Pittsburgh seam). The gilsonite
is indicated by G. The data for the artificial mixtures of
polyethylene and graphite show clearly that their mini-
Polyethylene- grophlte
mum AITs are controlled mainly by the low onset - mlxtreS

temperature for the pyrolysis and devolatilization of

polyethylene, even when the polyethylene content is
diminished markedly by mixing it with much larger
amounts of graphite. For example, in going from pure
polyethylene (100% volatility) to mixture that contains
three times as much graphite as polyethylene (25%
volatility), the AIT has increased only 50C from 400 to
450C. A comparable change in volatility for the coals
results in a 250C increase in AIT, from 400 to 650C.
For the polyethylene-graphite mixtures, it is not until Figure 3 Minimum AITs for artificial mixtures of polyethylene and
the graphite content exceeds 95% (< 5% volatility) that graphite of varying volatility. They are compared with the minimum
the AIT begins to increase markedly, and the mixture AITs for coals and gilsonite of comparable volatility

FUEL, 1991, Vol 70, October 1121

Autoignition temperatures for coal particles: M. Hertzberg

I,1,1,, ,~,1,~,, ,I,/

dusts whose characteristic diameters are 50pm, the
KEY
minimum AIT values listed (measured for 2040,~m
e.- Pocahontos Coal
- Pittsburgh Cool diameter dusts) are entirely representative of the values
*- Polyethylene obtained for the finer dusts. However, for the low vola-
tility dusts, such as anthracite, graphite or Pocahontas
seam coal dust, the AIT values listed in Tub/e 1 may not
be representative of the values that would be observed
for the finer dusts. Thus, for example, the Pocahontas
coal dust listed in Table I with a DS= 16pm had a
minimum AIT of 635C. But, as the data in Figure 4
show, that value is not representative of the AIT for
Pocahontas coal dust with O,= 3 pm, which has a
significantly lower AIT of 550C.
The data in Figure 4 also show that in the limit of fine
particle sizes below 5 pm in diameter, the minimum AIT
for the low-volatile, Pocahontas bituminous coal dust

3oo/I 2
MEAN PARTICLE
40 60 80 100 200
DIAMETER, DS, /an
400 600
actually approaches the same value measured for the high
volatile, Pittsburgh bituminous coal.
Essentially the same type of behaviour is displayed by
Figure 4 Particle size dependence for the minimum AIT for Poca- the lean limit concentrations for the two coals as that
hontas (I vB) coal, Pittsburgh (hvB) coal and polyethylene displayed by their AITs. The Pocahontas coal has a
nominal or proximate volatility that is about half the
value for Pittsburgh coal (Table I ), and for 4&50pm
with an invariant AIT of 4OOC, is again that the AIT of
diameter particles, its behaviour is normal. Its behaviour
the graphite-polyethylene mixtures is controlled entirely
for that size is normal in the sense that the lean limit
by the low onset temperature for the pyrolysis and
concentration for that size of Pocahontas coal is about
devolatilization of its polyethylene fraction, no matter
twice as high as the lean limit concentration for the
how small that fraction. For the corrected AIT line for
Pittsburgh seam coal in that same size range2. However,
the polyethylene-graphite mixtures, it is not until the
as particle sizes diminish below 50pm, the lean limit
pure graphite axis is actually reached that a discon-
concentration for the Pocahontas coal continues to
tinuous vertical step function appears and the line rises
decline, whereas the lean limit for Pittsburgh seam coal
vertically to the pure graphite value of 925C.
dust remains essentially constant. As a result, in the limit
of fine particle sizes below b 5 pm, the lean limit for the
Pocahontas coal approaches the value measured for the
PARTICLE SIZE DEPENDENCIES
Pittsburgh coal. Essentially the same type of behaviour
Data for the particle size dependence of the AITs also is also observed for the two types of coal if the particle
support these general viewpoints. The data are given in sizes are kept constant but the oxygen concentration in
Fiyure 4 for three of the dusts: polyethylene, Pittsburgh the dispersing air is increased. For both the high volatile
seam coal and Pocahontas seam coal. The data are for and the low volatile bituminous coals, the lean limit
narrowly classified particle sizes whose distributions have concentrations decrease with increasing oxygen content;
been reported earlier3. The data for polyethylene and however, the value for the Pocahontas coal decreases at
Pittsburgh seam coal dust reveal the existence of a faster rate than the value for Pittsburgh coal, until, for
characteristic diameters below which the minimum AITs enriched oxygen contents of -4O%, the two lean limit
become size invariant. That behaviour for the AIT data concentrations become essentially identica12. There are
in Fiyure 4 parallels the existence of similar characteristic thus three cases of analogous behaviour in which the
diameters for the lean limits of flammability for those explosibility or ignitability of the Pocahontas coal
same dusts3. In both cases, flammability limits or AITs, approaches that of the Pittsburgh coal in the limit of
the curves are a manifestation of processes that are under small particle diameters and/or increased heating fluxes.
devolatilization rate contro13.24. The higher volatility In all three instances the parallel behaviour can be
dusts tend to have larger characteristic diameters, whereas attributed to an increase in the real volatile yield of the
the lower volatility dusts have smaller characteristic Pocahontas coal until it finally approaches the volatile
diameters. This is clearly seen for the data in Fiyure 4 yield of the Pittsburgh coal in that limit of higher heating
and also for the lean limit data published elsewhere24.27. fluxes and finer particle sizes.
For polyethylene the characteristic diameters are near Preliminary confirmation of that expected increase in
IOOpm; for the high volatile bituminous coal it is near real volatility for the nominally low volatile Pocahontas
50pm; whereas for the low volatile bituminous coal, it is coal has been recently obtained with a laser pyrolysis
well below 20,~m. system 19.21.29. The volatility yield for the Pocahontas
A comparison of such data as those shown in Fiyure 4 coal was observed to increase from its proximate value of
with the data for the various dusts listed in Tub/e I, 16% to a value of 34% for IOOpm particles heated in a
reveals that the particle size variable may still be a hidden laser flux of 200 W cm 2. For Pittsburgh seam bitu-
variable for some of the dusts studied. For some dusts, minous coal particles, with comparable particle diameters
the AIT values listed in Table I are representative of the and high laser fluxes, the measured volatility yields
values expected for the finer sizes but for others the remained essentially constant at its limiting value of
tabulated values may be unrepresentative because of the 50-55% as laser fluxes increased and particle sizes
hidden particle size dependence. For the high volatility decreased.

1122 FUEL, 1991, Vol 70, October

 Autoignition temperatures for coal particles: M. Hertzberg

CONCLUSIONS 6 von Elbe, G. Fourth International Symposium on Combustion.

The Combustion Institute, 1952, p. 13
In the limit of fine particle sizes, the minimum AITs for 7 Chen, R. R., Fan, L. S. and Essenhigh, R. H. Twentieth
carbonaceous dust clouds dispersed in air vary from as International Symposium on Combustion. The Combustion
low as 400C for the totally volatilizable polyethylene to Institute, 1984, p. 1513
8 Essenhigh. R. H., Misra, M. K. and Shaw, D. W. Comhust.
as high as 925C for the non-volatile graphite. For the Flame 1989, 77, 3
coals, ignition temperatures fall in an intermediate range 9 Godbert, A. L. and Greenwald, H. P. Laboratory Studies of
between 425 C and 780-C depending on the coals rank the Inflammability of Coal Dusts, Bureau of Mines Bulletin
and its combustible volatile yield. The data show that 389. Washington DC, 1935
10 Karcz, H., Kordlylewski, W. and Rybak. W. Fuel 1980,59,799
the major factor determining the AIT of a carbonaceous
11 Hattwig, M. in International Symposium on Explosion Hazard
dust is its combustible volatile yield, which for a coal, Classification of Vapors, Gases, and Dusts, National Material
correlates inversely with its onset temperature for Advisory Board National Research Council. Washington. DC,
pyrolysis. The pyrolysis process generates combustible 1986, Ch. 9B
tars and gases, whose subsequent reaction with the 12 Wall, T. F., Gurvrajan, V. S.. Lucas, J. et al. Twenty-third
International Symposium on Combustion. The Combustion
oxygen of the air leads to thermal autoignition of the
Institute, Pittsburgh, 1990. p. 1177
system. As particle diameters increase from the finest 13 Cashdollar, K. L., Hertzberg, M. and Zlochower. I. A.
sizes, the cloud AITs tend to remain constant initially, Twenty-second International Symposium on Combustion, The
but eventually as the particles become very coarse, AlTs Combustion Institute, Pittsburgh, 1988. p. 1757
increase markedly. On the basis of the available data for 14 Zabetakis. M. Flammability Characteristics of Combustible
Gases and Vapors. Bureau of Mines Bulletin 627, 1965. pp.
the coals, for the oil shale dusts and for the artificial 32-44
mixtures of polyethylene and graphite, it is concluded 15 D3172, Proximate Analysis of Coal and Coke, and D3175,
that the thermal autoignition process for the carbon- Volatile Matter in the Analysis Sample of Coal and Cotie in
aceous dusts is controlled mainly by the rate of pyrolysis Annual Book of ASTM Standards, American Society for
Testing and Materials, Philadelphia, 1987, p. 380
and devolatiliL.ation of the particles within the dispersed
16 Stanfield, K. E. and Frost, I. C. Method of Assaying Oil Shale
cloud. by a Modified Fischer Report, Bureau of Mines Report of
Investigations 4477, 1949
17 Howard, J. B. in Chemistry of Coal Utilization (Ed. M. A.
Elliott),2nd Suppl.Vol.,Ch. 12, Wiley,New York, 1981,p.665
ACKNOWLEDGEMENTS 18 Seery, D. J., Freihaut, J. D., Proscia, N. M. et al. Kinetics of
Coal Pyrolysis, Final Report 955966-F, United Technology
The author wishes to express his appreciation to Yael Research Center under DOE Contract Yo. DE-AC22-84PC
Miron for her constructive review of this manuscript. He 70768, July 1989
is also indebted to Ronald S. Conti, Kenneth L. 19 Hertzberg, M. and Ng, D. L. in Fundamentals of the Physical
Chemistry of Pulverized Coal Combustion. NATO AS1 Series
Cashdollar, Isaac A. Zlochower and Richard Thomas for
E, Martinos Nijhoff, Dordrecht, 1987, p. 104
their contributions to various phases of the work 20 Gibbons-Matham, J. R. Phll Thesis University of London, 1988
described in this paper. He is also indebted to Dr Alan 21 Hertzberg, M.. Zlochower, I. A. and Edwards, J. C. Coal
W. Scaroni for his invitation to submit this paper as a Particle Pyrolysis Mechanisms and Temperatures, Bureau of
debate in print. Mines Report of Investigations 9 169, 1988
22 Niksa, S. Twenty-second International Symposium on Combus-
tion, The Combustion Institute, Pittsburgh, 1988, p. 105
23 Phuoc, T. X. and Maloney, D. J. Twenty-second International
REFERENCES Symposium on Combustion. The Combustion Institute. Pitts-
burgh. 1988, p. 125
Frank-Kamenetskii, D. A. Diffusion and Heat Exchange in 24 Hertzberg, M. and Zlochower, I. A. Sixth Annual International
Chemical Kinetics, Academy of Sciences USSR, Moscow, 1947. Pittsburgh Coal Conference. Vol. I. University of Pittsburgh,
English Translation, Princeton University Press, Princeton, Pittsburgh, 1989. p. I I
1955; Kuchta, J. M. Bureau of Mines Bulletin 680. 1985. . on. rr 25 Hertzberg, M., Zlochower, I. A. and Cashdollar, K. L.
9-l 2, 33-37 Twenty-first International Symposium on Combustion, The
Conti, R. S., Cashdollar, K. L., Hertzberg, M. et al. Thermal Combustion Institute, Pittsburgh, 1986, p. 325
and Electrical Ignitability of Dust Clouds, Bureau of Mines 26 Conti, R. S. and Zlochower, I. A. Paper presented at the Eastern
Report of Investigations 8798, Pittsburgh, 1983 Section Combustion Institute Fall Technical Meeting, 2-5
Hertzberg. M., Cashdollar, K. L., Ng, D. L. et al. Nineteenth November 1987, NIST, Gaithersburg
International Symposium on Combustion, The Combustion 27 Hertzberg, M. and Cashdollar. K. L. in Industrial Dust
Institute, Pittsburgh, 1982, p. 1169 Explosions, STP 958. American Society for Testing and
Hertzberg. M., Conti, R. S. and Cashdollar, K. L. Electrical Materials. 1988, p. 5
Ignition Energies and Thermal Autoignition Temperatures for 28 Hertzberg, M.. Cashdollar, K. L. and Lazzara, C. P. Eighteenth
Evaluating Explosion Hazards of Dusts, Bureau of Mines International Symposium on Combustion. The Combustion
Report of Investigations 8988, Pittsburgh, 1985 Institute, 1981, p. 717
Conti, R. S. and Hertzberg, M. in Industrial Dust Explosions 29 Hertzberg, M. and Zlochower. I. A. Twenty-third International
(Eds K. L. Cashdollar and M. Hertzberg), ASTM STP 958, Symposium on Combustion. The Combustion Institute, Pitts-
Philadelphia, 1987, p. 45 burgh. 1990. p. 1247

FUEL, 1991, Vol 70, October 1123