Page 386 - Dust Explosions in the Process Industries

Propagation of Flames in Dust Clouds 355

also applies to dust clouds. The only major difference is that a dust cloud needs to be
generated by raising dust deposits into suspension. This means that stage I and possi-
bly also stage 2 in Figure 4.63, the ignition and laminar propagation of the initial flame,
may not be relevant for dust flames. As already discussed, Greenwald and Wheeler
(1925) used black powder to stir up and ignite the primary dust cloud, whereas Fischer
(1957) used a turbulent gas flame. However, once the primary dust explosion is under-
way, the blast wave generated by it entrains dust further downstream, as already discussed.
Therefore, all stages of Figure 4.63, from stage 3 and downward, apply even to dust
clouds. The essential reason for the flame acceleration is turbulence generated in the
unburned cloud ahead of the flame due to wall friction when the cloud is pushed toward
the open tube end by the expansion of the part of the cloud that has burned. When the
flame front reaches the turbulent unburned cloud, the combustion rate increases. This,
in turn, increases the expansion rate of the combustion products and therefore also the
ow rate of the unburned cloud ahead. The result is an even higher turbulence level and
further increase of the combustion rate. During all these stages, compression waves are
emitted and propagate toward the open tube end. Because of heating of the cloud ahead
of the game due to adiabatic compression, each wave propagates at a slightly higher veloc-
ity than the previous one. Ultimately, therefore, they all catch up with the initial wave
and form a strong leading shock front. The turbulent flame front also, due to the posi-
tive feedback mechanism of combustion rate flow rate turbulence enhanced combustion
rate, eventually catches up with the leading shock wave. If the leading shock is sufficiently
strong, a switch can occur in the mechanism of flame propagation. Instead of heat being
transferred by turbulent diffusion behind the leading shock wave, the dust cloud may
become ignited in the highly compressed state inside the leading shock. If the induction
time of ignition is sufficientlyshort, the chemical reaction zone and the propagating shock
wave thlen become closely coupled and propagate through the cloud at constant veloc-
ity. This is detonation. (see Section 4.5). However, as already mentioned, flame propa-
gation at a constant high speed need not be a classical detonation but can also be a
high-speed turbulent deflagration supported by wall friction induced turbulence.
Figure 4.63 shows a tube with a comparatively smooth internal wall. However, if the
wall roughness is increased, the positive feedback loop of combustion acceleration
becomes more effective, and acceleration up to detonation occurs over a shorter distance.
Gas explosion experiments have been conducted in tubes in which the “wall friction”
was increased systematically by inserting in the tube a number of equally spaced, narrow
concentric rings in contact with the wall. Such experiments were in fact carried out by
Chapman and Wheeler (1926) in a small laboratory-scale tube of diameter 50 mm and
length 2.4 m, open at both ends. For methane/air, flame speeds of up to 420 mls were
measured as opposed to 1.2m/s without the rings. Chapman and Wheeler were fully aware
of the essential role played by flow-generated turbulence. Similar dramatic effects of such
equally spaced rings were found by Moen et al. (1982) for methane/air explosions in a
one-end-open large-scale tube of 2.5 m diameter and 10 m length.
These investigations are of considerable interest in relation to dust explosions in coal
mines, where the supporting structures of the mine galleries would seem to have the same
type of turbulence increasing effect as the concentric rings in tubes (Fischer, 1957). In
the process industry, the legs of bucket elevators are long ducts with repeated obstacles.
Rae (1971) analyzed coal dust explosion experiments in various large-scale tubes and
galleries of lengths in the range 100-400 m, conducted in the time period 1911-1971.