Page 392 - Dust Explosions in the Process Industries

Propagation of Flames in Dust Clouds 36 I

whether venting a primary explosion in a confined space could prevent the development
of secondary explosions in adjacent areas by reducing the expansion velocities and
hence the dust entrainmentpotential of the primary explosionsin those areas. The exper-
iments showed that a dust flame propagated down the gallery even if the mass of the dust
layer, per unit length of gallery, was considerably smaller than that correspondingto the
minimum explosible concentration if dispersed uniformly over the whole gallery cross
section. This is because the dust was dispersed only in the lower part of the gallery
volume and therefore gave real dust concentrations higher than the nominal values. In
accordance with this, it was observed that the dust flame thickness was in fact consid-
erably smaller than the height of the gallery. Such secondary dust flames were found to
sweep along the gallery floor all the way to the exit, even if the dust layer on the floor
was only 0.25 mm thick, representing a nominal dust concentrationreferred to the entire
gallery volume, of only 77 g/m3of maize starch, that is, at the limit for upward laminar
flame propagation.
Typical explosion pressures in the gallery were 0.2-0.4 bar(& if the gallery was
unvented and 0.07 bar(g) with vents close to the primary explosion chamber.
The fact that the dust entrained from the floor was distributed only in the lower part
of the gallery may throw light on the results from Fischer's (1957) experiments, where
stone dust barriers in the upper half of the gallery cross section under certain conditions
proved entirely ineffective in damping the propagation of the coal dust explosion.Fischer
suggestedthat the primary turbulent torus sweeping down the gallery entrained the coal
dust in the lower part of the gallery cross section and the stone dust in the upper part,
with little mixing of the two.
Experiments of the type conducted by Tamanini (1983) and also by the other work-
ers who used a primary explosion to initiate dust entrainment and the main explosion
depend very much on the nature of the primary explosion. Therefore, few generally
valid quantitative conclusions can be drawn from such experiments until the various
processes have been theoretically coupled.
Kauffman et al. (1984a) studied the propagation of dust explosions in a horizontal tube
of length 34.4 rn and internal diameter 0.30 m; that is, LID = 122.A main objective of
the experiment was similar to the one of Tamanini (1983), that is, to identify the mini-
mum quantity of dust deposited as a layer on the internal tube wall that can propagate a
dust explosion sweeping down the tube. The exhaust end of the tube terminated with a
90" bend of 2 m radius leading into a 2.5 m long tube with a number of vents in the wall
but with the far downstream end closed. The ignition source, located at the far upstream
end of the main tube, consisted of a 2.4 m long 50 mm diameter tube filled with stoi-
chiometric hydrogerdoxygen.In the first 3.4 m of the main tube, a dust layer was placed
in a \'-,'-channel running inside the tube parallel to the tube axis. This dust could be dis-
persed into a primary cloud by air blasts from a series of nozzles at the bottom of the
V-channel. In the remaining 33 m of the main tube, the dust layer rested directlyon the tube
wall, either as strips of widths 12.5mm or 90 mm along the tube bottom or as a thin layer
around the whole tube wall. The explosions were initiated by first dispersing the dust in
the V-channel, then igniting the hydrogedoxygen mixture, which would in turn ignite
the djspersed dust. The blast from this violent primary explosion would then sweep
down the main tube and entrain and disperse the dust from the layer on the tube wall, as
in the experiments of Greenwald and Wheeler (1925), Fischer (1957), Pineau (19871,
Tamanini (1983), and in the other investigations discussed by Rae (1971).