Page 383 - Dust Explosions in the Process Industries

352 Dust Explosions in the Process Industries

of his pioneering large-scalegallery experimentsin France.These experimentswere con-
ducted as a consequence of the disastrous coal dust explosion in the Courriers mine in
1906, where 1099 miners lost their lives. Similar work was subsequently initiated in
Poland, Russia, Germany, and the United States.
Greenwald and Wheeler (1925) used a horizontal explosion tube of internal diameter
2.3 m and length 230 m, that is, LID = 100,in their experiments. One end was normally
closed, the other fully open. A pulverized nut coal, ground to 85% by mass <74 pm par-
ticle size, and containing 33% volatiles was used. The ignition source was 800 g of
black powder igniting a primary cloud of 10kg of coal dust. The main quantity of coal
dust was spread along the gallery floor from the point of ignition at 61 m to the full open-
ing of the gallery at 230 m (see Figure 4.61). The quantity of dust spread on the gallery
floor was about 1500 g per m of gallery length, corresponding to a nominal dust con-
centrationin a fully dispersed state of 360 g/m3.No dust was spread out in the 61 m long
section between the normally closed upstream end of the gallery and the ignition point.

1000
NORMALLY CLOSED
END-
-
- VI
\
O w -
500
wl
w
5
2
LL
IGNITION
0- I
0 50 100 150 200 230
DISTANCE FROM NORMALLY CLOSED END [ml
Figure 4.61 Acceleration of coal dust explosions in a horizontal gallery of internal diameter 2.3 m
and length 230 m, showing the effect of venting at the upstream, normally closedend (From Creenwald
and Wheeler, 1925).

The main purpose of Greenwald andWheeler’s experiments was to investigatethe influ-
ences of the location and size of vents on the development of dust explosions in the gallery.
As Figure 4.61 shows, flame speeds of up to 800 m/s were generated with the upstream
end of the gallery fully closed. Whether the plateau of constant flame speed at 800 m/s
beyond 165 m indicates detonation is unclear. Lindstedt and Michels (1989) observed
violent, constant-velocitydeflagrations supportedby wall-frictioninduced turbulencefor
alkanes in air. Similar steady combustion phenomena may also exist for dust explosions
in long tubes and ducts. The flame speed would then be somewhat lower than for a
proper detonation. (Detonation of dust clouds is discussed in Section 4.5.)
Figure 4.61 demonstrates that venting at the upstream, normally closed, end reduced
the acceleration of the explosion appreciably. With a fully open upstream end, compar-
atively weak explosions of maximum flame speeds around 50 m/s resulted. In this case,
Greenwald and Wheeler made some interesting observations. The flame motion was
markedly vibratory, and the column of dust and air preceding the flame was expelled from
the gallery exit in puffs instead of in a continuous stream. The flame itself could be seen
to issue from one of the openings two or three times, with a slight in-rush of air occurring