Page 375 - Dust Explosions in the Process Industries
P. 375
344 Dust Explosions in the Process industries
Shevchuk et al. (1986) studied flame propagation in unconfined clouds of aluminum
dust in air at various levels of preignition turbulence. The clouds were generated from
a set of four dust dispersersdriven by a short blast of compressedair. Each disperserwas
charged with 1-10 kg of dust. After completion of dust dispersion, the dust cloud was
ignited after a desired delay. The highest level of preignition turbulence existed imme-
diately after completion of the dispersion. As the ignition delay was increased, the tur-
bulence decayed; and after a sufficiently long delay, the dust cloud was essentially
quiescent. Figure 4.51 gives some results.
2.c
- 1.5
E
I
a
-J
m
L
,4 1.0 - 6 XT.O.1s I
L
0 Figure 4.51 Radius of flame ball as a function of
L
v1
2
.‘ /” delay between completion of dust dispersion and
0 4$ time from central ignition of unconfined cloud of
2 0.5 . /%;:4;
10 pm diameter aluminum flakes in air; z is the
-0 1 0 effective ignition of the dust cloud. The dust con-
0.0 I centrations are normal averages (quantity of dust
The data points for z= 0.1 s and 62 g/m3are from three different but nominally identi-
cal experiments.Figure 4.51 shows that the initial radial flame speed decreased systemat-
ically with increasing ignition delay, or decreasinginitial turbulence,from about 30 m/s at
z= 0.002 s via 20 m/s at z= 0.1 s to about 1m/s at z=0.4 s. The ignitiondelay of 0.4 s was
probably sufficientlylong to render the dust cloud practically laminar at the moment of igni-
tion. However, after about 0.05 s, the flame was no longer laminar and acceleratedrapidly
to about40 m/s over the very shortperiod 0.05 to 0.07 s. Shevchuket al. suggested that this
“switch”from laminar to turbulent conditionsis triggered by flameinstabilitiesdue to non-
homogeneous dust concentration,which is inevitable in a real dust cloud. They defined a
special Reynoldsnumber for establishinga criterionfor the laminar-to-turbulenttransition:
(Radius of flame ball at transition point) x (Flame speed at transition point)
Re* =
(Kinematic viscosity of air)
and found that the transition generally occurred at Re” in the range 104-105.
4.4.5
SYSTEMATIC COMPARATIVE STUDIES OF TURBULENT GAS
AND DUST EXPLOSIONS
The dramaticinfluence of turbulenceon gas explosions has been studied extensively.The
investigations by Moen, Lee, and Hjertager (1982) and Eckhoff et al. (1984) are examples