Page 394 - Dust Explosions in the Process Industries

Propagation of Flames in Dust Clouds 363

In a subsequent study using the same 36 m tube facility as used by Kauffman et al.
(1984a), Srinath et al. (1985) determinedthe maximum overpressureand maximum flame
speed for a dried mixed natural organic dust and found 5.4 bar(g) and 607 m/s, respec-
tively.A numerical code developedby Chi and Perlee (1974) for premixed gas explosions
was used to solve the one-dimensionalcompressible flow equations for a flame propa-
gating through a tube of the same dimension as used experimentally.The code did not,
however, close the loop connecting flow, turbulence, and combustion rate, and an empir-
ically based relationship between the turbulent burning velocity and flame propagation
distance derived from the actual dust explosion experiments had to be employed. Under
these circumstances,the codepredicted a maximum explosionpressure of 5.89 bar(g) and
a maximum flame speed of 607 m/s, in good agreement with experiments. However, as
will be discussedin the next section,improved,more-comprehensivenumericalcode con-
cepts for dust explosion simulation are being developed. Further works on propagation
of dust explosions in long ducts and pipes are reviewed in Section 9.2.4.6 in Chapter 9.

4.4.8
THEORIES OF FLAME PROPAGATION IN TURBULENT DUST
CLOUDS: COMPUTER MODELS

4.4.8.1
Background
The discussionof flamepropagationmechanisms in the previous sections, in particular tur-
bulent propagation where turbulence is generated in situ by flow produced by the explo-
sion itself,has demonstratedthe vast complexityof the turbulent flame propagationprocess.
Simple experimental correlations are not sufficient for predicting explosion development
in complex practical situations.Atheory is needed that can unify all these experiments.In
its cornprehensiveform, the theory should includethe mechanisms of dust entrainment and
dispersion (see Chapter 3) as an integrated element in the complex feedback interaction
between the combustionof dust cloud, expansionof combustionproducts, gas flow ahead
of the flame, turbulence in the gas flow ahead of flame, intensified entrainment and dis-
persion of the dust ahead of the flame-and back to intensifiedcombustion.Increasingavail-
ability of computational power has facilitated considerable progress over the 1970s and
1980s.As reviewed in Section9.4.2.7 in Chapter 9, comprehensivecomputer codes forpre-
dicting dust explosion propagation in complex industrial geometries are currently being
developed. Encouraging progress has been made in the prediction of gas explosion prop-
agation in congested geometries as, for example, in modules on offshore oil and gas pro-
duction platforms or compact onshorerefineries and petrochemicalplants. The pioneering
work by Hjertager (1982, 1984, 1986),which uses the two-equation k-E model of turbu-
lence by Jones and Launder (1972, 1973) and the combustion model of Magnussen and
Hjertager (1976), should be mentioned specifically.More recently, Cant and Bray (1989)
developeda theoreticalmodel of turbulent combustionof premixed gases in closed bombs,
which may also be a useful starting point for dust cloud explosion simulation.
However, in the case of dust clouds, the two-phase nature of the problem adds con-
siderably to the complexity. The previous sections of this chapter give some elements