Page 634 - Dust Explosions in the Process Industries

Research and Development 60 1

normalized turbulence intensities of dust clouds and their normalized burning veloc-
ities for four combustible dusts. For the three of these requiring oxygen from the air for
their combustion-lycopodium, maize starch, and PMMA-positive correlations were
found, as would be expected from experience (see Figure 4.47 in Chapter 4). However,
Mitgau and Mitgau et al. also tested a dust of a special chemical compound having the
required oxygenfor its combustion within itself. In this case, when the supply of oxygen
from the atmosphere was not required for combustion of the particles, the effect of
increased turbulence was in fact to decrease the burning velocity. One mechanism that
could cause such a decreaseis increased cooling of the burning particles due to increased
relative velocity between each particle and the air surrounding it with increasing tur-
bulence of the dust cloud. Apparently, the mechanism of increased turbulent mixing of
combustion products and burning particles with unburned particles, which would
be expected to enhance combustion, was not sufficiently strong to counteract the
combustion-retarding mechanism by increased cooling of the preheated and burning
particles due to increased turbulence.
Therefore, enhanced turbulent replacement of gaseous reaction products by fresh air
around each particle is perhaps a more importantbasic combustion enhancementmech-
anism in clouds of particles requiring oxygen from the air for their combustionthan tur-
bulent mixing of combustionproducts and burning particles with unburned particles. In
other words, for these types of particles, which are normally encounteredin dust explo-
sions, it seems as if the velocity differencebetween an individualparticle and the air sur-
rounding it on a microscopic scale is a basic key factor causing the burning velocity to
increase with increasing “turbulence.”
In the development of comprehensive models to be used in practice, some pragma-
tism is still required. For example, an induction time for ignition may be taken as a
global characteristic of the combustion chemistry (shock tube or stirred reactor; see the
review of shock wave ignition in Section 9.2.3).An alternative approach is to consider
the laminar burning velocity as the fundamental parameter, as suggested by Bradley,
Chen, and Swithenbank (1988). Empirical relationships between turbulent burning
velocity and turbulence intensity are then established, using the laminar burning veloc-
ity as a normalizing parameter. Numerical ‘tflamelibraries” can then be established and
used for closing the positive-feedback loop of combustion-expansion-flow-turbulence-
combustion in numerical dust explosion simulation codes. The ongoing research and
debate on numerical modeling of premixed gas combustion should be watched care-
fully to ensure that any elements that may contributeto solving the dust explosion mod-
eling problem be explored. Korobeinikov and Vorobiev (1996) showed how catastrophe
theory may be applied in the mathematical analysis of the complex ignition and extinc-
tion processes involved, for example, in the propagation of flames in dust clouds and
premixed gases.
Understandingflame acceleration due to flame distortion and turbulenceproduced by
the propagating explosionitself is central for understanding both dust and gas explosions
in practice. Extensive experimentalresearchprograms have been conductedto study these
phenomena for gas explosions in obstructed geometries, as discussed, for example, by
Moen et al. (1982);Hjertager, Fuhre, and Bjoerkhaug (1988);and BaMte and Wingerden
(1992). By employing the experimentalfacilities used in these experiments and repeat-
ing the experiments in the various vessels, using dust clouds instead of premixed gas,
valuable insight could be gained. Systematic comparison of results with previous data