Page 333 - Dust Explosions in the Process Industries

302 Dust Explosions in the Process Industries

relative motion by means of correction factors to the spherically symmetric stagnantfilm
situation.
Ignition was induced by introducing a heat source of a given intensity in the energy
equation for a limited time. The model then predicted the particle heat-up, devolatiliza-
tion, and ignition of the volatiles and the subsequent flame propagation through the
spherical volume.

Figure 4.24 Computed pressure-versus-timepat-
terns for spherical explosions of coal dust of var-
ious particle diameters in air in a vessel of 0.10 pm
0 " / , 1 / 1 ~ 1 1 1 diameter: (a) 30 ,urn, (b)50 pm, (c) 100 ,um (From
00 bl 02 03 04 D5 06
t la1 Continillo, 1988).
Figure 4.24 shows an example of computations for laminar explosions of coal dusts
of various diameters in air, in a spherical vessel of 0.10 m diameter.
The predicted final pressure of about 12.5 bar(abs) is close to the maximum theoret-
ical adiabatic pressure. This is much higher than maximum pressures found in experi-
ments. The reasons are that the model accounts for neither heat losses nor endothermic
dissociation in the burned mixture.
Continillo (1988b) expressed some important view points concerning the use of com-
puter models for simulating dust explosions. A space resolution on the order of a few
pm is necessary for a detailed description of particle-scalephenomena. On the other hand,
the typical thickness of a real dust flame is on the order of 10 mm or more, whereas the
physical dimensions of process units in which dust explosions take place is on the order
of 1-10 m. This means that the ratio of the various length scales involved covers up to
7 orders of magnitude. Therefore, detailed comprehensivemodeling considering all the
relevant mechanisms across all 7n orders of magnitude is not really feasible, even by
means of extensivenumerical computing.In addition, such a model would require infor-
mation about a number of microscopic characteristics of the dust particles and their
interaction with heat and gas flows, which can be acquired only by complex, extensive
experimentation. Furthermore, as discussed in Chapter 3, the mechanics of generation
of dust clouds is very complex, and very small particles on the order of 1pm diameter
may not become dispersed into individual primary particles but appear as considerably
larger agglomerates constituting the effective particles in the dust cloud.
The optimal simulation model should include the minimum level of detail necessary
to reproduce the significant features of the explosion development with sufficient accu-
racy. The specific interpretationof this statementmay vary with the objectiveof the sim-
ulation. From an industrial safety point of view, the upper range of the length scale is
most important, whereas for studies of the combustionprocess as such, for example, for
predicting chemical conversion, the smaller scales may be of greater interest.
No matter what the objective,it is beyond doubt that computer simulationis the future
tool for predicting dust explosion developmentin industrialpractice. However, it is then