Confined Gas and Dust Explosions

E Salzano, Istituto di Ricerche sulla Combustione, CNR, Napoli, Italy

Introduction 1
Gas Explosions 2
Dust Explosions 3
Predicting the Course of Confined Explosions 4
Mitigation of Confined Explosions 6
Conclusions 7
References 7

Introduction

Explosion is a term that corresponds to a rapid heat release or pressure rise.1 In combustion science, this definition applies if the
exothermic reaction between the fuel and the oxidant (generally air) permits a rapid energy release, as compared to steady,
low-temperature reactions. For that to succeed, fast kinetics must characterize the overall combustion reaction of the analyzed
fuel–oxidizer system. Explosion limits are then defined as the pressure–temperature boundaries that separate the regions of slow
and fast reactions, where spontaneous explosion occurs.
In these terms, the definition of explosions is apparently in contrast with some of our real life experiences. As a matter of fact
many gaseous or vapor fuels (and dusts) are substantially unreactive (or nonexplosive) at atmospheric conditions. On the other
hand, we recognize the same substances as very hazardous (explosive) if mixed with air, provided an ignition source (a spark or a
hot spot). The latest assumption is consistent with the two apparently contrasting paradigms. Given an initial explosive region in a
premixed or partially premixed reaction system – opposite to the diffusive phenomenon where the slow mixing of reactant species
is the controlling mechanism – a self-sustained reaction wave (the flame or deflagration) is observed, assuming the concentration
of the fuel is within certain concentration limits (the flammability limits). In the flame region (i.e., in the combustion wave),
explosive conditions are always reached and the exothermic, fast combustion reaction takes place. Then, the combustion wave
propagates in the adjacent, unreacted layer by thermal and molecular diffusion. When gas explosions are analyzed, the undisturbed
flame velocity is known as the laminar burning velocity, or the fundamental burning velocity (SL,0) and represents the velocity at
which unburned gases move through the combustion wave in the direction normal to the wave surface. It is a fundamental
physicochemical property of the combustible mixture and depends on the flame kinetics through the characteristic reaction rate
G and on the transport processes through the density-weighted transport coefficient a:

SL,0 ∝ðaGÞ0:5 [1]

To this end, no reference to the fluid dynamics is given with respect to the explosion phenomenon. Slow, laminar regimes are
sufficient for the completeness of the combustion reaction and consequent severe pressure buildup due to hot gas expansion in any
adiabatic, confined volume. Explosive behavior may be then reproduced by using thermochemistry only. In isolated systems,
starting from Hess’ law and heat of formation of reactants and products, we may evaluate the final temperature (and the
corresponding adiabatic, maximum pressure, to be compared with the maximum allowable or failure/deformation pressure of
the analyzed system) once the reaction stoichiometry has been defined and appropriate functions for the thermochemical
parameters (i.e., specific heat at constant volume) of species with temperature are defined. Quite evidently, notwithstanding the
relatively simple mathematical issues, the main problem lies within the stoichiometry of the reaction. Large discrepancy of
experimental data can be observed by adopting oversimplified reaction schemes, which cannot be imputed to the nonadiabaticity
of the system. Indeed, the combustion reaction is a complex system and the definition of the composition of the products is not a
straightforward task unless pure, simple hydrocarbons are perfectly mixed with air. On the other hand, several codes are available
nowadays, some free of charge, which for most common substances will give the chemical equilibrium compositions and
properties of complex mixtures through minimization of Gibbs energy (e.g., Stanjan, CEA). Eventually, the pressurization of the
confined system due to the expansion of the hot reaction products is likely to produce structural damage to many weakly covered
equipment or buildings and we call these explosions, somehow in similarity with the effects of explosives in military applications.
The knowledge of the maximum pressure is nevertheless not sufficient for the prediction, prevention, and mitigation of gas and
dust explosions in confined systems because the kinetics of the entire explosive phenomenon is needed either for the prediction of
ignition likelihood (which is not treated in this article), or for the definition of the characteristic time of explosion texp, which
depends on the burning velocity and is an essential parameter for any prevention or mitigation methodology and for the efficiency
of explosion countermeasures.

Reference Module in Chemistry, Molecular Sciences and Chemical Engineering http://dx.doi.org/10.1016/B978-0-12-409547-2.11031-5 1

In this framework, gas, vapor, or dust explosions may be treated, at least conceptually, under the same theoretical backgrounds.
Nevertheless some differences may be easily foreseen for dust explosion prediction, starting from the necessary premixing phase
before ignition that in contrast with the natural tendency of ideal gas to expand over an entire volume.
In the next sections, we will define in detail the main issues of explosion science in confined systems, with the specific goal of
predicting and mitigating gas or dust explosions in closed vessels/equipment or buildings, that is, for the safety design of industrial
processes or for civilian purposes, whereas prevention systems are not treated.

Gas Explosions
The first comprehensive theory of flame propagation was given by Mallard and LeChatelier at the end of the nineteenth century.
Further advancements were soon performed by use of the Zeldovich–Frank-Kamenetski–Semenov theory for planar stationary
flames. Application of these classical theories is still considered satisfactory at least for simple, pure fuels. In the last years, the use of
well-known kinetic data (e.g., GRI-Mech) and kinetic-based codes as Chemkin or Cantera has largely increased. These codes are
able to reproduce the laminar burning velocity for any fuel mixture at the given initial temperature and pressure conditions by
solving the mass and energy conservation equation for large set of kinetic equation, starting from a hot spot condition. The
composition of combustion products and the final temperature are given and the structure of the flame is intimately resolved.
When comparing the results of fundamental theories of the combustion science or the results of the up-to-date kinetic codes with
experimental data for gas explosion in confined system, large discrepancies are found even for centrally ignited, freely expanding flame
in adiabatic spherical combustion chambers, or in tubes. As a matter of fact, it was recognized quite early that the characteristics of
premixed flames are profoundly affected by aerodynamics, starting from the work of Darrieus in 1938 and Landau in 1944, who
predicted independently that deflagrations are unconditionally unstable due to the thermal expansion of the gas and density changes
across the flame, and the work of Zeldovich on the diffusional–thermal stability of laminar flames. Nowadays, the effects of flow
nonuniformity, flame curvature, and flame-flow unsteadiness on the unstretched burning velocity are analyzed collectively by means
of a linear correlation with the flame stretch rate a,2 that is, the time rate of change of an infinitesimally small surface area element is
normalized over the area, as defined by Karlovitz, and a characteristic Markstein length L:

SL ¼ SL,0  aL
1 dAf
a¼ [2]
Af dt

By using this analysis, a good convergence of the experimental data with the calculated burning velocity has been found.3 Now
consider an explosion in a long tube filled with equidistantly placed circular orifice rings and a premixed fuel air mixture ignited at
one end. After ignition, the expansion of hot combustion products generates a turbulent flow field, which accelerates the flame
ahead it. When reaching the obstacle, the flame surface is distorted or perturbed. Hence, the energy release rate per unit volume is
increased. This effect, which further enhances the flow field velocity generating more turbulence, is often defined as the positive
feedback of explosion and is responsible for the production of shock waves in unconfined volumes (constant pressure) but may
also be recalled for flame acceleration in confined equipment. A characteristic turbulent burning velocity ST may also be defined
analogously with the laminar burning velocity.4,5
Several correlations for ST with respect to the root-mean-square velocity fluctuation, the turbulent intensity u0 and length scale L
have been proposed6:

0

ST u L
¼f g [3]
SL SL dL

where dL is the laminar flame thickness, and f and g are functions. However, the analysis of this phenomenon is still an open issue
and no universal correlation exists because ST depends on specific combustion regimes based on the value of the two ratios u0 /SL
and L/dL, and by a dimensional number: the turbulent Reynolds ReT, Damkhöler DaT, and Karlovitz KaT numbers.7,8 Besides the
pure laminar regime, flamelet (corrugated, wrinkled), distributed, and well-stirred regimes (as in internal combustion engines)
exist. Details on the regimes of combustion, flame dynamics, and other insights on the instabilities for the burning process are not
the goals of this article which is not focused on the fundamentals of premixed combustion science. What is essential here is the lack
of large-scale test data confirming the behavior of turbulent combustion at very large scales, which in turn is essential for the
definition of explosions in industrial applications or buildings. Moreover, the turbulent effects on the flame acceleration (i.e., on
the mass burning rate), which in turn affects the rate of pressure rise in confined systems, are loosely taken into account in the
engineering correlations for the design of mitigation systems for gas and dust explosion suppressors or venting systems. Eventually,
the only possibility of reproducing the overall transient, reactive phenomenon of the industrial explosion, and engineering
application is the adoption of the Computational Fluid Dynamics (CFD) technique, which is increasing in popularity among
the scientific and engineering community as a valuable prediction tool, in no small part to the ever growing computational power.
In the following sections, this approach will be analyzed in light of recent advances on the subject.
The continuous increase in flame speed due to the turbulent flow field may also lead to the extreme phenomenon of detonation.
This effect may be analyzed by starting from the classical pressure–volume (specific) diagram of fluid and the steady, planar,
 Confined Gas and Dust Explosions 3

one-dimensional flow principles of conservation of mass, momentum, and energy. In homogeneous backgrounds with perfectly
unobstructed geometries, weak deflagrations characterized by subsonic fluid velocities are typically expected. However, the
transition from deflagration to detonation (DDT) may occur in real industrial system (e.g., corrugated duct work or process
pipelines). The DDT phenomenon is very complex and far to be completely understood.9–13 In the simple but effective ZND model
(Zeldovich, von Neumann, and Döring), a detonation wave is described as a two-zone model: a strong inert shock wave (produced
by the deflagration in the case of DDT) compresses adiabatically the unreacted mixture beyond its autoignition temperature, thus
producing a reaction zone. The shock is in turn maintained by the backward expansion of the combustion products, thus providing
the thrust needed to drive the shock. In most experimental cases, the self-sustained detonation waves observed in tubes are
characterized by supersonic velocity in the shock discontinuity and by the existence of a sonic surface in the burned gas. This
condition is named Chapman–Jougouet (CJ) detonation. Following Lee,11 for a DDT to occur, the deflagration must first accelerate
to its maximum CJ deflagration velocity, which is about 0.5 the sound speed of the products. Then the formation of a detonation
results from the development of transverse perturbations at the shock front to form a self-resonating system where the gas-dynamic
oscillations are supported by the energy released by the chemical reactions.
The occurrence of detonation in accidental explosions is relatively rare, but the hazard may increase consistently when
considering highly reactive gases in large-scale equipment. In this case, the detonation phenomenon has dramatic effects on
structures. In hydrogen/air mixtures, the velocity of propagation is 2000–3000 m s1 and the increase in pressure is typically
about 20 times the initial pressure. Furthermore, the characteristic time of the detonation phenomenon is very short and any active
protective measure (e.g., suppressor, venting) is often ineffective, whereas passive protection (detonation/flame arresters) may be
positive. To this aim, two main parameters are generally adopted in the design industry for the prevention of DDT, particularly in
ducts or tubes, namely the detonation cell size and the run-up distance, which are tabulated for most common gases in air.14 A tube
diameter or an equipment scale smaller than the cell size, or conversely a tube length lower than the run-up distance should present
a lower hazard of detonation even if open issues are still existing (e.g., for methane/air mixture in corrugate tubes).15
Some other types of explosions such as strong deflagrations, strong or weak detonations have been postulated but never or
rarely observed experimentally. The knowledge on nonclassical types of explosions is still very scarce and may be confined to
combustion science rather than industrially occurring phenomenon. Other unknown behaviors may be seen for oxygen-enriched
or pure oxygen atmosphere and nonhydrocarbon mixtures.16 The recent special issue from Oran and Williams17 has analyzed in
detail the actual knowledge of the scientific community on the physics of the entire complex detonation phenomena.

Dust Explosions
Dust explosions are the result of the rapid combustion of fine solid particles, such as organic dusts (sugar or wood), metal dusts
(aluminum and magnesium); polymer based dusts, and carbonaceous dusts. The magnus opus of Eckhoff18 is the starting reference
for establishing basic knowledge and research needs on this subject. The author invites the reader to refer to the cited textbook for
any in-depth analysis on the several chemical and physical aspects related to this phenomenon, starting from powder science to
technology.
Following Eckhoff19, the range of explosive powder/dust concentrations ranges from 100 g m3 to a few kilograms per meter
cube. Particle size ranges typically from nanometer to half of millimeter, which corresponds to specific surface area of few meter
square per gram (or about 1 m2 cm3). Genuine nano-powders have a surface area >60 m2 cm3 and should be treated sepa-
rately.20 Most standards provide a basis for a combustible dust to be defined as an explosive if the particle sizes are  420 mm.
In general, we can say that, with the exception of pure metal particles, basic theoretical knowledge on the combustion of
suspended dust clouds in closed systems may be characterized by the same thermochemical and fluid-dynamic analysis defined for
premixed gaseous explosions, provided there is intense gasification/pyrolysis/devolatilization phenomena on the single dust
particles in the preheating zone. Multiphase combustion poses, however, several issues and complexities in the full comprehension
of dust explosion mechanisms and for the sound engineering application of prevention and mitigation measures. The flame
thickness is clearly determined by the volatile gases driven out of the particles ahead of the flame. However, for many organic
materials, the combustion of the remaining solid char particles occurs at a slower rate in the tail of flame, where heterogeneous
combustion may take place (Nusselt-type flame). These processes are controlled by the diffusion of oxygen to the surface of
individual solid particles. In many organic dusts, both Nusselt-type flame and premixed volatile gas flames are present. In some
cases, as for polymer based dusts, a preliminary melting phase has been observed, followed by vaporization and finally vapor phase
burning. Pure metals, such as aluminum, may have high gasification temperature and even more complex behavior.
The ease with which a dust ignites and the violence (defined as the maximum rate of pressure rise in the case confined
explosion) increase with decreasing particle size. However, dusts are almost never present as a single size but rather in a range.
Hence, the tail end of this size distribution may represent the natural source for explosion. Furthermore, in order for an explosive
dust cloud to be formed from a layer/deposit, the same must be exposed to a process that disperses the particles in the air at the
extent that the dust concentration drops into the explosive range, either intentionally (handling and transportation in mills, dryers,
filters, and cyclones) or unintentionally (e.g., in the case of cloud generated from dust layer by blast wave produced by a primary
explosion or by gas flow passing across the layer).21 Either in laboratory equipment, where specific dispersion systems are adopted
or, very likely, in real case accidents, the dust/powder suspension is never homogeneous. Migration of heavy particles inside
turbulent eddies results in local concentration gradients. Furthermore, the competition between turbulent dispersion and inelastic
particle collision produce coherent particle congregation.22,23
4 Confined Gas and Dust Explosions

Another important issue regards the measured maximum flame temperature (hence the maximum pressure) reached by
accidental dust explosions, which results largely lower than the theoretical data, even hundreds of degrees. This deficit of enthalpy
may be explained by the formation of soot particles due to imperfect reactant (either the pyrolized/gasified gas or the dust)
mixing.24 In many industrial applications, this point is essential because mitigation systems rely on the maximum pressure as
calculated or tested experimentally in closed bomb tests.
As for gas explosions, Borghi’s idealization of turbulent regime can be still conceptually adopted, and several correlations for the
turbulent burning velocity ST developed for gaseous mixtures can be applied for dust explosions. However, the influence of
turbulence on dust explosions is less evident than in gases. Furthermore, results are still uncertain and more in-depth information is
needed for the development of proper correlations between turbulent intensity and burning velocity.24 Finally, self-sustained
detonation waves can develop in pulverized organic or metallic particles suspended in air. Because heterogeneous dust–air mixtures
are insensitive to detonation, however, transition from DDT in tubes requires a sufficient tube length–diameter ratio and initiation
energy stronger than that used in gaseous DDT.25

Predicting the Course of Confined Explosions

Figure 1 shows typical explosion histories for methane–air mixtures at stoichiometric concentration in nonadiabatic, empty, closed
vessels of different geometry. Similar sigmoidal behavior is found for dust explosions. With specific reference to the safety design of
industrial equipment or buildings, the prediction of the course of an explosion consists essentially of the reproduction of the
pressure history recorded in the system. In spherical, adiabatic systems, under the laminar assumption, that means the prediction of
the rate of pressure rise, which depends on the scale of the system and on the reactivity of the fuel/oxidant mixture, that is, on the
burning velocity, provided the flame thickness is negligible with respect to the overall characteristic length of the combustion
chamber. Starting from the assumption that, at any moment after the ignition, fractional pressure rise is proportional to burned
mass mb over the total initial mass m0,26 that is:

P  P0 mb
¼ [4]
Pe  P 0 m0
it can be demonstrated that:

ð1=gÞ "
0 1=g #2=3
dP 3 ðPe  P 0 Þ P 0 P Pe  P
¼ 1 SL [5]
dt rb P P Pe  P 0

where P0 is the initial gas pressure, g is the heat capacity ratio, rb is the equivalent radius of burned volume, Pe is the maximum
pressure which can be considered as equivalent to the adiabatic flame pressure Pad and the laminar burning velocity is dependent
on pressure and temperature by the classical correlation27:

að’Þ
bð’Þ
SL T P
¼ [6]
SL ð T 0 , P 0 , ’ Þ T0 P0

where SL(T0,P0,’) is the reference laminar burning velocity at the given initial pressure T0 and P0, and a and b, are specific functions
which depends on the equivalence ratio ’ (the ratio of fuel/oxidant concentration over the same ratio at stoichiometric
concentration) or, equivalently, of the fuel concentration. This equation is valid either for gas or dust.28
Equation [5] is the starting equation for the definition of the gas explosions severity index KG or for the equivalent dust index
KST, which correspond to the maximum rate of pressure rise observed in an 1 m3 spherical vessel. These two indexes are based on
the assumption that, given the fuel mixture, the maximum rate of pressure rise depends only on the scale of the confined system,
based on the so-called cubic-root law29,30:

dP  p ffiffiffiffi
KG ðor STÞ ¼ 3
V [7]
dt max

where V is the volume of the combustion chamber. The cubic-root law assumes that the burning velocity is constant, and that the
maximum rate of pressure rise is found when the flame is close to the vessel wall, hence rb is equal to the equivalent radius of the
vessel. Quite clearly, the correlation is considerably ambiguous. In the case of gases, the value of KG is found experimentally in
quiescent gaseous fuel mixtures and the effects of the turbulence are not included. On the other hand, for dust explosions, intense
initial turbulence is necessarily introduced for dust dispersion. Hence, the KST is totally arbitrary because the state of turbulence is
arbitrary itself. Furthermore, it is completely in-appropriate for layered dusts, for which the combustion is dominated by the dust/
air mixing process.31 Finally, the KST value has been proved to vary strongly with several material properties, particularly the particle
diameter, the dispersion methodology, and the humidity.32,33 In this framework, Amyotte et al.34 has, for example, discussed on
the substitution effects for dust explosion based on KST and particle distribution.
Despite the drawbacks, the two indices have been largely adopted for the development of practically all the existing national and
international standards for the design of mitigation systems and particularly for venting, which is the cheapest and the most
 Confined Gas and Dust Explosions 5

Figure 1 Experimental pressure histories for explosions of methane in air at stoichiometric concentration, measured in different vessels and initial
conditions. The term t0 is to the ratio of the flame path (or the characteristic length scale L of the vessel) to the flame speed SF (¼ru/rbSL);
Spherical: Central ignition; V ¼ 0.15 dm3; L ¼ 0.14 m; Cylindrical: Central ignition; V ¼ 0.05 dm3; L ¼ 0.2 m (half of the tube length). The variation of the
pressure slope after the flame has reached the vessel wall; the lowest maximum pressure with respect to the adiabatic pressure; and the highest
maximum rate of pressure rise with respect to the lowest volume can be observed; Tube: End ignition; V ¼ 0.03 dm3; L ¼ 1.2 m. Strong heat exchange at
the tube walls reduce consistently with the maximum pressure and change the pressure history with respect to the spherical case; Cylindrical,
turbulent: Central ignition; V ¼ 0.05 dm3; L ¼ 0.2 m (half of the tube length). Initial turbulence provided by a rotating fan 2 m s1. The increase of the
rate of pressure rise with respect to the corresponding quiescent explosion in the same vessel can be observed, even if the maximum pressure is
higher due to shortest duration of heat exchange to the vessel walls.

common measure for confined gas and dust explosions. The use of gas and dust deflagration indices should then be absolutely
calculated in standard equipment and is limited to the application required by standard methods.
As discussed earlier, for both gas and dust, the effects of self- and geometry-derived turbulization of the flame and hence on the
overall development of the explosive phenomenon are substantially unpredictable with simplified analysis unless spherical, empty
combustion chamber, even if several attempts have been presented in the literature by adopting a lumped parameter. In the recent
years, distributed parameter models CFD codes have become a routine tool for the simulation of either confined or partially
confined explosions. Due to the excessive computational efforts in the direct solution of Navier–Stokes equations, the simulation
of any industrial system is always based on the Reynolds-averaged Navier–Stokes (RANS) code. The turbulence is generally
implemented by empirical submodels, typically the k–e model or similar, however, still empirical, more advanced models. Due
to the short time scales and computational stiffness, the combustion reactions cannot be solved by Arrhenius equation. As a
consequence, the mean burning rate, in any finite volume, may only be evaluated by using empirical correlations for the turbulent
burning velocity and for the propagation of the reaction region within the numerical domain. A ‘subgrid’ or ‘porosity’ formulation
is always considered so as to take into account the turbulence induced by obstacles when the volume of the same obstacle is smaller
than the volume of the single computational cell.
6 Confined Gas and Dust Explosions

The use of RANS for the reproduction of large-scale, compressible, reactive systems is questionable and very large uncertainties
are intrinsically produced by the several assumptions on the complex combustion phenomena involved in confined and partially
confined explosions. On the other hand, this technique is the only available option for engineering analysis of large-scale industrial
explosions, particularly in the presence of obstacles and when flow fields are produced by the venting phenomenon, even if the
high-level skill and large experience of operators are needed. In the recent years, several commercial codes are on the market with
large success, for example, Flame Acceleration Simulator (FLACS) (Gexcon) or FLUENT (ANSYS). These codes are effective in
reproducing high turbulent regimes in relatively small scale whereas they have some difficulty in reproducing quasi-laminar
regimes. On the other hand, assessing turbulent regimes is safer because an increase in the rate of pressure rise and maximum
pressure (in the case of venting system) are typically expected.
The development of CFD codes for the prediction of dust explosions is strongly required in the process industry for the
definition of obstacle-induced turbulence and the explosion feedback (similar to gases). An ambitious program is that of DESC,23
which is based on the RANS code developed for gas explosions, named Flame Acceleration Simulator. More recently, the use of
large eddy smulation (LES) has been proposed. This technique is still under development and limited to smaller volume due to the
available computational power.35

Mitigation of Confined Explosions

The definition of the strength of enclosure or process unit is the first parameter for the evaluation of resilience of equipment and
building with respect to the internal explosion. To this aim, a first reference is the design pressure, which should never be exceeded
by the internal pressure produced by the deflagration. If analyzing the classical design standards (e.g., API 620, API 650, UL 2085,
and ASME codes for unfired vessel and similar others) the large part of equipment and all buildings are just too weak to withstand
any gas or dust explosion, even in the almost-negligible partial volume (layered) deflagration case. This aspect forces the designer
toward the application of efficient prevention measures as inerting the atmospheres, working outside the flammability limits, or
avoiding the static charge accumulation by grounding. In the case of explosions, the typical choices of designers are addressed in
two mitigation measures defined as suppression or venting. Both methodologies restrict the development of damaging pressures
during incipient stages and must operate in as short a time as possible (Figure 2, data elaborated from Cooper et al.36).
The mechanisms of suppression of the explosion is essentially based on quenching but free radical scavenging, wetting, and
inerting mechanisms are also present. Injection devices are based on chemical detonators installed inside the vessel or, more
frequently, on high-discharge rate bottles installed on the outside of the vessel and pressurized with nitrogen. Used standards are
EN 14373 or NFPA 69.
The type – and the corresponding design – of venting systems for gas or dust explosion is typically based on empirical
correlations published in current standards as, among many, NFPA 68, EN 14491, EN 14494, and EN 14497. The cited correlations
are generally overconservative, particularly for low-strength equipment. Furthermore, many fundamental aspects in combustion
described in the previous article of this work, as turbulence effects for gas explosion, are neglected or simply disregarded. Hence,
unless ideal suppression or venting actions are carried out, the cited semiempirical correlations are then unable to reproduce
correctly the venting phenomenon, even in the case of an empty, lab-scale apparatus.

Figure 2 Pressure history in closed vessel with mitigation system activated at the set pressure Pv. Data elaborated from Cooper et al.36; 1: Initial
undisturbed pressure rise, after ignition; 2: Ideally mitigated explosion by suppressing or venting systems; 3: Vented explosion. The values of P1–P4 are
peak pressures reached inside the mitigated system: P1 ¼ peak pressure reached after vent opening. The raise of the pressure is due to the inertia of the
vent time, and to the time required to establish the sufficient flow of unburned gas from the vessel for the internal pressure to decrease; P2 ¼ peak
pressure related to the external explosion (in open atmosphere or in the case of ducted venting system). The flame front emerging from the vent
ignites the unburt gas or dust previously expelled from the vessel; a partial occlusion of venting section and/or back-flow of burnt gases in the vented
chamber; P3 ¼ peak pressure (often named reduced pressure) reached inside the equipment. The raise of pressure is due to turbulence induced by
the pressure field between the low-pressure venting area and the pressurized, internal sections, and to the turbulence induced by the geometry of
the equipment. The acceleration of flame affects the characteristic venting ratio as in eqn [2]. The pressure is followed by low-frequency Helmholtz-type
oscillations; P4 ¼ peak pressure due to high-frequency acoustic-type oscillations, this particular pressure peak could not occur in most practical
situations.
 Confined Gas and Dust Explosions 7

Several scientists, however, have produced significant literature and correlations on this field, starting from the fundamental
paper of Bradley and Mitchelson26 and the other pioneering works of Tamanini37and Rota38. These contributions are essentially
based on the concept that the maximum explosion pressure reached in the combustion chamber (the reduced pressure, Pred) is
proportional to the ratio of two characteristic times:


Pred tmit
¼f [8]
P0 texp

where tmit is the characteristic time of mitigation (e.g., venting, suppression) and P0 is the absolute initial pressure: the shorter the
explosion time, the higher the final pressure. In the case of venting, Bradley and Mitchelson expressed the reduced pressure through
several correlations based on the dimensionless ‘vent ratio,’ A0 /S0 , that is, the ratio of the dimensionless vent area, A0 :

Av
A0 ¼ CD [9]
As
and the flame Mach number, S0 :

Sf Sl
S0 ¼ ¼ ðE  1Þ [10]
c0 c0
where CD is the discharge coefficient, Av is the vent area and As is the internal surface of the vented equipment, E is the expansion
ratio (the ratio between the density of unburned and burned gas, ru/rb), and c0 is the acoustic velocity in the unburned gas. The
analysis of the ratio A0 /S0 reveals that it basically represents the ratio of two characteristic times defined above where tmit is

Vv
t exp ¼ [11]
As Sl ðE  1Þ

is the characteristic time of combustion of the mixture inside the vessel (with volume Vv) once the flame area is assumed to be
represented by the internal surface As, and tv is

Vv
tv ¼ [12]
CD Av c0
These equations are valid for empty vessel. Nevertheless, the flame propagation is affected by turbulence induced by the vent
itself, which is generally taken into account by a ‘turbulization factor’ or DDT interaction, as defined by Molkov39. Lumped-
parameter models have been then developed as, for example, the SCOPE (Shell Code for Over-pressure Prediction in gas
Explosions) model40 in order to reproduce the turbulence induced by the venting itself or obstacles. Further details on gas or
dust explosion mitigation systems are in the cited literature. See also Zalosh41 and Amyotte and Eckhoff 42 for further reference on
venting systems of gas and dust explosions.
Despite these efforts, the complex behavior of vented explosions are, however, not reproducible with any integral or lumped-
parameter models unless a factor of 2 for the reduced pressure. It is again clear that the only future for the design of explosion
mitigation in real vented systems is the RANS–CFD or more accurately, the LES-based CFD.35,43,44

Conclusions

The description of the explosion phenomena given above applies for any type of gaseous fuel/oxidant mixture in a confined or
partially confined system.
For engineering purposes, with reference to the design of appropriate prevention and mitigation systems, the characterization of
both main thermochemical and kinetic parameters is vital.
CFD must be considered as the next step in future assessments for the prediction of explosions and for the sound design of
prevention and mitigation systems. Further fundamental knowledge on combustion modeling and numerical development for the
LES-based CFD codes is strongly needed. However, CFD is not currently a cost efficient way to conduct several calculations in a
timely manner. There still exists a world for both simple codes and complex codes for now and it is up to the user to provide an
educated guess.

References
1. Glassman, I.; Yetter, R. Combustion, 4th ed.; Academic Press Elsevier: San Diego, USA, 2008.
2. Law, C. K.; Sung, C. J. Structure, Aerodynamics, and Geometry of Premixed Flamelets. Prog. Energy Combust. Sci. 2000, 26, 459–505.
3. Law, C. K. Combustion at a Crossroads: Status and Prospects. P. Combust. Inst. 2007, 31, 1–29.
4. Sivashinsky, G. I. Instabilities, Pattern Formation, and Turbulence in Flames. Annu. Rev. Fluid Mech. 1983, 15, 179–199.
5. Matalon, M. Intrinsic Flame Instabilities in Premixed and Nonpremixed Combustion. Annu. Rev. Fluid Mech. 2007, 39, 163–191.
6. Gülder, O. L. Turbulent Premixed Flame Propagation Models for Different Combustion Regimes. P. Combust. Inst. 1990, 23, 743–750.
8 Confined Gas and Dust Explosions

7. Borghi, R. On the Structure of Turbulent Premixed Fames. In Recent Advances in Aeronautical Science; Bruno, C., Casci, C., Eds.; Pergamon Press: London, 1984.
8. Bradley, D.; Lawes, M.; Liu, K.; Mansour, M. S. Measurements and Correlations of Turbulent Burning Velocities Over Wide Ranges of Fuels and Elevated Pressures. P. Combust.
Inst. 2013, 34, 1519–1526.
9. Liñan, A.; Williams, F. A. Fundamental Aspects of Combustion; Oxford University Press: New York, 1993.
10. Oran, E. S.; Gamezo, V. N. Origins of the Deflagration-to-Detonation Transition in Gas-Phase Combustion. Combust. Flame 2007, 148, 4–47.
11. Lee, J. H.S The Detonation Phenomenon; Cambridge University Press: Cambridge, 2008.
12. Oran, E.; Gamezo, V. N.; Ogawa, T. The laminar-to-Turbulent-Detonation Transition: Hot spots, Turbulence and Stochasticity. In 61st Annual Meeting APS Division of Fluid
Dynamics, San Antonio, TX, November, 2008.
13. Shepherd, J. E. Detonation in Gases. P. Combust. Inst. 2009, 32, 83–98.
14. Ciccarelli, G.; Dorofeev, S. Flame Acceleration and Transition to Detonation in Ducts. Prog. Energy Combust. Sci. 2008, 34, 499–550.
15. Gamezo, V. N.; Zipf, R. K.; Sapko, M. J.; Marchewka, W. P.; Mohamed, K. M.; Oran, E. S.; Kessler, D. A.; Weiss, E. S.; Addis, J. D.; Karnack, F. A.; Sellers, D. D. Detonability of
Natural Gas–Air Mixtures. Combust. Flame 2012, 159, 870–881.
16. Di Benedetto, A.; Cammarota, F.; Di Sarli, V.; Salzano, E.; Russo, G. Anomalous Behaviour During Explosions of CH4/O2/N2/CO2 Mixtures. Combust. Flame 2011, 158,
2214–2219.
17. Oran, E. S.; Williams, F. A. The Physics, Chemistry and Dynamics of Explosions. Philos. Trans. R. Soc. A 2012, 370, 531–532.
18. Eckhoff, R. K. Dust Explosions in the Process Industries, 3rd ed.; Gulf Professional Publishing/Elsevier: Boston, MA, 2003.
19. Eckhoff, R. K. Understanding Dust Explosion: The Role of Powder Science and Technology. J. Loss Prevent. Proc. 2009, 22, 105–116.
20. Eckhoff, R. K. Influence of Dispersibility and Coagulation on the Dust Explosion Risk Presented by Powders Consisting of nm-Particles. Powder Technol. 2013, 239, 223–230.
21. Eckhoff, R. K. Current Status and Expected Future Trends in Dust Explosion Research. J. Loss Prevent. Proc. 2005, 18, 225–237.
22. Geurts, B. J.; Vreman, B. Dynamic Self-Organization in Particle-Laden Channel Flow. Int. J. Heat Fluid Flow 2006, 27, 945–954.
23. Skjold, T. Review of DESC Project. J. Loss Prevent. Proc. 2007, 20, 291–302.
24. Proust, C. A Few Fundamental Aspects About Ignition and Flame Propagation in Dust Clouds. J. Loss Prevent. Proc. 2006, 19, 104–120.
25. Zhang, F.; Gronig, H.; van de Ven, A. DDT and Detonation Waves in Dust–Air Mixtures. Shock Waves 2001, 11, 53–71.
26. Bradley, D.; Mitchelson, A. The Venting of Gaseous Explosions in Spherical Vessels. I – Theory. Combust. Flame 1978, 32, 221–236.
27. Metghalchi, M.; Keck, J. C. Laminar Burning Velocity of Propane–Air Mixtures at High Temperature and Pressure. Combust. Flame 1980, 38, 143–154.
28. Dahoe, A. E.; Zevenbergen, J. F.; Lemkowitz, S. M.; Scarlett, B. Dust Explosion in Spherical Vessels: The Role of Flame Thickness in the Validity of ‘Cube-Root Law’. J. Loss
Prevent. Proc. 1996, 9, 33–44.
29. Bartknecht, W. Explosions; Springer: Berlin/Heidelberg/New York, 1981.
30. Lewis, B.; Von Elbe, G. Combustion, Flames, and Explosions of Gases; Academic Press: New York, 1987.
31. Li, Y.-C.; Kauffman, C. W.; Sichel, M. An Experimental Study of Deflagration to Detonation Transition Supported by Dust Layers. Combust. Flame 1995, 100, 505–515.
32. Cashdollar, K. L. Overview of Dust Explosibility Characteristics. J. Loss Prevent. Proc. 2000, 13, 183–199.
33. Di Benedetto, A.; Russo, P.; Amyotte, P.; Marchand, N. Modelling the Effect of Particle Size on Dust Explosions. Chem. Eng. Sci. 2010, 65, 772–779.
34. Amyotte, P. R.; Pegg, M. J.; Khan, F. I. Application of Inherent Safety Principles to Dust Explosion Prevention and Mitigation. Process Saf. Environ. Prot. 2009, 87, 35–39.
35. Di Sarli, V.; Di Benedetto, A.; Russo, G. Using Large Eddy Simulation for Understanding Vented Gas Explosions in the Presence of Obstacles. J. Hazard. Mater. 2009, 169,
435–442.
36. Cooper, M. G.; Fairweather, M.; Tite, J. P. On the Mechanisms of Pressure Generation in Vented Explosions. Combust. Flame 1986, 65, 1–14.
37. Tamanini, F. Turbulence Effects on Dust Explosion Venting. Plant/Oper. Progr. 1990, 9, 52–60.
38. Rota, R.; Canu, P.; Carra, S.; Morbidelli, M. Vented Gas Deflagration Modeling: A Simplified Approach. Combust. Flame 1991, 85, 319–330.
39. Molkov, V. V. Unified Correlations for Vent Sizing of Enclosures at Atmospheric and Elevated Pressure. J. Loss Prevent. Proc. 2001, 14, 567–574.
40. Puttock, J. S.; Yardley, M. R.; Cresswell, T. M. Prediction of Vapour Cloud Explosions Using the SCOPE Model. J. Loss Prevent. Proc. 2000, 13, 419–430.
41. Zalosh, R. Explosion Venting Data and Modeling: Literature Review; The Fire Protection Research Foundation: Quincy, MA, 2008.
42. Amyotte, P. R.; Eckhoff, R. K. Dust Explosion Causation, Prevention and Mitigation: An Overview. J. Chem. Health Saf. 2010, 17, 15–28.
43. Ferrara, G.; Di Benedetto, A.; Salzano, E.; Russo, G. CFD Analysis of Gas Explosions Vented Through Relief Pipes. J. Hazard. Mater. 2006, 137, 654–665.
44. Molkov, V.; Makarov, D.; Verbecke, F.; Mansurov, Z.; Zhumabaev, M. LES Model of Vented Explosion: Hydrogen–Air Mixtures. In Proceedings of the 5th International Seminar on
Fire and Explosion Hazards, Edinburgh, UK, 23–27 April 2007.