of non-uniform gaseous mixtures I. Sochet1,2 , T. Lamy2 , J. Brossard2 1 Laboratoire Energétique Explosion Structure (LEES), 63 Av. De Lattre De Tassigny, 18020 Bourges cedex, France 2 Ecole Nationale Supérieure d’Ingénieurs de Bourges (ENSIB), 10 Bd Lahitolle, 18020 Bourges cedex, France
Received 19 October 1998 / Accepted 15 July 1999
Abstract. An experimental investigation is performed to characterize the detonability of small gaseous
clouds with a concentration gradient. Two types of gaseous mixtures are used: (i) a heavy gas C3 H8 /O2 (equivalence ratio φ: 0.70 ≤ φ ≤ 1.66); (ii) a light gas H2 /O2 (0.50 ≤ φ ≤ 1.25). The mixtures are initially confined in a hemispherical volume which is characterized by an initial radius R0 = 0.04 ÷ 0.08 m. When the confining is ruptured, the gaseous mixture diffuses into the surrounding air. The concentration distri- bution is a result of molecular diffusion, gravity and turbulence. Schlieren chronophotographies enable the illustration of the dispersion of the cloud. By means of pressure profiles of blast waves generated by the explosion, the limit between the two explosion phenomena (total and partial explosive charge) is defined. The limit time delay, which leads to a given concentration distribution and for which detonations cannot be observed, is investigated with respect to initial gaseous composition and initial volume of confining. The critical nominal initiation energies in uniform and non-uniform media are characterized.
1 Introduction From a theoretical point of view, the characterization
of shock wave propagation in non-uniform media has been principally carried out by various, mainly Russian research Accidental explosions include a variety of modes of pres- teams. Numerous works, such as Bird (1961), Zel’dovich sure wave propagation resulting from a detonation or a and Raizer (1967), are mainly devoted to powerful ex- deflagration wave. As a flame is propagating in a com- plosions at high altitude and are related to temperature bustible gas, the burned gases create a pressure wave dur- and/or density gradients. Bird (1961) considers the prop- ing their expansion, in the initial unburned gas. If the agation of a shock wave in a stationary gas in which the pressure wave is weak, then the combustion front propa- density varies linearly. The shock strength changes in the gates at a subsonic velocity and in an autonomous way non-uniform region. Reflections and successive reflected throughout the unburned gases. The propagation is real- waves are produced. These reflected waves have a consid- ized by the coupling of the chemical reaction and heat and erable influence on the rate of change of shock strength mass transfer phenomena resulting in a deflagration wave. for increasing density. Zel’dovich and Raizer (1967) focus For a strong pressure wave, the shock wave increases the on propagation of shock waves in a non-homogeneous at- temperature and pressure of the unburned gas: this gas is mosphere with an exponential density distribution. They called shocked gas. It allows the initiation of the explosive show that the shock front surface and the strength of the reaction. The energy released by the reaction sustains the shock wave do not change in the same way in different shock wave. Hence, the shock wave and the combustion directions: in particular, in the direction of the density wave support each other. The self-sustained propagation decrease, the shock wave is accelerated. Since then, many can be considered as the coupling of the shock wave and developments have been proposed taking into account the the reaction zone. The coupling propagates at supersonic counterpressure effect. The study of shock wave propaga- velocity with respect to the upstream unburned gas. In tion in reactive non-uniform media is very complex since practical initial conditions of an explosive cloud, the deto- the reactivity is a function of the non-uniformity. Sev- nation is not instantaneously created. The ignition process eral authors suggest different possibilities to create non- generates a deflagration wave which can possibly acceler- uniformity in reactive mixtures: (i) by temperature gradi- ate and lead to a detonation wave. ent; (ii) by density gradient; (iii) across an inert region. Zel’dovich et al. (1970) are the first researchers to raise Correspondence to: I. Sochet clearly a criterion to initiate a detonation in non-uniform (e-mail: Isabelle.Sochet@ensi-bourges.fr) 364 I. Sochet et al.: Experimental investigation on the detonability
media. Initially, Zel’dovich et al. (1970) investigated the
conditions for the onset of detonation waves in reacting gases, maintained initially at non-uniform temperature (linear distribution). Three regimes of the development of chemical reaction were proved to be possible: thermal explosion, formation of a shock wave propagating away from the reaction zone and a steady state detonation wave. These regimes depend on the temperature gradient, which can be small, steep or intermediate. Later, Zel’dovich et al. (1988) considered the problem of the evolution of a two- component mixture with non-uniform initial distribution Fig. 1. Experimental setup of one component. Depending on the physical size of the non-uniformity domain, the kinetic parameters and the inert gas (He, Ar, CO2 , SF6 , Air) in a vertical tube by heat of reaction, various types of explosion process were Thomas et al. (1991). It appears that the governing pa- revealed. However, the numerical model of Dorofeev et al. rameters are: the width of the inert region when the con- (1989) shows the existence of a critical temperature dis- centration gradients at the interface are high, the inert tribution length. There is a minimum internal energy of concentration profile when the form of the inert region is preheated gas mass which initiates detonation. The value distributed by mixing, the reactivity of the gas, the heat of this energy is close to the minimum energy of blast ini- transfer and the wall friction. The whole of these experi- tiation and depends only on mixture properties. The con- mental studies confirm that the critical conditions for ini- tinuation of this analysis, investigated by Dorofeev et al. tiation and propagation processes are difficult to predict (1995), included multi-dimensional numerical modelling in a reliable manner. and the effect of non-uniformity shape: the minimum non- Hence, from these theoretical and experimental stud- uniformity length does not depend strongly on the distri- ies and with a view to contributing to industrial safety, bution function (gaussian, polygonal). Following the same it would appear necessary to extend the knowledge of the idea, He (1991) shows that the Zel’dovich criterion is a non-uniform gaseous explosion processes by experimental necessary condition but not sufficient to initiate a detona- investigations. This problem is related to practical situa- tion. He investigated the effect of an initial non-uniform tions in which gaseous mixtures are ignited accidentally, temperature distribution in a limited environment (heat in some cases leading to the formation of detonation waves pocket) in a uniform cold gas. The numerical model leads and consequently causing major accidents. This is why we to the critical size of a heat pocket of gas which can initiate develop our first experimental and theoretical investiga- a self-propagating detonation in a cold gas. More recently tions (Sochet et al. 1993; Sochet and Brossard 1994) based Khokhlov et al. (1997) studied deflagration-to-detonation on a compositional gradient, as would exist in an acciden- transition (DDT) in non-uniform media. They considered tal gas cloud. The basic idea of this paper is to characterize the process of initiation of detonation that arises from the detonation limit of a small explosive gaseous charge the explosion of a reactive gas with a non-uniform distri- diffusing into surrounding calm air. We describe the ex- bution of induction times. The non-uniformity is consid- perimental technique and present a synthesis of numerous ered as the result of the mixing of high-entropy products experimental results as a function of the initial and ini- and low-entropy unreacted fuel. The criterion of DDT in tiation conditions of the non-uniform explosive gaseous unconfined flames was formulated in terms of the one- mixture. dimensional thickness of the Chapman-Jouguet detona- tion (xd ), the laminar flame speed and the laminar flame thickness in the reactive gas. Khokhlov et al. (1997) show that there are two key elements to the 2 Experimental setup theory for uncon- fined DDT: (i) the size Ld 103 xd of the region that The similitude laws are representative of physical phe- can trigger DDT in a mixture of hot burning product and nomena involved in the detonation process. Consequently, fuel; (ii) the intensity of turbulent motions required for the choice of small scale experiments are clearly justified the region of size Ld to undergo DDT. (Brossard et al. 1988). The experimental setup used in this study is similar to that described in previous papers From an experimental point of view, a few studies have (Sochet et al. 1993; Sochet and Brossard 1994). The ex- been conducted. Donato et al. (1981) investigated a one- perimental device (Fig. 1) allows the following: dimensional detonation wave propagating through a fuel concentration gradient obtained by diffusion from a lean – to obtain an explosive gaseous charge with a space- mixture into a rich one and conversely from a rich mix- time dependant concentration (Sochet and Brossard ture into a lean one. It resulted that the detonation adjusts 1994); itself immediately to the compositional gradient and con- – to study the detonability conditions of this charge; sequently to the conditions of local energy release. The – to characterize the shock wave generated by the deto- detonation reinitiation across an inert region of air in a nation. horizontal tube had been highlighted by Bjerketvedt et The uniform gaseous mixture is initially confined in al. (1986) and Teodorczyk et al. (1996), and across an a hemispherical volume (soap bubble). The initial radius