OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible. This is an author-deposited version published in : http://oatao.univ-toulouse.fr/ Eprints ID : 19758To link to this article : DOI : 10.1007/s00193-018-0814-4 URL : http://dx.doi.org/10.1007/s00193-018-0814-4Any correspondence concerning this service should be sent to the repository administrator: staff-oatao@listes-diff.inp-toulouse.fr AbstractAfter the detonation of an oxygen-deficient homogeneous high explosive, a phase of turbulent combustion, called afterburning, takes place at the interface between the rich detonation products and air. Its modelling is instrumental for the accurate prediction of the performance of these explosives. Because of the high temperature of detonation products, the chemical reactions are mixing-driven. Modelling afterburning thus relies on the precise description of the mixing process inside fireballs. This work presents a joint numerical and experimental study of a non-reacting reduced-scale set-up, which uses the compressed balloon analogy and does not involve the detonation of a high explosive. The set-up produces a flow similar to the one caused by a spherical detonation and allows focusing on the mixing process. The numerical work is composed of 2D and 3D LES simulations of the set-up. It is shown that grid independence can be reached by imposing perturbations at the edge of the fireball. The results compare well with the existing literature and give new insights on the mixing process inside fireballs. In particular, they highlight the fact that the mixing layer development follows an energetic scaling law but remains sensitive to the density ratio between the detonation products and air.

Small-scale experiments are a good means of carrying out explosion and shock wave measurements. Commonly, the shock wave is tracked thanks to pressure sensors and sometimes with a high-speed camera. In the present study, these methods were used to analyze the interaction of a shock wave with an obstacle of simple geometry. The primary aim of the study was to demonstrate the need to correlate these different methods in order to analyze certain phenomena related to the three-dimensional interaction of a shock wave with an object. The correlation between the overpressure and the visualization made it possible to carry out a complex analysis. The visualization was carried out simultaneously on two planes, from the front and top views, thanks to the optical setup. Shock wave characteristics were taken at ground level downstream of the obstacle with pressure gauges. The correlation of the images obtained allows the identification of the waves on the profile and their contribution in intensity.

Computation of gas dispersal in urban places or hilly grounds requires a large amount of computer time when addressed with conventional multidimensional models. Those are usually based on two-phase flow or Navier-Stokes equations. Different classes of simplified models exist. Among them, two-layer shallow water models are interesting to address large-scale dispersion. Indeed, compared to conventional multidimensional approaches, 2D simulations are carried out to mimic 3D effects. The computational gain in CPU time is consequently expected to be tremendous. However, such models involve at least three fundamental difficulties. The first one is related to the lack of hyperbolicity of most existing formulations, yielding serious consequences regarding wave propagation. The second is related to the non-conservative terms in the momentum equations. Those terms account for interactions between fluid layers. Recently, these two difficulties have been addressed in Chiapolino and Saurel (2018) and an unconditional hyperbolic model has been proposed along with a Harten-Lax-van Leer (HLL) type Riemann solver dealing with the non-conservative terms. In the same reference, numerical experiments showed robustness and accuracy of the formulation. In the present paper, a third difficulty is addressed. It consists of the determination of appropriate drag effect formulation. Such effects also account for interactions between fluid layers and become of particular importance when dealing with heavy-gas dispersion. With this aim, the model is compared to laboratory experiments in the context of heavy gas dispersal in quiescent air. It is shown that the model accurately reproduces experimental results thanks to an appropriate drag force correlation. This function expresses drag effects between the heavy and light gas layers. It is determined thanks to various experimental configurations of dam-break test problems.