Understanding detonation behaviours in non-uniform explosive mixtures is one of the oldest and most challenging problem on explosion dynamics. Variable initial states are commonly encountered in most situations of practical interest such as safety of goods and people, advanced aerospace propulsion and defence security. They can be accidental such as leaks of hydrogen from pipes, tanks or the inner envelopes of nuclear reactors, those of hydrocarbon gases, the clouds of combustible particles in silos or mines. They can also result from technological constraints, such as the non-uniform distributions of fuel, oxidizer and burned gases in chambers of rotating detonation engines (RDE). The safe distribution and use of hydrogen combined with more thermally efficient propulsion modes are one of the thematic sets favoured by the scientific community to contain global warming. Obviously enough, anticipating all configurations of non-uniform explosive distributions is basically impossible. Rather, this project addresses four generic configurations designed as representative limits of a large range of real-life situations that consider one-phase gaseous or two-phase gas-liquid fluids. - Initial layers of composition normal to the direction of the detonation propagation. The configuration retained in this work is the succession of vertical layers of inert and premixed reactive gases normal to the propagation direction of the detonation. This is an idealization of jet-injection technologies (i) in RDE chambers when the jets are parallel to the exhaust, since the detonation rotates azimuthally, and (ii) for the mitigation of deflagrations or detonations that accidentally propagate in reactive gases. This work will address this configuration with either jets of an inert gas in a premixed reactive composition or jets of premixed reactive gas in an inert gas (subtask 3.2). - Initial layers of composition parallel to the direction of the detonation propagation. The configuration retained in this work is two superimposed horizontal layers parallel to the propagation direction of the detonation, with reactive or inert, liquid or gaseous fluids. This is an idealization of the stratified layers that can be formed accidentally from slow leaks from tanks and pipes, or designed specifically to shape discontinuity waves. This work will address this configuration with two gaseous layers, one inert and the other premixed-reactive (subtask 3.1), an inert liquid layer below a premixed-reactive gaseous layer (subtask 4.1), and a liquid fuel layer below a premixed-reactive gaseous layer (subtask 4.2). This project thus aims at contributing to compressible reactive fluid dynamics with reliable and controlled experimental observations on several interface phenomena specific to high-pressure, high-velocity physics that include oblique and normal shock interactions, mixing, vaporization, and fragmentation. The importance of the latter phenomena is that they drive the physics of both the causes and the effects of deflagration and detonation existences in many configurations of practical interest. The results will also contribute help numerical simulations to get more realistic if not predictive.