This action addresses the combustion of metals, which is involved in various industrial fields such as metallurgy, aeronautics, nuclear and gas industry. The associated hazards have motivated the study over the few past years of their ignition, combustion and extinction. Mostly experimental, these studies aim at understanding the conditions under which the combustion of a metallic component exposed to high pressure and oxygen concentration may initiate and propagate. It has been shown in particular that three main steps can be identified in the combustion of iron or steel rods. During an induction period, the sample is heated until temperature allows a significant superficial oxidation reaction. A phase of static combustion follows, in a relatively small volume of liquid metal and oxide. Finally, dynamic combustion takes place, when the liquid mixture flows and leaves its initial reaction zone. The current knowledge cannot explain the details of these observations, due to the great variety of the involved physico-chemical processes (transport phenomena, phase changes, homogeneous and heterogeneous reactions), to their interactions (identification of the limiting processes), and in most part to the lack of knowledge regarding the exact mechanisms and kinetic laws of the metal/oxygen reaction.

For these reasons, a macroscopic modelization appeared as a more promising and amenable approach to rationalize the observations. A first step along these lines has been taken in collaboration with Pr. Kirill Kazakov (LABEX invited professor – Moscow State University, 09-12/2012). A consistent model has been established, which can account for all the aspects of the experimental results. Furthermore, it has been shown that two kinds of transitions from the static to dynamic combustion regimes can exist. A fast transition occurs when the oxygen flow velocity is large enough so that the shear-induced vorticity created at the liquid/gas interface balances the vorticity induced by baroclinic effect. Otherwise, the transition to dynamic combustion is slow, with a characteristic time proportional to the square of the liquid thickness and inversely proportional to its kinematic viscosity.

 

Three important results have been obtained regarding the dynamic combustion regime.

• The oxide fraction f in the liquid phase is given by (Q_m + c_p (T_m-T₀))/Q. For iron, Q_m=250 J/g, Q=6 kJ/g and c_p=0.45 J/g K, which yields f=15.5%, in very good agreement with data from the literature.
• The normal combustion velocity is given by 2 v*c₀ / f n, where n is the density of metal atoms in the liquid phase, c₀ the oxygen concentration in the incident gas flow and v* the frictional velocity at the gas/liquid interface. This yields u_n=0.54 cm/s for iron, close to the experimental value of 0.61 cm/s.
• In the absence of flow, the normal combustion velocity is proportional to the square root of the oxygen partial pressure. This is a remarkable result, considering that such dependence is generally attributed to sorption reactions that take place at liquid/solid or liquid/gas interfaces.

Complementary experiments have been conducted during year 2014, in collaboration with PIMM laboratory (UMR 8006), in order to refine the modelization developed in 2012. In parallel, a new collaboration with Grigory Ermolaev (Khristianovich Institute of Theoretical and Applied Mechanics, Novosibirsk, Russia) has been initiated, regarding the kinetic aspects of metal combustion.

---