Aluminum powder as recyclable energy carrier : population balance modelling of oxide smoke formation

Abstract

Against the background of our society’s endeavour towards a sustainable energy economy, in the past decade, a variety of different technologies has emerged to tackle the challenge of global warming. While particular concepts have already been developed to maturity, metal powders have recently been hypothesized as recyclable and carbon-free energy carriers as part of an on-demand oxidation-reduction cycle, on the consumer side of which the metal powder is burned while releasing heat and condensed oxides as main reaction products. This thesis focuses on aluminum which not only qualifies as a potential metal fuel due to its high energy density, availability and handling safety, but has also been investigated in detail for its reaction kinetics. At high temperatures, aluminum particles can be burned exothermally in oxidizing atmospheres in a similar way to carbon-based particulate fuels. The produced aluminum oxide is solid under ambient conditions and may form through two distinct chemical pathways. On the one hand, vaporized aluminum initiates homogeneous gas phase combustion, leading to the condensation of aluminum oxide into very fine smoke droplets. On the other hand, heterogeneous reactions at the particle surface cause a direct conversion of aluminum into aluminum oxide, rendering the fuel particle biphasic. At present, the formation of oxide smoke fines poses major challenges to the oxide recovery from dust flames and the closure of the metal fuel cycle. The smoke’s size distribution influences natural deposition and emission mechanisms and, consequently, plays a decisive role in the design of gas-particle separation devices. With the objective of elucidating the oxide smoke dynamics and identifying operating conditions promoting the formation of larger smoke droplets, we propose a comprehensive modelling approach that permits a prediction of the smoke size distribution alongside gas and particle surface compositions. Physically, on the spatially localized level, the oxide droplet size distribution is influenced by the ambient gas phase composition and shaped by the mutual competition of nucleation, condensational surface growth, evaporation/dissociation and coagulation. The heat release and dispersion temperature, on the other hand, are affected by chemical reactions, phase transition and radiation. In this thesis, the oxide smoke droplets are described in a Eulerian fashion by harnessing a population balance description that is informed by a complete set of droplet formation and interaction kinetics and allows for analyzing the interaction, competition and mutual reinforcement of the relevant physical processes. In a first step, the population balance framework is applied in a perfectly stirred reactor and a partially stirred reactor. These simplified model formulations are representative of the dynamics in a single grid cell of a spatially inhomogeneous laminar reactive flow solver or a one- point, one-time probability density function description. As key novelty in this context, the partially stirred reactor model is extended to account for the presence of a reactive surface with small-scale variability in terms of surface composition. Detailed gas phase and heterogeneous surface kinetics, including NOx formation, are taken into account. In a next step, we present a fully Eulerian framework for modelling the combustion of a single spatially resolved aluminum particle. In order to describe the reacting gas-droplet dispersion, we combine the population balance equation governing the smoke size distri- bution with tailored balance laws for gas phase species as well as the dispersion mass, mo- mentum and enthalpy, while the detailed kinetic framework is augmented by the transport parameters governing species differential diffusion, droplet diffusion and thermophoresis. The major novelties of our physical model lie with the prediction of the smoke size distri- bution at every location in the flow domain, the accommodation of size-sensitive kinetics and transport processes as well as the prediction of possible NOx pollutants in a spatially resolved fashion. Based on a comparison of our model predictions with available experi- mental measurements, we calibrate the droplet formation kinetics and, finally, validate the model while attesting a very good agreement. Ultimately, the spatio-temporally resolved single particle model is instrumented to estimate the emissions of a burning aluminum particle over the course of its conversion. Here, a particular feature is the incorporation of a time-varying particle morphology, including an oxide lobe. Our predictions of the particle burning times and residue sizes for different initial particle diameters are found to agree well with available experimental data. In order to demonstrate the controllability of the combustion products’ sizes, an analysis of the effect of a varying pressure on our predictions is performed for all three configurations. Lastly, we present the fundamentals of a modelling framework for turbulent metal dust flames encountered in practically relevant metal dust burners. Within the scope of a partially stirred reactor, the model is shown to allow for the individual assessment of polydispersity and turbulence influencing the particle-laden flow.