Advancing 3D understanding of aluminum agglomerates in propellant environment using reconstruction techniques

Aluminum is commonly used in solid propellants to enhance energy density, but it tends to form large agglomerations during combustion, leading to incomplete combustion and reduced propulsion efficiency. To better understand the agglomeration and combustion behavior of aluminum particles, we developed a dual-perspective high-speed microscopic imaging system that captures the evolution of agglomerations from two aligned viewpoints. A two-step unsupervised segmentation algorithm based on K-means clustering and a Neural Radiance Field (NeRF)-based reconstruction framework were employed to resolve the 3D distribution of molten aluminum droplets and oxide caps. The results revealed a linear increase in the oxide-to-metal ratio over time. A combustion model was further developed to describe the burning process of aluminum particles in multi-component oxidizing atmospheres, incorporating O₂, CO₂, and H₂O as oxidants. The model assumes diffusion-limited combustion with oxide deposition, and was validated against literature and experimental data, showing good agreement in predicting particle size evolution and burning time. Sensitivity studies showed that oxidizer concentration has a significantly greater impact on combustion rate than temperature. The proposed imaging and modeling approach improves the understanding of aluminum agglomerate evolution in realistic propellant environments, and provides valuable guidance for optimizing propellant formulations to reduce incomplete combustion and improve solid rocket motor performance.