Abstract: Aluminum powder, as a high-energy-density metallic fuel, holds significant research value in detonation propulsion and dust explosion safety. To elucidate the mechanism by which aluminum particle size and its distribution affect the cellular structure of detonation waves, this study conducts systematic two-dimensional numerical simulations of aluminum–gas two-phase detonation based on the Eulerian–Lagrangian framework and a hybrid surface-kinetic-oxidation-diffusion combustion model. First, in the aluminum–air system, quantitative relationships between the cell size and particle parameters are established for both monodisperse and log-normally polydisperse suspensions. It is found that the logarithm of the cell size exhibits a linear correlation with both the logarithm of particle diameter and the variance of the size distribution. A unified conversion formula between monodisperse and polydisperse cell sizes is established and supported by a statistical-moment analysis of the log-normal distribution, with a correlation coefficient ranging from 2.0 to 2.5, which is consistent with theoretical expectations of the hybrid combustion model. Second, the investigation is extended to various reactive gas environments, including hydrogen/water vapor, methane/oxygen, and ethylene/oxygen mixtures. The quantitative law is shown to hold across all the reactive gas systems and parameter ranges examined, while the chemical activity of the gas is shown to significantly modulate the cell size and detonation instability by altering the heat release intensity. Finally, through one-dimensional detonation wave analysis and theoretical decoupling, it is revealed that the synchronized evaporation of particles in monodisperse systems leads to a concentrated, intense heat transfer phase, which is the microscopic origin of the "double-peak" pressure structure behind the wave and enhanced local cellular perturbation. In contrast, the asynchronous phase change in polydisperse systems weakens the concentrated heat transfer effect, resulting in a smoother post-wave flow field. This study establishes a cross-scale correlation linking particle size distribution, microscopic phase change, and macroscopic cellular structure, providing theoretical foundations and numerical support for hypersonic propulsion fuel design and industrial dust explosion risk assessment.