Industrial fires and explosions initiating from self-ignition of combustible porous dust deposits represent a serious hazard for human beings, environment and industry. Understanding the fundamental basis of combustible dust ignition behaviours at different geometries is of importance to prevent and mitigate the accidental risks. A correlation of self-ignition temperatures (SITS) measured by hot-oven tests and minimum ignition temperatures of dust layers (MITIs) determined by hot-plate tests has been established previously. However, this analogy approach based on Frank-Kamenetzkii model is limited by ignoring the influence of oxygen diffusion. In this work, an improved method is developed by implementing a correction factor for the pre-exponential factor caused by the boundary geometry. This method is testified by comparing with experimental data, previous analogy method and numerical simulation. Results show that our proposed method performs a better predictability of MITLs and simplicity. The improved analogy method indicates that the different boundary geometries of a dust deposit significantly impact the apparent pre-exponential factor, while have an ignorable influence on the activation energy, which is also verified by numerical investigations. Furthermore, the numerical model with the corrected kinetic parameters provides a satisfactory explanation compared with experimental observations regarding to temperature and concentration evolutions of dust layers.