A three-layer digital twin architecture for granular bed profile prediction in rotary kiln

Granular bed profile in rotary kilns plays a critical role in determining heat transfer, gas distribution, and material residence time. However, due to enclosed high-temperature and rotating conditions, accurately and rapidly predicting the bed profile remains a significant challenge. To overcome this, this paper proposes a three-layer digital twin (DT) architecture for predicting the bed profile in rotary kiln. The first layer is the modeling layer, which integrates mathematical mechanism models of material motion into a simulation core to generate simulation data of kiln processing. The second layer is the data layer, where material outlet height data is collected and used to iteratively update simulation results, improving prediction accuracy. The third layer is the interaction layer, where a virtual terminal is designed to retrieve and visually present the results. A small-scale rotary kiln test platform is applied to validate proposed DT. Results indicate that the proposed DT architecture can effectively predict granular bed profile in both steady or dynamic states within a very short time. This study provides a potential solution for modeling and predicting granular flow in rotary kilns and shows promise for broader engineering applications with further optimization.