Search papers, labs, and topics across Lattice.
This paper introduces a hybrid CFD-DEM modeling approach to simulate solid particle erosion, capturing complex multiphase flows and particle dynamics. The high-fidelity simulation data, combined with experimental results, is used to train an AI-based fusion model within a digital twin platform. This fusion model corrects for limitations in traditional CFD predictions, leading to significantly improved erosion prediction accuracy.
By fusing AI with CFD-DEM simulations, this work achieves significantly more accurate erosion predictions than traditional methods, bridging the gap between simulation and real-world performance.
Accurate prediction of solid particle erosion is critical for the design and reliability of equipment in oil & gas, chemical, and energy sectors. This work presents a hybrid modeling approach that couples Computational Fluid Dynamics (CFD) and Discrete Element Method (DEM) to simulate complex multiphase flows with realistic particle dynamics, including shape effects, impact behavior, and transient interactions with surfaces. These high-fidelity simulations are used to generate rich datasets across varying flow and solid-loading conditions. To address the discrepancy between traditional CFD predictions and experimental results, the coupled simulation outputs are integrated into an AI-based fusion model built on an AI-augmented, simulation-based digital twin platform. By training on both synthetic and experimental data, the fusion model corrects for physical model limitations and significantly improves erosion prediction accuracy. The proposed methodology offers a more comprehensive understanding of erosion mechanisms and provides a powerful tool for design optimization, operational planning, and predictive maintenance. By merging data-driven intelligence with advanced physical modeling, this approach establishes a scalable and transferable solution for erosion analysis in real-world industrial systems.
