En
期刊文章
文章精选
当期目录
过刊浏览
高被引
高下载
特邀文章
专刊合集
待刊文章
Volume 105
您当前的位置：首页 > 期刊文章 > 过刊浏览 > Volumes 96-107 (2025) > Volume 105
Li, S., Wang, D., Yu, H., Guo, J., Chen, G., & Fan, L. (2025). Unraveling multi-parameter coupling dynamics and building a machine learning-based predictive model for viscous dissipation rate in pin-type stirred mills. Particuology, 105, 315-324. https://doi.org/10.1016/j.partic.2025.08.006
Unraveling multi-parameter coupling dynamics and building a machine learning-based predictive model for viscous dissipation rate in pin-type stirred mills
Shengdong Li a, Dexi Wang a, Honglei Yu b *, Jinyuan Guo a, Gong Chen b, Lin Fan a
a School of Mechanical Engineering, Shenyang University of Technology, Shenyang, 110870, China
b School of Chemical Equipment, Shenyang University of Technology, Liaoyang, 111000, China
10.1016/j.partic.2025.08.006
Volume 105, October 2025, Pages 315-324
Received 13 June 2025, Revised 8 August 2025, Accepted 12 August 2025, Available online 20 August 2025, Version of Record 3 September 2025.
E-mail: braveyhl@163.com
Highlights

• Reveal different nonlinear influence laws of geometric parameters of agitator on grinding effect.

• Discover importance of variable group size to the fitting accuracy of geometric parameters.

• Analyze multi-factor coupling mechanism and weight ranking based on fusion of machine learning and numerical simulation.

• Determine gradient boosting model as optimal model for predicting viscous dissipation rate and propose engineering guidance.


Abstract

The rod and pin stirred mill is a key device for micron-sized powder production, yet the quantitative understanding of its grinding mechanism under multi-parameter coupling remains insufficient. This study develops a coupled flow field model based on computational fluid dynamic to investigate how agitator diameter, shaft diameter, and rotational speed influence viscous dissipation. Results reveal a positive correlation between these parameters and viscous dissipation rate, following a power-law relationship. Specifically, the agitator diameter shows a two-stage linear effect, while the shaft diameter exhibits Gaussian-type nonlinear growth. Numerical simulation combined with machine learning enables sensitivity analysis, indicating that rotational speed has the most significant impact, followed by shaft diameter and agitator diameter. The Gradient Boosting model demonstrates the highest prediction accuracy. These findings provide a quantitative basis for the engineering design of high-performance stirred mills.

Graphical abstract
Keywords
Pin-type stirred mill; Numerical simulation; Machine learning; Viscous dissipation rate
文章导读