Scholarlink

<p>This study aims to model and optimize the mixing performance in a baffled cubic mixer, belonging to the tumbler category of mixers widely used in the pharmaceutical industry, using artificial neural networks (ANN), with a focus on understanding the effects of baffle geometry and operating parameters. Three baffle configurations were investigated to improve mixing homogeneity: (i) four flat baffles located at the midpoints of each wall, (ii) four flat baffles placed at each corner of the mixer, and (iii) a cross-shaped baffle (+) with four axial arms. Additional variables studied include baffle width and powder cohesion strength. Experiments were conducted under varying operational conditions, including rotational speeds ranging from 10 to 20&#xa0;rpm, fill levels between 35% and 50%, a top-bottom loading profile, and a fixed mixing time of 20&#xa0;min. A total of 80 experiments were performed to construct the ANN training dataset. The optimal ANN architecture, composed of 15 neurons in the hidden layer, achieved excellent predictive accuracy, with a mean squared error of 3.053 × 10⁻¹¹ and a coefficient of determination (R²) close to 1. Sensitivity analysis using Garson’s algorithm revealed that baffle width is the most influential factor, contributing 32% to the overall effect on mixing performance. This finding highlights the critical role of baffle design in enhancing particle flow and mixing uniformity. Therefore, selecting the optimal width for each baffle shape and position is essential to achieve the desired mixing performance. Notably, the use of baffles led to a reduction in the relative standard deviation (RSD) compared to the unbaffled case, clearly demonstrating the effectiveness of baffle insertion in enhancing mixing homogeneity. The results demonstrate the strong potential of ANN models for accurately capturing complex mixing behavior and guiding the design and optimization of mixing systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Modeling and Prediction of Pharmaceutical Mixing Performance in a Cubic Mixer Equipped with Baffles Using the Artificial Neural Network Model

  • 期刊论文

  • 发表日期：

    2025年9月24日

查看全文
  • Amina Bouhaouche,
  • Kamel Daoud

摘要

This study aims to model and optimize the mixing performance in a baffled cubic mixer, belonging to the tumbler category of mixers widely used in the pharmaceutical industry, using artificial neural networks (ANN), with a focus on understanding the effects of baffle geometry and operating parameters. Three baffle configurations were investigated to improve mixing homogeneity: (i) four flat baffles located at the midpoints of each wall, (ii) four flat baffles placed at each corner of the mixer, and (iii) a cross-shaped baffle (+) with four axial arms. Additional variables studied include baffle width and powder cohesion strength. Experiments were conducted under varying operational conditions, including rotational speeds ranging from 10 to 20 rpm, fill levels between 35% and 50%, a top-bottom loading profile, and a fixed mixing time of 20 min. A total of 80 experiments were performed to construct the ANN training dataset. The optimal ANN architecture, composed of 15 neurons in the hidden layer, achieved excellent predictive accuracy, with a mean squared error of 3.053 × 10⁻¹¹ and a coefficient of determination (R²) close to 1. Sensitivity analysis using Garson’s algorithm revealed that baffle width is the most influential factor, contributing 32% to the overall effect on mixing performance. This finding highlights the critical role of baffle design in enhancing particle flow and mixing uniformity. Therefore, selecting the optimal width for each baffle shape and position is essential to achieve the desired mixing performance. Notably, the use of baffles led to a reduction in the relative standard deviation (RSD) compared to the unbaffled case, clearly demonstrating the effectiveness of baffle insertion in enhancing mixing homogeneity. The results demonstrate the strong potential of ANN models for accurately capturing complex mixing behavior and guiding the design and optimization of mixing systems.