Scholarlink

<p>In centrifugal pumps, high axial thrust loads can lead to rapid thrust bearing wear, resulting in either sequential pump failure or frequent operational breakdowns. This study focuses on an experimental and numerical investigation of the effects of key parameters such as axial clearance, flow rate and impeller speed on axial thrust. An experimental setup was designed to examine the influence of these parameters, with data collected and analysed using the Taguchi method and statistical analysis techniques, including ANOVA. Response surface methodology (RSM) and artificial neural networks (ANNs) were employed to model complex parameter interactions and capture nonlinear relationships. Subsequently, machine learning algorithms such as teaching–learning-based optimization (TLBO) and genetic algorithms (GA) were applied to optimize the parameters for minimizing axial thrust and enhancing predictive accuracy. The findings reveal that axial thrust increases with greater axial clearance and impeller speed but decreases as the flow rate increases. The optimum parameter combinations for minimizing axial thrust were identified as 855 rpm, 0.6 mm and 12 L/min using GA, and 863 rpm, 0.6 mm and 12 L/min using TLBO. These results underscore the importance of carefully considering axial clearance, flow rate and impeller speed when designing centrifugal pumps to achieve improved performance and durability.</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

Experimental investigation of axial thrust in a centrifugal pump using design of experiments (DOEs) and machine learning algorithms

  • 期刊论文

  • 发表日期：

    2025年8月20日

查看全文
  • Kamal Singh,
  • Achhaibar Singh,
  • Dinesh Kumar Singh

摘要

In centrifugal pumps, high axial thrust loads can lead to rapid thrust bearing wear, resulting in either sequential pump failure or frequent operational breakdowns. This study focuses on an experimental and numerical investigation of the effects of key parameters such as axial clearance, flow rate and impeller speed on axial thrust. An experimental setup was designed to examine the influence of these parameters, with data collected and analysed using the Taguchi method and statistical analysis techniques, including ANOVA. Response surface methodology (RSM) and artificial neural networks (ANNs) were employed to model complex parameter interactions and capture nonlinear relationships. Subsequently, machine learning algorithms such as teaching–learning-based optimization (TLBO) and genetic algorithms (GA) were applied to optimize the parameters for minimizing axial thrust and enhancing predictive accuracy. The findings reveal that axial thrust increases with greater axial clearance and impeller speed but decreases as the flow rate increases. The optimum parameter combinations for minimizing axial thrust were identified as 855 rpm, 0.6 mm and 12 L/min using GA, and 863 rpm, 0.6 mm and 12 L/min using TLBO. These results underscore the importance of carefully considering axial clearance, flow rate and impeller speed when designing centrifugal pumps to achieve improved performance and durability.