Analysis of the effectiveness of magnetic filters of shovel hydraulic system and successive hydraulic failure prediction using data-driven approaches

• Author / Creator

  Odeyar, Prerita

• Surface mining dominates the world's production of minerals. Currently, almost all non-metallic minerals (more than 95 %), most metallic minerals (more than 90%), and a significant fraction of coal (more than 60%) are mined by surface mining methods. The hydraulic shovel-truck system forms the backbone of the surface mining industry. With the complexity of the shovel and the tough working conditions, hydraulic system components are more susceptible to failures and are often the most expensive system to repair. The failures are highly unpredictable, associated with high follow-on failure rates, and cause lengthy downtime. Wear and debris-related failures are the most common cause of hydraulic failures. The most common contaminant of wear and debris failures is iron contamination which is a constant issue as it is the by-product of machine operation and component failures. As the component wear gradually increases in the hydraulic system, it leads to debris accumulation in the oil that might trigger multiple component failures allowing the contamination to spread rapidly, resulting in catastrophic damage. Hence, companies are constantly aiming to improve the hydraulic system condition and prevent major failures by using different methods like condition monitoring methods, increasing hydraulic filter capacities, and implementing new methods like introducing new filters and using statistical data-driven techniques to predict and stop failures.
  Magnetic filters are the newly introduced hydraulic filters that use the most advanced magnetic technology to prevent contamination and system wear failures. The first part of the research aims to quantitively evaluate magnetic filter performance on hydraulic system components and test their effectiveness using failure data of hydraulic shovels.
  The second part of the research focuses to further enhance the mitigation of catastrophic hydraulic failures and increase component life with the use of data-driven techniques. The aim is to assess the likelihood of successive failures of hydraulic components in the next 1000 hours of operation after a component failure. Historical failures are studied using different machine learning algorithms and probability of successive failures are predicted based on the failure patterns identified.

• Subjects / Keywords

  • Data-Driven
  • Hydraulic filters
  • reliability
  • Machine Learning
  • Mine Equipment

• Graduation date

  Fall 2022

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-w4qj-s097

• License

  This thesis is made available by the University of Alberta Library with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.