Usage
  • 44 views
  • 55 downloads

Taming entropy: designing, evaluating and applying decision-support systems for risk management in industrial and non-industrial settings

  • Author / Creator
    Naef, Michelle E M
  • After being assigned a process fire investigation in my second year of professional practice, and later working in maintenance leadership following multiple fatality incidents, I wondered why process safety outcomes did not seem to be improving despite constant discussion and an incredible pace of technological innovation in both automation and information systems. I pursued research in the field of chemical engineering and risk management to answer the question: why are process safety incident frequencies rising despite advanced process control and rapidly improving technology? This question seemed to be a classic process control problem, a “delta” or difference between the way things are and the way business and operations leaders said things should be. Over a decade of experience working in process facilities and control rooms had provided clues to the nature of the underlying problem, and I focused my research on the latent value I believe exists in improving the quality of human decisions at the human-computer interface.

    The research summarized in the following dissertation began with a small pilot study where I pursued a practical application of a novel predictive analytic method developed by Suresh et al., (2019). Using the layered approach to user interface assessment, I developed a visual representation of the causal maps generated using the hierarchical approach and asked participants to narrate their problem-solving process when presented with a simulated process fault. It became clear that developing a defensible, reliable method for evaluating risk-based decisions at the human-computer interface could be a valuable tool in determining what types of information system and automation interface features had real utility to field workers. I developed a large-scale study demonstrating an application of the situational design model to a process industry challenge. 35 participants faced an abnormal situation similar to the one presented to the pilot study participants. The instruments and techniques applied can be further refined and adapted to offer significantly more analytic power to design feature assessment and to industrial training and assessment processes.

    In a parallel research stream, I examined the structure of information systems in the process industry, observing that many of the classification structures used in analytics and in human-centric risk management activities were similar, but not standard. Miscommunication and challenges in organizing relevant data seemed to be persistent challenges that limited the effectiveness of these efforts and required frequent intervention by subject matter experts. I wondered how asset management ontology knowledge bases could be applied more effectively to improve understanding of process systems in risk management contexts, and potentially improve the efficacy of predictive analytics by reducing the number of spurious or misleading connections made by algorithms and optimization processes. I developed a case study and applied an asset management ontology as a communication and learning tool for a group of stakeholders engaged in COVID-19 response planning in a commercial building. The results demonstrated the value of industrial risk management approaches to a new application, and the use of ontology knowledge bases as a risk communication tool.

    Major loss incidents in the process industry are complex, multi-variate networks that connect people, assets, environment, and information systems. Mapping these connections, uncovering the unexpected consequences of different interactions and better equipping decision-makers at the front line is a necessary step in reducing the frequency of major loss incidents. This research offers theoretical, methodological and empirical contributions supporting new directions for industry in pursuing safer, more reliable operation. 

  • Subjects / Keywords
  • Graduation date
    Fall 2023
  • Type of Item
    Thesis
  • Degree
    Doctor of Philosophy
  • DOI
    https://doi.org/10.7939/r3-pxg2-db34
  • License
    This thesis is made available by the University of Alberta Libraries 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.