Analyzing the risks associated with railway transportation of hazardous materials and developing process models for railway incidents with high potential for release using machine learning and data analytics

• Author / Creator

  Ebrahimi, Hadiseh

• The Canadian economy relies heavily on its transportation network. It supports hundreds of thousands of jobs, contributes billions to the economy, and facilitates the movement of goods within the country as well as internationally. Railways provide affordable and efficient transportation to over 84 million passengers each year, and they transport approximately 70% of all intercity surface freight and half of Canada's exports. Rail transportation of hazardous materials is an activity that is important to most industries but is commonly associated with the oil and manufacturing sectors. Hazardous materials (hazmat) are defined as explosives, flammable and combustible substances, toxic substances, oxidizing substances, and corrosive substances, among others. Between 2011 and 2017, the quantities of fuels and chemicals transported by Class 1 railways (Canadian National Railway, CN, and Canadian Pacific Railway, CP) increased by 42.5%. Railway incidents transporting hazmat can have severe consequences for people that require mitigation, especially in areas where there is a high population density. In order to prevent and minimize the negative impacts of railway incidents, risk assessment is key to planning and improving safety. Several factors contribute to the risk analysis of hazmat transportation, such as hazmat-related incident rates in transport infrastructure, the consequences of hazmat release, and the probability of hazmat release. The objectives of this study are developed based on these factors of the risk assessment.
  The primary objective of this study is to identify the impact of human factors on the likelihood of railway incidents and to identify the leading factors and their associations. It is determined that most deficiencies occurred in the areas of organizational oversight, supervision, and organizational culture. In addition, supervisory and organizational factors are highlighted as important factors in the prevention of railway loss incidents. The second objective of this study is to develop and illustrate with a case study a methodology for developing enhanced risk maps. According to the risk maps, land-use planning should consider the appropriate allocations of hospitals, medical centers, route access, and emergency services to reduce and prevent future losses. As a third objective of the study, a prediction model is developed to predict evacuations in railway incidents and to identify their causes, and contributing factors using text mining and co-occurrence analysis. It is determined that Random Forest (RF) is the most accurate model for predicting evacuations. Furthermore, the type of incident (i.e., leak and spill), the action on means of contaminant (MOC)(i.e., overturning and derailment), the railyard operation and loading operations (i.e., loading, unloading, transloading, and handling), and the type of hazardous material (i.e., petroleum crude oil, diesel fuel, sulfuric acid, nitrate ammonium) are considered as contributing factors to evacuation. Finally, the study aims to develop a machine learning model capable of predicting the probability of hazmat release, identifying the underlying causes and contributing factors, and evaluating these factors in an effort to reduce hazmat release. There are many factors that can contribute to hazmat release incidents, including the location of tank cars within a train, the derailment of tank cars, the speed of the train, and the test year of the last tank. Analyzing the reports of the railway incidents using text mining indicate that the primary contributors to hazmat releases are the type of incident (i.e., release, leaking), the action on MOC (i.e., derailment, strike, puncture), and the type of hazmat involved (i.e., methanol, propane, aviation fuel).

• Subjects / Keywords

  • Risk assessment
  • railway transportation
  • machine learning
  • human factors
  • hazardous materials release
  • railway accident

• Graduation date

  Spring 2023

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-d5hg-3937

• 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.