Insights
Perspectives on supply chain analytics, risk, and data-driven decision-making.
Research Projects
Predicting Hazardous-Material Releases on Canada’s Rail Network
When a train carrying hazardous materials is involved in an accident, what determines whether those materials will actually be released—and can that risk be anticipated? In this research, my co-authors and I combined more than two decades of Canadian rail-incident data with information on weather, geography, train operations, track characteristics, and hazardous-material types, and used machine-learning models to identify the conditions associated with higher release risk. The analysis shows that factors such as track type, hazardous-material class, the number of dangerous-goods cars involved, operational activity, and environmental conditions can provide meaningful signals about the likelihood of a release, demonstrating how analytics can help move transportation safety from reactive assessment toward more proactive risk management.
To translate the research into practice, we also developed the Rail HAZMAT Release Predictor, a web-based decision-support tool that applies the models to different incident scenarios. The findings can inform decisions ranging from predictive maintenance and shipment configuration to emergency preparedness and region-specific mitigation strategies—illustrating how data and AI can turn historical safety information into actionable intelligence for regulators, transportation operators, and emergency planners. Read the full study.
Rethinking Risk Assessment: What Changes When Resilience Matters?
Traditional risk matrices rank threats according to two familiar questions: How likely is an event, and how serious would its consequences be? But two organizations facing the same disruption may experience very different outcomes depending on how prepared they are to absorb, respond to, and recover from it. In this research, my co-authors and I developed a practical approach that explicitly incorporates resilience into the traditional risk matrix. Using survey data from 60 small and medium-sized businesses in Southern Ontario, we assessed 23 operational risks—including supply disruptions, cyber risks, extreme weather, technology failures, workforce disruptions, and infectious disease—and compared their rankings with and without resilience. The results were striking: every risk changed position once resilience was considered, and even the highest-priority risks changed. For example, extreme weather and natural disasters moved into the top five when resilience was incorporated, while human error dropped out. The broader message is simple but important: understanding exposure to a disruption is not enough; effective risk decisions should also consider an organization’s capacity to withstand and recover from it. This perspective can help organizations prioritize mitigation investments and scarce resources more realistically while strengthening preparedness and business resilience. Read the full study.
Integrating Strategic and Operational Decisions for Emergency Response
Effective emergency response requires more than deciding where resources should be located before a crisis. It also requires determining how those resources should be deployed once an uncertain event occurs. In this research, my co-authors and I developed a two-stage stochastic optimization framework that connects these two levels of decision-making. Before an emergency, the model determines where response facilities should be located and what types and capacities of equipment should be pre-positioned; once an incident occurs, it determines how available resources should be dispatched based on the location and severity of the event. Rather than planning around a single assumed emergency, the approach incorporates multiple possible incident scenarios and their probabilities, allowing decision-makers to explicitly balance preparedness, response capability, coverage, and cost under uncertainty.
We demonstrated the framework through an application to hazardous-material rail incidents on a realistic transportation network in Ontario, Canada. The analysis illustrates an important principle with much broader relevance to emergency planning: strategic preparedness and operational response should be designed together rather than in isolation. It also shows how optimization can help decision-makers evaluate practical trade-offs—for example, whether to establish additional response facilities, redistribute existing equipment, invest in higher-capacity resources, or accept different response-time and coverage levels. This type of integrated approach can provide a useful decision-support framework wherever emergency preparedness involves uncertain events, geographically distributed resources, and the need to coordinate long-term investments with rapid operational response. Read the full study.
From Data to Action: Why Information Sharing Matters in Complex Systems
During a disruption, the problem is often not a lack of data—it is getting the right information to the right organizations at the right time, in a form they can understand and act upon. My master’s research examined this challenge in the context of emergency management and critical infrastructure, where transportation, energy, communications, public authorities, first responders, and private operators may be highly interdependent, yet operate with different systems, terminology, priorities, and information needs. I developed a conceptual, ontology-based information-sharing framework that addresses three fundamental questions: when should information be exchanged, what information should be shared, and how should that information be managed throughout its life cycle? The framework considers factors such as the evolving situation, system status, interdependencies, available resources, operational functionality, geographic location, and the roles of different stakeholders, with the broader goal of improving situational awareness, coordination, and resilience. Its applicability was explored through a real emergency scenario involving a major 2009 snowfall event in Lombardy, Italy.
The same principle extends well beyond emergency management. Modern supply chains are networks of interconnected organizations, technologies, infrastructure, and decision-makers, and their performance increasingly depends on how effectively data can be transformed into shared, actionable intelligence. As AI, advanced analytics, IoT, digital platforms, and other emerging technologies accelerate the digital transformation of supply chains, the ability to integrate information across organizational boundaries becomes even more important. Technology can generate and analyze unprecedented amounts of data, but its real value emerges when organizations can share, interpret, and act on that intelligence collaboratively—particularly when disruptions require fast, coordinated decisions across the supply chain. Read the full study.