Human-Centered Simulation Modeling to Facilitate Critical Infrastructure Resilience Planning

Abstract

Planning for timely, effective, and efficient post-disaster infrastructure restoration is critical to prevent prolonged disruptions. Infrastructure resilience planning aims to anticipate damages and disruptions in service, envision the restoration process, and provide recommendations to shorten disruptions. Simulation modeling has computational capabilities to capture domain-oriented complexities such as interdependencies. Several resilience planning initiatives have been conducted collaboratively by emergency management and infrastructure experts in the last two decades in the US. However, no simulation models were used in the initiatives. Modeling methodologies that can adequately facilitate collaborative infrastructure resilience planning are missing in the restoration modeling literature. This dissertation aimed to facilitate collaborative infrastructure resilience planning. The planning is a complex process as it involves domain- and user-oriented dimensions. The human-centered design process was applied to design and develop a modeling methodology that can support stakeholders in resilience planning by considering both domain- and user-oriented dimensions. For this purpose, eighteen interviews were conducted with experts who participated in the previous resilience planning initiatives to understand the procedure, identify challenges, and explore opportunities of using simulation modeling. Human-centered simulation modeling was created as a conceptual design framework for developing simulation models. This framework identifies essential design components in developing the modeling methodology consisting of user-interaction, system representation, and computation core. This framework enables simulation modeling developers to ensure that the user's needs, strengths, and concerns are taken into account. There is a lack of modeling methodologies to utilize the design framework and address the identified challenges, especially for complex resource availability scenarios. We developed a process-based discrete-event simulation modeling methodology that is resource-aware and topologically-explicit. This methodology was built as a combination of discrete-event simulation modeling and network modeling. It was applied to simulate a hypothetical electric power network's restoration as a case study to demonstrate the modeling performance, and several sensitivity analyses were performed to validate it computationally. It can provide insights into the restoration process and sequence, scheduling, resource allocation, and service restoration timeframes. This work contributes to restoration modeling literature by considering time-dependent and multi-source resource availability in simulating the post-disaster infrastructure restoration process.