Disaster recovery planning of transportation networks

The way to disaster-resilient transport infrastructures is paved with effective recovery planning. Disaster Recovery Planning of Transportation Networks (DRPTN) is a critical and complex concept that often relies on recommendations of decision models and decision support systems. To develop such decision systems and draft a reliable recovery plan, a well-structured modeled problem with equitable decision attributes is essential. This dissertation aims to address problem structuring and methodological identification of DRPTN decision attributes. To do so, the research begins with a systematic analysis of the DRPTN optimization models, divided into four phases: problem definition, problem formulation, problem-solving, and model validation. For each of these phases, challenges and opportunities are articulated, as well as suggestions to overcome the identified gaps. To address the knowledge gaps in the problem structuring of DRPTN models, I developed a prescriptive decision aid mechanism to assist in harnessing experts' knowledge and recommend decision attributes for DRPTN problems. Afterward, I implemented this framework in a real-world DRPTN problem case study to test its performance, analyze the outcomes, and produce a systematically selected set of DRPTN decision attributes. The findings of the research have been reported in three sections, which are outlined in detail below. The findings of the gap analysis suggest the presence of critical challenges within decision attributes of existing DRPTN models, including insufficient efforts to justify and support the adopted decision factors with theoretical arguments or formal selection processes. Furthermore, the problem-solving phase of DRPTN modeling would benefit from adopting meta-heuristic algorithms when explicit or implicit justifications exist, such as convexity, linearity, or complexity analysis of the mathematical programming. In the problem formulation phase, more effort could integrate traffic management measures and post-event travel demand models into the formulation of network recovery. Addressing the validation phase of DRPTN models, a benchmark system, multi-aspect simulation advances, and a systematically developed level of confidence is needed to support the reliability of the outcomes. The gap analysis results suggest that the method-rich but methodology-poor phenomenon appears as a challenge for disaster recovery models. With respect to the developed attribute-selection methodology, the framework implementation enabled the use of experts' collaborative input in a structured manner and promoted a disciplined decision process. The multi-stage, non-hierarchical architecture of the framework allowed for the critical and thorough evaluation of candidate attributes in a relatively user-friendly manner. The framework could act as a mechanism to harness decision-makers' knowledge and help them to isolate those elements of the decision context which are most relevant to the problem. Therefore, the recommended set is supposed to result from a thorough, systematic process and collaborative decision-making; hence, it offers tenable attributes for both DRPTN practice and research. Finally, the process has led to six attributes for the case study’s road network disaster recovery planning: 1) access level to service-providing nodes, 2) integration of link travel delay and traffic flow, 3) travel time improvement per recovery duration, 4) travel time improvement per unit of resource, 5) centrality measures, and 6) link capacity. Using the recommended set of attributes in a DRPTN model is expected to provide effective and efficient recovery solutions that maximize mobility and accessibility in the network. Analysis of the results suggests that the framework leads to an improved attribute set compared to the attributes selected in an unassisted manner. The sensitivity analysis confirms that the outranked outcomes are relatively robust against the assigned preferences. This argument was also supported by an information entropy analysis. Both analyses suggest that "certainty" was an incentive factor for participating experts while evaluating candidate decision attributes. Throughout this research, I was able to 1) identify knowledge gaps and opportunities in optimized DRPTN decision models through conducting a systematic critical literature review and suggest solutions for detected challenges, 2) formalize the decision process of selecting attributes with a few innovative mathematical formulations and modeling approaches, 3) assist and harness the knowledge of subject-matter experts with a decision aid mechanism customized for this purpose, 4) offer the methodology as a toolkit for further application in both science and practice, and 5) suggest a set of decision attributes of DRPTN for the case study. Finally, besides these main contributions, I also had the chance to observe and report on some new technical improvements, understandings, and knowledge that can be useful for scientists and practitioners in decision analysis, traffic engineering, and disaster management. The dissertation concludes with an emphasis on the art of problem structuring in the DRPTN context.

Author dc:creator

• Zamanifar, Milad

Advisor dc:contributor.advisor

• Hartmann, Timo

Licence dc:rights.uri

• https://creativecommons.org/licenses/by/4.0/

Language dc:language.iso

en

Identifier URI

https://doi.org/10.14279/depositonce-16150

OAI identifier oai:identifier

oai:depositonce.tu-berlin.de:11303/17369