Integrating Qualitative Information and Quantitative Information for a Decision Support System for Safety-Critical Facility Inspections

Bulk fuel storage facilities are inspected to identify mitigation strategies needed to prevent fuel leaks and possible accidents. Inspection schedules drive important decisions around safety, security and environmental protection, but are limited by personnel and resources available, particularly in remote and infrastructure-challenged settings where data and resources are scarce. To overcome these limitations, we propose a methodology to use quantitative information integrated with expert qualitative information as inputs in a mixed-integer linear program to prioritize bulk fuel facility inspections in resource-limited settings. The quantitative information includes numerical data relevant to fuel storage facilities’ environmental risk indices. The qualitative data includes expert inspection comments and observations. Statistical analysis, natural language processing methods and sentiment analysis are used to convert the qualitative data into a quantitative form that can be input into the MILP.  A case study of the optimization decision support provided by this approach for the U.S. Coast Guard in Western Alaska and the U.S. Arctic is presented, along with a framework for integrating quantitative and qualitative data to provide decision support addressing data, resource and personnel limitations in resource-challenged settings.