Decision support architectures for the recovery of interdependent infrastructure with large‐scale damage and socially vulnerable populations. Issue 5 (14th March 2022)

• Record Type: Journal Article
• Authors: Rand, Krista; Sun, Minghui; Fleming, Cody H.
• Is Part Of: Risk analysis. Volume 42:Issue 5(2022)
• Page Start: 1124
• Page End: 1135
• Publication Date: 2022-03-14
• Subjects: recovery -- resilience -- social vulnerability -- system theoretic process analysis; Technology -- Risk assessment -- Periodicals; 658.403
• DOI: 10.1111/risa.13910
• Languages: English
• ISSNs: 0272-4332

Abstract

Abstract: Infrastructure damage has household‐level consequences after a major disaster. Losses are experienced due to factors such as unavailable services and impaired mobility. Socially vulnerable residents, in particular, have few resources with which to adapt. Decision support tools for making justifiable, transparent, repeatable decisions that center the needs of users during recovery are currently nonexistent. In part, this is because infrastructure recovery is a complex process, often involving the coordination of multiple entities. The recovery problem can be rendered more tractable by applying tools suitable for modeling complex systems and processes. System theoretic process analysis (STPA) can be used for goalsetting in a complex, dynamic system such as community civil infrastructure. STPA is used here to devise a decision support tool architecture suitable for coordinated multiagency recovery efforts. The example application is a long‐term recovery process with widespread infrastructure damage, population displacement, and other disruptions in system use due to a major disaster. In the example, losses and hazards are defined to reflect recovery challenges commonly faced by vulnerable populations experiencing partial or total displacement. This extension of STPA then reverses these hazards, starting with the most hazardous system states and progressing sequentially to less hazardous states until recovery is complete.