Supporting rational decision-making in civil engineering

  • Andrea Verzobio

Student thesis: Doctoral Thesis


The management of civil engineering structures, such as bridges and dams, is fundamental for ensuring their continued safe and economical operation, where decisions such as whether or not to suspend operations are based on uncertain knowledge concerning the state of the structure. Modern development in technology has made available several accurate monitoring devices providing Structural Health Information, that can be used to support such decision-making through more informed assessment of the structural state. The wide-spread adoption of these devices has led to the development of Structural Health Monitoring (SHM) decision support system, to identify appropriate courses of action based on observed data from the monitoring system. Essentially, it is a two-step process, which includes the judgment of the structural state based on the SHM information, and the decision about the optimal action based on the knowledge of the structural state. When engineering knowledge concerning the state of the system is uncertain, as the monitoring system does not directly observe the state of the structure, Bayesian inference and Expected Utility Theory provide the only consistent way to judge and to make decisions, respectively, as all alternative inferential methods for decision support are susceptible to logical inconsistency.;However, we must recognize that in the real world the process followed by decision makers may be distorted. The goal of the research proposed in this contribution is twofold: we investigate how heuristic behaviours may affect human judgment and decision-making in civil engineering, and also how decision-making can be distorted when multiple agents, even rational but with different appetites for risk, are involved in the decision chain. Firstly, most agents in everyday life apply heuristic approaches rather than a formal Bayesian procedure in order to make inference to support decisions. In particular, without the use of formal algorithms to support rational interpretation of data, humans apply simple strategies or mental processes to interpret data, which are prone to systemic errors. This may happen with data that come from various data sources, such as SHM but also engineering expert knowledge. Innovative frameworks to support rational decision-making are then required, in order to minimize the risk of biased judgments or decisions. For instance, being able to predict the behavior of an irrational manager is necessary when we set a general policy for bridge management, and we know that someone else who is going to enact the policy may behave irrationally.;In this doctoral thesis, we start reviewing the literature of heuristics and cognitive biases in order to identify the most relevant as regards human judgment and decision-making for civil engineering structures. We identify Kahneman and Tversky's representativeness as a heuristic for which SHM-based decision-making is particularly susceptible, where simplified rules for updating probabilities can distort the decision maker's perception of risk. Therefore, we reproduce mathematically this observed irrational behavior to investigate how it distorts human judgment. In addition, it is recognized that heuristic behaviors may affect expert knowledge. Consequently, we propose a method for eliciting engineering expert knowledge in order to assess civilengineering structures: the process is required in order to support the collection of valid and reliable data, by minimizing the adverse impact of cognitive biases. Secondly, the decision process can be distorted when multiple agents are involved, not only in the case of irrational behaviors, where the distortion is expected, but even in the case of rational behaviors. Indeed, decision makers may differ in their decisions under uncertainty according to thei
Date of Award13 Aug 2020
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde
SupervisorDaniele Zonta (Supervisor), Enrico Tubaldi (Supervisor) & John Quigley (Supervisor)

Cite this