ABSTRACT
In managing systems of distributed infrastructure, such as bridges within a transportation network, maintaining the functionality of the system as a whole, rather than any individual component, should be the primary goal. The structure of the overall network will therefore have a large impact on the relative values of monitoring and maintaining the individual components; for example, in a set of bridges connected in series, it is essential that all bridges be maintained in an operational condition, while for a parallel system of bridges, the failure of a single bridge will not necessarily disconnect the entire system. This paper provides a theoretical investigation into the relative values of monitoring bridges in transportation networks with various topologies and under various assumptions on what constitutes a failure of the bridge network. A probabilistic model of the bridges, which incorporates variables relevant to the state of the bridges, including interdependencies between these variables for similar structures, is used to represent the system of bridges, probabilistically defining their states (i.e. failed or not). The network topology, defining the connectivity between these bridges, is used to assess how the status of individual bridges affects the overall state of the network. The value-of-information metric is then used to assess the potential benefits of monitoring individual bridges in the system under different assumptions on the network topology and correlations in bridge state variables. This metric is based on the difference in expected management costs for the system with and without additional information, such as might be provided by detailed bridge inspections or structural health monitoring activities, and therefore can be used by managing agencies to decide whether and where to implement these strategies.
Figure 1 presents results for three system topologies (cumulative, series, and parallel), showing how value of information increases as measurements are made in each system. First, it should be noted how the interdependencies between components affect the relative values of observations. Second, the similarity in results for series and cumulative systems indicate that it may be possible to more efficiently compute the value of information in series systems by modeling them as equivalent cumulative systems. value of information (VoI) vs. number of sensors for cumulative, series, and parallel 5-component systems. Factor <italic>ρ</italic> indicates the correlation between component variables (0 for independent components, 1 for identical components). https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9781315207681/cd556cd4-4dcf-4efe-8e29-56fc67b8bfbd/content/fig149_1.tif"/>
