Using data to explore trends in bridge performance

Abstract

Asset management organisations collect large quantities of data on the inventory, condition and maintenance of their bridge structures. A key objective in the collection of these asset data is that these can be processed into useful information that can inform best practice for the design of new structures and the management of existing stocks. As a leading bridge asset owner, Highways England, UK, is applying insights from mining of its asset data to contribute to continual improvement in the management of structures and its understanding of their performance. This paper presents the application of modern data science tools and optimal decision tree learning to Highways England's asset information database comprising bridge inventory, inspection records and historic and current defects for its stock of thousands of bridges. Trends are observed in the factors affecting the current condition of bridges and their rate of deterioration. Optimal decision trees are used to identify the most influential factors in the performance of bridge structures and present complex multifactor trends in a format readily digested by managers and decision makers, to inform standards and policy.