A data heterogeneity modeling and quantification approach for field pre-assessment of chloride-induced corrosion in aging infrastructures

Aging infrastructures (e.g. roads, bridges and water mains) in America are deteriorating and becoming structurally deficient and their reliability and safety issues become matters of great concern. For the reinforced concrete infrastructures in marine environments, one of the leading failure causes is chloride-induced corrosion, which consists of a complex electrochemical process of chloride ingress. Inspecting chloride ingress conditions involves the costly and time-consuming procedures of extracting cores and performing laboratory analysis. Based on the limited resources, it will be desirable to develop pre-assessment approaches in evaluating chloride-induced corrosion conditions before extracting cores. Existing approaches mainly rely on engineering experience and/or visual inspection, which may be subjective or subject to visual inspection error. Existing approaches in analyzing trajectory profiles are often restricted by the oversimplification of homogeneity assumption and failed to address the potential heterogeneity among profiles data. This paper proposes an evidence-based analytical approach for chloride ingress pre-assessment by comprehensively exploring, quantifying and analyzing the historical heterogeneous chloride ingress profiles data and associating them with inexpensive external factors information, which are often readily available from concrete suppliers and bridge inventory databases. Given inexpensive information of a location to be inspected, the propose work can provide rich pre-assessment results, which will facilitate engineers to prioritize their resources and schedules and first inspect those most at-risk locations. A real-world case study is provided to illustrate the proposed work and demonstrate its validity and performance. (C) 2017 Elsevier Ltd. All rights reserved.