Can interdependency considerations enhance forecasts of bridge infrastructure condition? Evidence using a multivariate regression approach. Issue 8 (2nd August 2020)

• Record Type: Journal Article
• Authors: Lavrenz, Steven M.; Saeed, Tariq Usman; Murillo-Hoyos, Jackeline; Volovski, Matthew; Labi, Samuel
• Is Part Of: Structure and infrastructure engineering. Volume 16:Issue 8(2020)
• Page Start: 1177
• Page End: 1185
• Publication Date: 2020-08-02
• Subjects: Interdependency -- deterioration -- system-of-equations method -- three-stage least squares -- bridge management -- condition prediction; Structural analysis (Engineering) -- Periodicals; Structural engineering -- Periodicals; Buildings -- Performance -- Periodicals; 620.005
• DOI: 10.1080/15732479.2019.1693606
• Languages: English
• ISSNs: 1573-2479

Abstract

Abstract: In recent years, there has been an increasing trend towards viewing complex infrastructure systems in a holistic manner, to better understand their behaviour. This paper extends the holism concept to facilitate comprehension of the deterioration of the three bridge components: the deck, superstructure, and substructure. The hypothesis is that these bridge components deteriorate, not in isolation, but collectively as a system of systems, and therefore their deterioration exhibits substantial interdependency. To investigate these interdependencies, this paper uses National Bridge Inventory (NBI) data. The paper explores the efficacy of using a multivariate three-stage least squares (3SLS) model to describe these interdependencies and to quantify the effects of other explanatory factors on the deterioration of these components. The results of the 3SLS model are compared to traditional linear models estimated using ordinary least squares (OLS), to demonstrate the 3SLS model's ability to return more precise estimates of bridge deterioration effects. The results show that the 3SLS model statistically outperforms the OLS models by an average of 104% and 173%, based on root mean square error and mean absolute percentage error, respectively. These results support the hypothesis that bridge components exhibit system-of-systems behavior as their deterioration levels are influenced by the condition of each other. The results also suggest that such holistic nature can be captured using a simultaneous equation model. Improved condition prediction can enhance confidence in agency bridge management functions and result in a more efficient use of limited public resources for infrastructure maintenance and repair.