2019 CSCE Annual Conference - Laval (Greater Montreal)

2019 CSCE Annual Conference - Laval (Greater Montreal) Conference


Title
CLUSTER BASED REGRESSION MODELING FOR PREDICTING CONDITION RATING OF HIGHWAY TUNNELS


Author(s)
Dr. Sahar Hasan, Purdue University
Dr. Emad Elwakil, Purdue University (Presenter)
Abstract

Highway tunnels are long-term projects that committed with higher capital costs and additional levels of maintenance because of its complex systems. However, fewer deterioration models have been built compared with other highway components such as bridges. The limited research in this area has raised the need to develop models for predicting the tunnel condition based on tunnel inventory database (NTI); that provides objective decisions for future maintenance plans.  This paper has investigated the significant impact of dependent variables on tunnel conditions from three aspects: Geometric, Inspection, Structure Type and Material. The proposed methodology has based on two phases analysis; two- step cluster analysis and regression modeling.  Nine models have been developed with high coefficient of determination (R2=90.8%), classified in terms of service in tunnel and ground condition. Further analysis using average validity percentage (AVP) method was used to examine the validity of built models and come out with satisfied results (83%). The developed model benefits highway authorities to prioritize the maintenance and make informed investment decisions in an objective manner based on the historical data.