Date of this Version
Steelman, J., Garcia, F. (2020). Supporting Bridge Management with Advanced Analysis and Machine Learning. NDOT Research Report SPR-1(19) M088.
A supplemental Artificial Neural Network (ANN)-based tool was developed to support the Nebraska Department of Transportation (NDOT) in optimizing bridge management investments when choosing between refined modeling, field testing, retrofitting, or bridge replacement. Load ratings typically increase by approximately 15% to 20% when using detailed finite element analysis (FEA) instead of AASHTO approximate analysis methods. The ANN tool is implemented in an Excel spreadsheet to accept ten input parameters readily available to NDOT engineers performing typical load ratings, and predicts FEA- equivalent critical girder distribution factors (GDFs), removing unnecessary conservativism from approximate AASHTO GDFs, potentially justifying load posting removal for existing bridges, and enabling more optimized design for new construction. The Excel tool outputs direct ANN-predicted GDFs and adjusted GDFs penalized to account for ANN error by reliability calibration philosophically consistent with AASHTO Load and Resistance Factor Rating. The study included detailed FEA for 174 simple span, steel girder bridges with concrete decks. Subsets of 163 and 161 bridges within these available cases comprised the ANN design and training datasets for critical moment and shear live load effects, respectively. The reliability calibration found that the ANN live load effect prediction error with mean absolute independent testing error of 3.65% could be conservatively accommodated by increasing the live load factor by less than 0.05. The study also demonstrates application of the neural network model validated with a diagnostic field test, including discussion of potential adjustments to account for noncomposite bridge capacity and Load Factor Rating instead of Load and Resistance Factor Rating.