Bridge Load Posting Prediction

University Transportation Centers Program2021PDFTech ReportKameshwar, SabarethinamMia, Md. ManikBandaru, SaiLouisiana State University (Baton Rouge, La.)United States. Department of Transportation. University Transportation Centers (UTC) ProgramTransportation Consortium of South-Central States (Tran-SET) University Transportation Center for Region 6United States. Department of Transportation. Office of the Assistant Secretary for Research and TechnologyTransportation Consortium of South-Central StatesLouisianaUnited StatesBridgesCluster analysisDeteriorationLoad limitsLogistic regression analysisMachine learningMarkov chainsMonte Carlo methodPredictive modelsBridge load postingRandom forestFuture prediction20STLSU0169A3551747106Final Research Report Aug. 2020 \u2013 Aug. 2021There are approximately 13,000 bridges in Louisiana facilitating movement of people, goods, and services. At present, about 12% of the bridges are load posted, i.e., they are deemed to lack the strength to safely carry all legal loads. With time bridges will age and deteriorate; at the same time, legal loads might increase. Load posted bridges disrupt the movement of goods and commerce. Therefore, objective of this research was to estimate the number of load posted bridges in Louisiana over the next 50 years. For this purpose, herein, a data based approach was used. For a given bridge type, the approach first developed three random forest models, to predict the future deck, sub-structure, and super-structure condition ratings of all the bridges of that type over the next 50 years. The inputs to these random forest models included a large number of bridge parameters obtained from the National Bridge Inventory (NBI). Herein, the bridge types were obtained from the Louisiana Department of Transportation and Development (LADOTD). Next, for a given bridge type, another random forest model was developed which used bridge parameters from NBI along with condition ratings to predict the load posting decision for a specific year in the future. The number of load posted bridge in each type were aggregated to obtain an estimate of the number of load posted bridge over the next 50 years. The random forest models identified several bridge parameters that significantly influence bridge load posting, e.g., condition ratings, age, span length, and roadway width among others. The results showed that while bridge types like light weight concrete pre-cast slab units and timber bridge had a large number of load posted bridges at present, concrete slab type bridge may be of concern in the future.133