1356396 - Identification of Key Model Parameters in an Infrastructure Decision Support Tool Using Sobol Sensitivity Method.

Decision support models are widely used by asset managers to make effective and efficient decisions related to the maintenance, repair, and renewal of aging transportation infrastructure. These decision support models often involve the selection of many parameters to realistically represent the system while constraining the computational cost which increases as the number of parameters increases. Also, these decision support tools for infrastructure asset management often use models with input parameters whose properties are either unknown or whose values are unpredictable thus making it difficult to determine with certainty the right parameters for the models. Traditional or local sensitivity analysis offers some insights into selecting appropriate values but does not account for interactions among the parameters. The Sobol method, a global sensitivity analysis, was implemented on the output from a decision support model for planning multi-asset highway infrastructure maintenance and rehabilitation capturing interactions to investigate the impact of the input parameters on the variance of the output, and the method identified the most influential parameters. The decision model can be refined neglecting the less-important parameters thus decreasing the dimension of the input parameters of the model and its complexity. Results show that this technique has potential to simplify the model complexity and that the refined models can perform comparably as the original model.