Variability-Based Resilient Design Method for Structures under Undefined Uncertainties

The need for resilient infrastructure continues to increase due to evolving environmental demands and the recognized economic benefits of increased lifecycles. Greater lifecycle expectations require consideration of a structure’s resilience in the face of uncertainties that are not always known, or readily quantifiable. This paper proposes a method for assessing the reliability of structural systems where sources of variability cannot be readily defined, and traditional methods of uncertainty quantification are inadequate. The proposed variability-based framework utilizes machine learning, L2-norm, and proximity indicators to assess the sensitivity of the structure’s expected performance due to minor perturbations in its conditions. The utility of the proposed method is demonstrated through its application to a simple truss bridge, where the loci of conditions that produce a robust performance prediction, as well as disproportionate changes in structural response, are sought in the face of variability.