Event Date
Engineering and the sciences depend on various types of models to understand processes and to perform design. Traditionally, these models have relied heavily on accurate descriptions of the physical phenomena that govern the behaviors of interest. In recent times, however, attention has shifted from such physically-based models to ones that depend primarily on collected data pertaining to the behaviors of interest. Provided sufficient data is available, data-driven models have proven to be effective, addressing some problems that are intractable using physically-based models. This presentation describes several data-driven analysis models applied to concrete materials and structures. The examples involve structural topology optimization; determination of concrete meso-scale properties; nondestructive assessment of reinforcing steel corrosion in structural concrete; and identification and characterization of defects in airfield runway pavements.