Abstract:
This paper presents a process for generating an "as-built" finite element model of an object using nondestructive evaluation (NDE) data, contrasting it with the traditional "as-designed" approach that relies on idealized specifications. The goal of as-built modeling is to create a more realistic and accurate numerical representation of an object by incorporating actual geometry and material properties obtained through methods like X-ray computed tomography (X-ray CT) and ultrasonics (UT). The process involves several steps: acquiring data from multiple NDE modalities, fusing this data to form a comprehensive dataset, and then using this dataset to create a solid model for finite element meshing. This approach is particularly valuable for performance assessment and reducing the uncertainty in numerical simulations. The authors illustrate this method using a cylindrical test object containing different materials (aluminum, epoxy, and cellulose) and demonstrate how data fusion between X-ray CT (which is good at distinguishing materials with different densities) and UT (which is effective at identifying interfaces not visible to X-ray CT) can create a more complete model. The paper highlights that while X-ray CT could distinguish aluminum from epoxy, it failed to differentiate between epoxy and cellulose due to similar X-ray attenuation, a gap that was filled by the ultrasonic data which successfully identified the epoxy-cellulose interface. Ultimately, this as-built model provides a more accurate representation for subsequent finite element analysis, offering better insight into the object's real-world performance.
