D. Jack
Baylor University,
United States
Keywords: nondestructive inspection, finite element methods, digital twin
Summary:
The use of carbon fiber reinforced materials has become mainstream in the aerospace, automotive, and sporting industries due to their versatility, performance, and high strength to weight ratio. This presentation seeks to address two challenges present today. First, parts are often designed under a defect-free assumption, and then a factor of safety approach is imposed to allow for unknown manufacturing variabilities. Essentially creating an inherent overdesign and additional qualification costs. The second challenge is the inability to properly validate new and advanced computational models that couple the final part performance with the underlying structure and properties due to uncertainties in the underlying structure itself. The present work presents a new approach where variations in the internal structure, sometimes termed defects, are captured using high-resolution non-destructive techniques and then directly incorporated into a finite element simulation allowing for what is in essence the true digital twin of a structure. The talk highlights some of the current work by the research team at Baylor University to quantify using non-destructive testing of a variety of common manufacturing induced and service induced defects. This work presents several novel methods developed by the present research team to nondestructively characterize the internal features within a laminated composite, and then feed the features into a finite element model domain to estimate the true, not necessarily the as designed, part performance. Inspections are performed using various aspects of high-frequency ultrasound, eddy current, and thermography, to create a three-dimensional image of internal features, and results are cross correlated with each other and with micro-X-ray computed tomography and all results are in excellent agreement. The characterized three-dimensional features are then incorporated into a finite element model domain, and the finite element results for the strain field are then compared to results of the strain field during loading from digital image correlation (DIC). The novelty of the presented method is the combination of physical testing, non-destructive testing for the geometric extraction, to structural predictions using the inspection data directly, thus enabling a true digital twin and connecting the real-world structure with the material properties to ascertain the final part performance.