65. PROCESSING AND VISUALIZATION OF RECONSTRUCTED TOMOGRAPHIC NUCLEAR IMAGING MODELS FOR STRUCTURAL NON-DESTRUCTIVE INSPECTION
Department: Computer Science & Engineering
Research Institute Affiliation: California Institute for Telecommunications and Information Technology (Calit2)
Faculty Advisor(s):
Falko Kuester
Primary Student
Name: Thomas Wypych
Email: twypych@ucsd.edu
Phone: 555-555-5555
Grad Year: 2010
Abstract
Hazardous waste materials committed for disposal are often placed in concrete casks for indefinite containment; to ensure a safe long-term storage, these concrete vessels are evaluated for structural integrity. This assessment is conventionally performed using permeability tests of the stressed containment panel specimens, where measurement of laminar flow through the cracked panel serves as the primary indicator of structural integrity. Although these tests are adequate to determine the level of containment breach, any assessment of a panel’s internal structure – and nature of the failure – is limited to superficial observations of cracks on the panel surface: the structural integrity below the surface is occluded from view. In order to provide insight to the structure beyond the panel surface, we propose a non-destructive inspection mechanism to allow detailed modeling of fatigued panels, and will complement the existing flow-based integrity tests and surface observations with subsurface structural models. Nuclear imaging technology has been proven as an effective method of non-invasive inspection, both in conventional X-Ray and more advanced computed tomography (CT) modalities. The examination system employs a laminography-type scanner, using a gamma-driven nuclear imager and experimental computed tomography techniques to perform detailed inspections of these concrete containment test specimens. Development of the scanning tool involves a number of formidable engineering tasks addressing the system design, data processing and handling, as well as presentation and visual analytics. The ground-up simultaneous development of the scanning hardware and tomography reconstruction system, based on interchangeable hardware and hardware simulation, has uniquely allowed independent development of these interdependent systems. Additionally, the current design for an in-progress modular reconstruction and integrated visualization system provides distributed data processing and handling, effectively abstracting the inspection tool seamlessly from scan-to-screen. Our work is aimed at providing a new and useful visual inspection tool to the structural engineering community, as well as to explore the considerable computer science challenges associated with both delivering high quality visual results and the increased complexity from modifying the conventional reconstruction methods. Specifically, these include: computationally costly adaptations in the reconstruction process to compensate for the inclement scanning environment, support for limited angle reconstructions, and other advanced techniques favoring quality over speed.