Modelling of Radiographic Testing to Derive a Detection Positioning and Sizing Algorithm for Pore Defects

Licentiate thesis, 2012

When laser welding light weight materials, e.g. aluminium and titanium based alloys, small sub-millimeter pore defects occasionally forms. Due to their small sizes, less than some 0.2 mm, as individuals they have very low impact on the structural integrity. The structural integrity might however be more affected if they can no longer be considered as individuals and thus their interspacings are important. If one could detect and measure the full 3-D position of each small pore it would be possible to make better lifetime predictions. This is in contrast to only measure their positions in 2-D, the output in conventional radiography, together with being conservative in interspacing calculations. The information could also contribute to deeper understanding of the welding process variations. The main objective of this work has been to develop a methodology to detect, size and position these small pores. To develop the automatic computer algorithms synthetic radiographs have been used together with a design of experiment approach to set it up. Probability of detection (POD) curves are frequently used within e.g. the aero industry to quantify the capability of the nondestructive evaluation. Thus, to evaluate and benchmark the algorithm POD like curves has been produced using synthetic radiographs. There are three papers included in this thesis. The first paper describes the mathematical model of the radiographic testing together with a parameterized weld geometry model. The second paper presents an improved X-ray detector model together with addition of spatial correlated quantum noise. The third paper describes the automatic algorithm to detect, position and size the pores together with a method to set its control parameters. The final results include a proposed computer algorithm together with an initial proposal for an inspection procedure based on digital radiographic testing.