Detection and 3-D positioning of small defects using digital radiography, 3-D point reconstruction, and tracking

The weight and quality concerned aero industry is constantly trying to decrease the excess material in their components. With decreased weight the fuel consumption can be lowered and the negative environmental effects from aviation decreased. One possible way to remove excess material, for example in gas turbine engine components, without decreasing safety is to increase the capability and reliability of the quality control. Such quality control improvements would facilitate less uncertain fatigue life predictions, which would then support cutting some of the excess material in the component without lowering the safety factors. When laser welding light weight materials, for example titanium based alloys, small submillimeter pore defects occasionally forms in clusters. Due to their small sizes, less than some 0.2 mm, as individuals they are believed to have low impact on the component fatigue life. However, existing fatigue life models predict that pore defects with small interspacings should instead be considered as a single larger pore, which might then have a higher impact on the fatigue life. These submillimeter pore 3-D interspacings is an example of increased nondestructive evaluation capability of value when predicting the fatigue life of manufactured components. The aim of this work has been to assess the applicability of using digital radiography and 3-D point reconstruction algorithms as the at the manufacturing in-line quantitative nondestructive evaluation to detect, size, and 3-D position small submillimeter pores. There are six papers included in this thesis. The first paper describes a simple mathematical model used to produce realistic synthetic radiographs representing the radiographic inspection. The model is extended in the second paper with an improved X-ray detector model, spatial correlated quantum noise, and additional experimental comparison. In addition, the validity range of the presented model is explored. In the third paper, a 3-D point reconstruction algorithm based on tracking is derived, set up, and pre-evaluated utilizing synthetic radiographs, and experimentally verified on a weld sample. The fourth paper contains a high contrast-to-noise ratio experimental verification of the 3-D point reconstruction algorithm. In addition, a method and an algorithm to measure the radiographic setup geometry is derived and experimentally verified. In the fifth paper the detectability limitations in 3-D point reconstruction are explored in connection to component fatigue life models. In the sixth paper the capability of the 3-D point reconstruction algorithm is experimentally explored on a commercially available setup utilizing the radiographic magnification technique