CT scanning has become an important technique in biomedical applications. Imaging of internal features of tissue or bone with higher contrast and resolution than ever before has been achieved using CT. Recently, use of CT to observe crack propagation in metal has begun in some researchers. Progression of fatigue crack in metals is a well-known phenomenon. It is generally believed that fatigue crack in metals will develop slowly and naturally (consistent with the classical P-K model) 12, 13 , that the crack will propagate at a certain growth rate and will eventually stop in a finite life 14 . However, many previous fatigue tests lack sufficient detail information 15, 16 as shown by early fatigue crack images acquired using OCT.
It is very important to be able to determine the fatigue life of a component because defects that cause early fatigue crack initiation are the primary cause of structural failure in the range of the fatigue lifetime of the component. However, the fatigue life of the component cannot be estimated by a single constant value, but will be related to many intrinsic and extrinsic factors. A number of fatigue life prediction models have been developed in which the crack growth rate is critical. Factors that affect the fatigue life of a component include: the applied load, the geometry of the component, the material composition and thickness of the component, the mean stress and the applied stress range and frequency of the fatigue loading cycle, the extrinsic environment (ambient temperature and moisture content) and the intrinsic properties of the component (e.g. length, diameter, volume and defect contents). These parameters are often the major factors that can be modified (or controlled) during fatigue testing (for example, such as the load and the mean stress) and through non-destructive inspection to reduce the risk of failure and extend the fatigue life of the component. Using CT imaging, the internal crack growth can be monitored at various stages of the fatigue test, and non-invasive techniques can also be used to control and optimize the fatigue process and the accelerated lifetime of the component being tested.
Some users may wish to use the BGS software in server mode, especially when constructing many non-destructive inspections to be reported as part of an overall bridge design. The user can configure some options such as cross sectional view in Figure 3 and output in comma separated format in Figure 4. Figure 5 shows the styles available for the report layout. Here, most of the discussion of the structural drawings will be centered on crack and void results as the cross sectional view is a major strength. Figure 6 is the report layout for the structural drawings. Each structural drawing has a tab associated with it to report the results. Several tabs can be added to the report, one each for major types of results and associated attributes, and many more can be defined by the user.
The results for voids and cracks are also displayed differently depending on the setting of the options, each with a different look and format to provide intuitive mapping of the results to each other and the overall scan. When using dedicated servers of the software, the results are displayed and indexed separately for each of the major types of results. Each result can be individually sorted.
The crack scans in this report were captured using a laboratory CT scanner that has been adapted for the purpose of the study. The laboratory scanner captures the x-ray dataset, digitally stores it to disk and then processes it in order to obtain the result. The x-ray data, including the phase contrast image, is first digitally subtracted from the absorption contrast image by using the inversion algorithm given by Muller [ 8 ] to enable one to obtain the phase contrast image. The data is transferred to the software, where the user must choose if the phase contrast image should be displayed on the fly during the scan, or if it should be displayed once the scan is complete. Depending on the software chosen, the display is either continuously updated (Figure 8), or on request by the user (Figure 9). The phase contrast image is typically acquired using isotropic voxels with a resolution near or equal to the resolution of the absorption contrast image (resolution of 0.15 mm in the two dimensional plane, i.e. a voxel length of 0.3 mm).