3D image analysis for evaluating internal deformation/fracture characteristics of materials

In material engineering, it is widely recognized that deformation / fracture or (D/F) characteristics are important because D/F characteristics help us to understand the behavior and need to be analyzed to safely use developed materials. In the past, many analyzing methods had been proposed; however, all these methods are limited to the surface of materials. In fact, nano-scale structural features like pores complicate their D/F; moreover, there is no relation between their D/F of the surface and the internality. Therefore, internal D/F characteristics have to be analyzed on nano-order. Nowadays, synchrotron radiation facilities can observe 3D-CT images on nano-order; therefore, Some researchers are attempting to analyze internal D/F characteristics from the movements of pores. However, it is difficult to extract many pores and find the corresponding relation of each pore by visual observation. In order to clarify 3D, local, high-accuracy D/F characteristics inside materials, we propose a method that extracts pores and obtains each corresponding relation of pores automatically. Inside 3D-CT images, over ten thousand pores and artifacts (noises) make it difficult to match pores. Furthermore, pores do not always have corresponding relations between preand postD/F because of different range of imaging CT as well as united and separated pores. Our method solves these problems by extracting the pores, matching pores using RBFT (Radial Basis Function Transformation) and obtaining the strains of the material, which shows its D/F characteristics. Two kinds of experiments were executed. The first one was the experiment with a trial material. This experiment enabled carrying out in the same environment as real materials. The trial material was made of a rolled sheet of a dispersion-strengthened copper alloy with alumina particles. The Constructional image of the trial material shows in Fig. 1. There were about 200 pores inside the trial material. In-situ tensile test was performed. Then we took 4 3D-CT images (Fig. 2). This experiment had two types; extracting only pores and matching pores correctly. In the result, pores were extracted from the first (1st), the second (2nd) and the third (3rd) 3D-CT image without fail; furthermore, all matching rates with three combinations of 1st-2nd, 1st-3rd and 2nd-3rd 3D-CT images were 100%. The second one was the experiment using virtual displacements. In the trial material, there were about 200 pores; therefore, we could obtain matching rates by comparisons between our results and the answers of visual observation. In real materials, the total number of pores is too large to get the answers of pore matching by visual observation. To deal this problem, we simulated the post-D/F models based on Poisson’s ratio which is one of the mechanics of materials. The pre-deformation material (Fig. 3) had 1090 pores inside. The result of pore matching was correct with respect to all 1090 pores. Our system had proven its effectiveness by experiments with the trail material and virtual displacement. However, these experiments were simple. In the future, we will conduct experiments with virtual displacement which has some noises aim at adapting real materials. 250 300