Two bio-siliceous Onnagawa (ONG I and ONG II) shale samples have been hydraulically fractured under two constant differential stresses (60 and 85 MPa, respectively) to investigate the fracture network's connectivity evolution by a postmortem analysis. The pressure inside the drilled borehole in a cylindrical core sample is increased above the confining pressure (10 MPa) until failure by hydraulic fracture. The two samples failed at two different borehole pressures (ONG I: 42 MPa, ONG II: 16 MPa). Fractured samples were scanned in an industrial X-ray CT machine and the tomographic images of the fracture network were extracted for a postmortem investigation. From the fracture volume segments, obtained by thresholding the frequency distribution of the fracture network's voxel values, a quantitative estimation of fracture connectivity was carried out. The connectivity was quantified based on the relative entropy of size distribution of fractures (${H_r}$), a method adapted from information theory. Fracture connectivity estimation shows that ${H_r}$ is at a maximum value when the fractures show a significant distribution with very limited connectivity. The value of ${H_r}$ is at a minimum and close to 0 when a well-linked fracture network is formed. In both samples, this minimum was attained at the threshold of 43k indicating the highest connectivity and the best representation of the fracture network. The extracted fracture network of ONG I showed a multi-winged hydraulic fracture network while a planar conventional two-winged hydraulic fracture network had been generated in ONG II with a lower fracture volume.

X-ray computed tomography (XCT) has become a popular method in fracture characterization of rocks. After Wellington & Vinegar (1987) first utilized XCT for petrophysical studies, it became a popular method in analyzing and visualizing the internal structure of rocks in the early 1990's (e.g., Johns et al., 1993;Orsi et al., 1994;Vinegar et al., 1991). Later with the use of it to characterize the porosity of fault zones (Antonellini et al., 1994) and strain localization in shear bands (Desrues et al., 1996) in deformed rocks, its implications in geophysical applications were realized. However, these were not powerful enough in both penetration and resolution to obtain high resolution images of high-density formations. This led the subsequent studies of Otani et al. (2000) to perform their experiments in a more developed and powerful industrial type XCT scanner, which was able to reconstruct the internal structure of soil in a nondestructive manner. They further showed that the structure of soil samples can be investigated quantitatively based