An intensive sensitivity test was made on the block size distribution for various joint system models, as well as truncating and censoring processes made in the joint survey. The unknown characteristic features of rock joints are the size and spacing of joint planes due to the limited access for field sampling, but they are very sensitive to the number and volume of blocks framed. By knowing the impact of rock joint modeling based on truncated and censored data, the field sampling can be modified for a specific project to minimize the adverse impact on the final engineering analysis.
Rock joint systems in geological formations are surveyed routinely and characterized for geotechnical and geohydrological engineering design and analysis. The joint characteristic parameters include joint orientations, sizes, spacings, surface roughnesses and apertures or some other parameters such as joint connectivities and permeability or shear strength values depending on their applications in the engineering analysis. However, joint data surveyed in the field are truncated and censored due to the limited field access, and thus statistically biased. The block size (volume) distribution, blocks formed by joint planes in the rock mass, is a pertinent characteristic parameter in these engineering analyses, since it can describe or be related to the engineering criteria intuitively and/or quantitatively. In this paper, a series of sensitivity tests were made to study the impact on block size distributions with various joint models and their characteristic parameters. The block volumes were calculated by the finite element approach developed recently by Young (1992). The statistics of block size histograms were compared to quantify those impacts that can be introduced into the engineering analysis and judgements, as well as better joint sampling plans for the field survey. The unknown characteristic features of rock joints are the size (or extension) of joint planes and the spacing between joint planes (or density) due to the limited access available for field sampling and the practicality of measuring every joint plane. These parameters are directly related to the block size and the number of blocks formed, and are tested extensively in this paper for their sensitivity on the block formation. The sensitivity tests were made for different statistical models and various ranges of censoring and truncations in each model. By knowing these impacts of rock joint modeling based on truncated and censored data, the field sampling can be modified for specific projects to minimize the adverse impact on the final engineering analysis.
The purpose of joint simulation is to generate a set of joints whose statistical distribution will duplicate the field model of joints observed from joint mapping. Due to the limited access for sampling or sampling window size, the field mapping data is truncated and censored. So, the field joint model has to be inferred from the limited number of joint mapping data. Various statistical models were studied for joint analysis for these purposes by many investigators (Arnold 1941, McMahon 1971, Baecher 1977, Dershowitz 1984, Young 1990).
