It is necessary to grasp joint distribution in rock masses and modelling it for reliable and effective analyses of groundwater flows. Joint information which could be obtained from the site is limited because of different conditions of site investigations. Therefore, statistical plans to estimate the joint distribution parameter must be discussed for a parametric estimation. We suggest serial job flows for joint sampling, estimation, and modelling methods (2-D, 3-D) of joint distribution through limited information to be obtained at sampling points.
Recently, with the increase of constructions of large underground structures, studies concerned with rock joint distribution, which will affect foundations at great depth, have been made actively, though, there are only few studies on development of modelling systems of jointed rock masses as a serial system to be integrated whole procedures of surveys, statistics, and modelling as a stage now. Therefore, the purpose of this study is to make a model of jointed rock masses in order to integrate the results of studies1)2)3) since before on characteristic elements of joint distribution. Further, we tried to formulate to systematize the results by introducing new methods, and developed a calculation system for modelling.
In this system, in order to consider the local deviations of regularity of joints, we classified the objected area into regions according to the distributional characteristics being statistically judged as a similar one, and we clarified the joint population in the objected area into several partial populations (joint sets) as clusters through judges of special similarity of joint distribution according to the orientation of joints. Then, based on a probabilistic space model defining each characteristic element such as 1) orientation 2) persistence 3) density 4) aperture 5) filling 6) roughness 7) conductivity by each joint set, we recognized probabilistic structure through stochastical point estimation or interval estimation of parameter to be necessary for expressions which using functions. Moreover, we made a model by means of simulating the joint system based on the probabilistic space. (Fig. 1)
For the purpose of making a stochastical simulation model of joint distribution, probablistic structure of each characteristic elements of joint distribution shall be defined.
Generally, when considering that the number of distributed joints is large and survey able area is limited, it is natural that joint distribution survey has been carried out by sampling surveys. However, the results of sampling surveys shall be apply able to estimate the probabilistic structure which shall be defined as abovementioned clause 2. As for this kind of sampling survey method, it is a general method since before to set scan lines or scan windows on the surface of rock masses and to describe about the crossed joints which is longer than the certain defined length (Q.,). Further, in order to set sampling points, it will be preferable to select the points according to the homogeneous sampling method of which work efficiency is high.
