A study of profile characterization for rock joints

Summary This study is part of a longer-term effort to characterize the shear and normal behavior of rock joints. The profile characterization of rock joints is an important part of this effort since the topography of a joint has a major influence on its tribological behavior. Although many parameters have been proposed by various authors to characterize joint topography, they do not represent unambiguous intrinsic properties of a joint. Such parameters are usually sensitive to the approach used to estimate their values, and, for joint measurements, are dependent on the sampling interval as well as the sample size. Thus, consistent values for these parameters can be difficult to determine. This paper presents an investigation of the sensitivity of some joint profile parameters using numerically-generated profiles which allow one to investigate the characterization technique and some aspects of the sampling methodology.Specifically, Gaussian-distributed profiles characterized by an exponential cosine autocorrelation function and an exponential autocorrelation function are generated. Gaussian profiles can be completely characterized by the first three even spectral moments, and, in this study, these spectral moments are estimated by three approaches — the finite difference approximation approach, the crossing-counting approach, and the autocorrelation function (acf) approach. The estimated values of these spectral moments are then compared with their analytical values to investigate the discrepancies caused by different estimation approaches. The variations of the spectral moments of the exponential cosine representation against the profile structure, the sampling interval and the sample size are also described. In addition, the influence of the sampling interval on the inferred roughness angle of profiles characterized by an exponential cosire function is briefly discussed.