Characterization of spatial variability of soils

The spatial variability of soils contributes significant level of uncertainty in civil engineering designs. Reliability Based Designs provide a consistent framework to quantify the uncertainties and one of the primary steps in the reliability analyses of geotechnical designs is to characterize the in-situ spatial variability of soil properties. And such a reliability analysis needs rational stochastic soil models and a priori variability estimates.; Recent evidences suggest to say that the stochastic nature of soils may be fractal in nature. For long-memory processes, the scale of fluctuation and variance are found to be dependent on the sampling domain sizes. Fractal processes were generated by Fast Fourier Transform (FFT) and Monte Carlo simulations were employed to establish this domain dependency relationship of scale of fluctuation and variance. The effective scale of fluctuation and variance corresponding to respective domain sizes for different correlated random fields are established. A simulation based equivalent finite scales of fluctuation to simulate fractal processes are also established.; An extensive spatially distributed soil data sets from different parts of the world were collected to examine the fractal behaviour in soils. Among the seven soil sites analyzed, five soil sites showed the presence of fractal nature in the CPT data. Further, the domain dependency relationship of scale and variance observed in the simulations was verified with these real world data, and the distributions of scales were established at increasing sampling domain lengths. A priori estimates for scales of fluctuation as a function of domain lengths and spectral exponent of fractal model are established. These estimates are believed to be readily used for reliability analyses at any arbitrary site.