Spatial probabilistic evaluation of sea bottom soil properties and its effect on foundation design

Oğuz, Emir Ahmet
The spatial correlation length (SCL), or the scale of fluctuation, is a parameter for describing the spatial variability of a soil property and is one of the important parameters used in random field theory. Studies reporting the SCL of soil properties of offshore/nearshore soils are rather limited in the literature. In this study, the vertical SCL is determined using site investigation data from two nearshore and one large offshore sites in Turkish waters. In nearshore sites, a total of 41 boreholes and Standard Penetration Tests (SPT) reaching to 35 m depth from seabed, in water depths of up to 26 m; and in the large offshore site, 65 cone penetration test (CPT) soundings (having 10-200 m lengths in seabed) in water depths of up to 64 m, are utilized. Based on extensive data, the vertical SCL is calculated using four different autocovariance functions. Among these four functions, the squared exponential function gives the highest and cosine exponential function gives the smallest SCL values. The vertical SCL values based on SPT-N value, CPT tip resistance, friction ratio and sleeve friction are observed to be within typical ranges reported in the literature for similar soil groups, both onshore and offshore. The vertical SCL based on SPT-N value is larger than the SCL based on CPT tip resistance, friction ratio and sleeve friction. The vertical SCL based on SPT-N is slightly larger in sandy mixture soils as compared to clayey soils (about 2 m in sand mixtures and about 1.7 m in clays). In CPT data, deep water (greater than 10 m water depths) clays and sands have greater SCL than shallow water clays and sands. In both SPT and CPT cases, “constant mean with depth” approach always gives larger SCL as compared to “depth-dependent mean (or trend)” approach. The effects of SCL and coefficient of variation of soil parameters on settlement and bearing capacity of a shallow strip foundation are demonstrated by using random finite element analysis tools; RSETL2D and RBEAR2D, where all soil parameters are assumed to be lognormally distributed. The results of this study add to the limited database of spatial correlation lengths based on real data and could be useful for future studies on reliability assessment of nearshore and offshore foundations. Moreover, the results of random finite element analyses indicate that the variability of soil parameters, in terms of coefficient of variation and spatial correlation length, has a significant effect on settlement and bearing capacity of shallow strip footings.