Effect of Geotechnical Parameter Variability on Soil-Pipeline Interaction

Abstract

Soil properties are stochastic in nature, and at a particular location, the soil properties will have an associated mean and probability distribution. However, in soil-pipeline interaction studies on oil and gas pipelines where the maximum response is associated with stronger pipe-soil spring properties, the soil properties used to determine the maximum soil resistance and soil spring stiffness are usually taken as deterministic, assuming no variation in soil properties. Alternatively, for soil-pipeline interaction scenarios where the maximum response is associated with weaker pipe-soil spring properties, the geotechnical engineer may assign deterministic lower-bound values in an attempt to apply conservatism. This paper examines the effect of variability of soil properties on the maximum soil resistances and soil stiffness values for both drained and undrained soil loading conditions. Then, the resulting probabilistic distributions are applied to two example soil-pipeline interaction problems common in pipeline design, a longitudinal landslide and a strike-slip earthquake fault displacement. The results show that the stresses and strains in the pipeline resulting from drained soil loading are relatively insensitive to the probabilistic distribution of soil properties, and for the most part, the use of average values may be adequate. However, for some soil-pipeline interaction scenarios and loading conditions, and given the large natural variation in undrained soil properties, the use of single deterministic average soil properties under undrained loading conditions may underestimate important pipeline strains or stresses. The use of probabilistic soil properties for some pipeline loading scenarios is recommended.