Fatigue analysis of offshore well conductors: Part I – Study overview and evaluation of Series 1 centrifuge tests in normally consolidated to lightly over-consolidated kaolin clay

Abstract An important aspect of deepwater well integrity assurance is conductor fatigue analysis under environmental loads acting on the riser system during drilling operation. Fatigue damage arises from stress changes in a structure due to cyclic loading. In practice, the lateral cyclic soil response is typically modelled using Winkler p – y springs. An appropriate soil model for conductor–soil interaction analysis is the one based on which the absolute magnitudes of stresses and their changes can accurately be predicted for well integrity evaluation. The API recommendations for p – y curves, which are often used for conductor–soil interaction analysis, have originally been developed for piled foundation and are inappropriate for well fatigue analysis. To that end, an extensive study involving four series of centrifuge model tests and FE numerical analyses was conducted to fundamentally study conductor–soil interaction under a wide spectrum of loading conditions. The tests simulated conductor installations in normally to over consolidated clays, and medium-dense clean sands. Soil models were developed specifically for conductor fatigue analysis for each of the soil types. The test results and soil models are presented in two papers. The first paper, Part I, presents an overview of the study and first series tests in normally to lightly over-consolidated kaolin clay and discusses the observations made with regards to monotonic and cyclic soil resistances and their relationship to conductor fatigue modelling. The second paper, Part II, presents centrifuge test results in normally to lightly over-consolidated Golf of Mexico (GoM) clay, over-consolidated natural clay and medium-dense clean sands along with the respective soil models developed for conductor fatigue damage prediction. Overall, the accuracy of fatigue life predictions using these novel soil models is very high – generally within about a few percentage of the measured values.