The surrogate model is applied to a reliability analysis of the stability of untrenched pipelines. Experience has shown that the stability of a pipeline is sensitive to a number of variables associated with soil properties, structural modelling, and hydrodynamic loads. Because uncertainties are unavoidable in these key design parameters, it is vital to evaluate pipeline instability failure using a probabilistic approach. However, a reliability analysis of on-bottom stability in the time domain is often a very computationally expensive process. The stability of a pipeline exposed to time-varying environmental loads is a highly nonlinear phenomenon with pipe–soil–fluid interactions. In this work, surrogate model technology was utilised to realise the reliability analysis which is also quite time consuming. A Fourier model was used to calculate the hydrodynamic forces acting on a pipeline. The surrogate model was constructed using experiment sampling and, in conjunction with a Monte Carlo simulation, was employed to perform the reliability evaluation. Moreover, the influences of various random variables on pipeline instability are discussed through a sensitivity analysis. An accurate assessment of the conservatism in the different variables could lead to fewer conservative stabilisation requirements. The results show that using a surrogate model for reliability analysis not only reduces the computational cost significantly but also produces a highly accurate evaluation of pipeline stability.
Dynamic stability analysis of pipeline based on reliability using surrogate model
Journal of Marine Engineering & Technology ; 12 , 2 ; 75-84
2013-04-01
10 pages
Article (Journal)
Electronic Resource
English
Surrogate model uncertainty quantification for active learning reliability analysis
Elsevier | 2024
|A New Robust Surrogate Model: Reliability Based Hybrid Functions
British Library Conference Proceedings | 2011
|