Optimal inspection and design of onshore pipelines under external corrosion process

Abstract

Pipeline systems can be safely designed and operated by using conservative safety margins and approximations. However, when expected consequences of failure are accounted for, optimal designs or optimal inspection/maintenance plans cannot be found using overly conservative assumptions. Specifically, pipeline corrosion cannot be modeled using popular but overly conservative linear corrosion growth models. In this paper, a novel polynomial chaos corrosion growth model is constructed from extensive field data, and employed in the optimal design of an example buried pipeline. The optimal corrosion thickness, time to first inspection and time between successive inspections are considered as design variables. The design objective is to minimize total expected life-cycle costs, which include costs of construction, inspections and repair, and expected costs of failure. Expected numbers of failures, repairs and replacements are evaluated by a probabilistic analysis using Latin hypercube sampling, and a novel approach is presented in order to smoothen these expected numbers w.r.t. design variables. The resulting objective function is discontinuous, and presents many local minima; hence, global optimization algorithms are required. A multi-start simplex algorithm is employed, but results are also compared with a crude exhaustive search. Results are obtained for several cost configurations, reflecting different failure consequence scenarios. A discussion is presented with respect to the optimal inspection schedules and optimum corrosion thicknesses found herein.