Probabilistic Integrity Assessment of Pipelines and Pressure Vessels With Localized Corrosion

The acceptability of localized corrosion for pipelines in service is discussed and the methodology for probabilistic assessment of corrosion damage using the 2004 revision of Det Norske Veritas document DNV-RP-F101 is presented with regard to predicting the remaining life of onshore, offshore pipelines and also pressure vessels for the oil and gas industry. Capacity assessment of corrosion defects in pipelines and pressure vessels is a challenge for the industry, and has resulted in several methods and codes in the recent years. The methods include ASME B31G for pipelines and other standards and recommended practices that can be applied for pipelines as well as pressure vessels such as BS7910 Annex G and API 579. In 1999 DNV issued a recommended practice for the assessment of corroded pipelines, DNV-RP-F101, which was developed in co-operation with the pipeline industry. A revision of this document was issued in October 2004. The latest revision provides guidelines for probabilistic assessment of the pipeline reliability and remaining life. Pipeline corrosion is often of localized nature and depends on many factors such as fluid composition and partial pressures, temperature, pH, flow rate and efficiency of corrosion inhibition. These factors may not be easy to quantify with great certainty and a probabilistic approach is particularly justified. An approach is presented in this paper where the actual measured corrosion damage is fitted to an extreme value distribution. The future corrosion rate distribution for internal CO₂ corrosion is predicted using the latest de Waard and Milliams model and an inhibitor distribution, to predict the corrosion of each pipeline segment. An Gumbel type extreme value approach is applied to estimate the present condition and the local corrosion flaw distribution that may cause a leak of the pipeline in the future. The future corrosion is estimated using the anticipated future operating conditions of the pipeline to predict the corrosion rate distribution expressed in terms of a Weibull distribution. The paper highlights three cases as examples where the approach has been applied for assessing the probability of failure and reliability during service of two offshore pipelines carrying oil and gas with wet CO₂, and one stainless steel pressure vessel in a process plant occasionally exposed to trace amounts seawater originating from leaking heat exchangers.