Print Email Facebook Twitter Probabilistic approach for the optimization of pipework inspection and maintenance Title Probabilistic approach for the optimization of pipework inspection and maintenance Author Knijnenburg, L.J. Contributor van der Stap, A.C.M. (mentor) Renting, F.W. (mentor) Faculty Civil Engineering and Geosciences Department Hydraulic Engineering Programme Offshore & Dredging Engineering Date 2017-06-02 Abstract Fluctuating oil and gas prices, coupled with initiatives toward energy security, drive the global storage market of oil and gas. A tank farm, often referred to as a terminal, is an industrial facility for the storage of the oil and/or petrochemical products. Vopak's biggest terminal, Vopak Terminal Europoort (VTE), has an immense capacity of 25 million barrels. The terminal contains over 200 on-site pipelines to handle the products in and out the terminal. (VTE) was mostly build in the early seventies and after years of operation and alterations, the older pipework might not be able to handle the required pressures, forces and flows required by new projects. Inspections are performed to measure and monitor the pipework degradation. Completely and thoroughly inspecting all pipework would be the best practice to determine the pipework integrity. Unfortunately, inspecting all the existing pipework would be extremely expensive. In addition to the costs of outsourcing the inspections itself, it would be very time consuming, resulting in an immense impact on operations and planning. Also, due to the presence of insulated pipelines, buried pipelines and inaccessible pipe sections, the accessibility of pipework for inspections can be challenging and therefore extremely expensive. Pipework degradation due to corrosion is an uncertain process. The temporal variability of corrosion makes the decision making regarding maintenance and inspection actions challenging. Therefore, Vopak is calling for an alternative approach in addition to their already existing Risk Based Inspection (RBI) strategy, that more structurally accounts for pipeline degradation uncertainties. The aim of this thesis is to demonstrate a decision making process for inspection and maintenance of pipework, with a structured probabilistic model. The inherent uncertainties of pipeline degradation in time and space are accounted for within a Two-Stage Bayesian model framework. The wall loss, with its variability, is modelled by a Gamma process. Within this framework, also the epistemic uncertainty due to limited available data and lack of knowledge is modelled. The observation of the amount of wall loss, within a certain time span, provides information on the corrosion rate. In the Bayesian framework, the epistemic uncertainty will be reduced each time new observations become available. The amount of localized corrosion is modelled with a Poisson process. The proposed model makes use of the binomial distribution to account for the possibility that some locations may have no observable corrosion and therefore impose no risk to the non-inspected part of the pipeline. In the reliability analysis of a pipeline of interest, this distribution is combined with the calculated corrosion rate distribution, to describe the Exceedance Probability (EP) of an introduced threshold that is associated with reliable pipework operation. The output of the proposed model presents the (EP) in time for a non-inspected pipework of interest. The current- and future state of a pipeline is determined by combining the calculated (EP) with the corresponding consequence of failure, which results in an expression for a Risk of pipeline failure. The decision to either perform extra inspections, maintenance or replacement of pipe sections is formulated based on the Risk of pipeline failure results. Through multiple Case Studies, the results from this approach proved that it can advice on the next inspection date by not accepting the transition of the (EP) towards a higher Risk region. In addition to the already existing (RBI) strategy, the proposed model can be used to analyze a complete pipework system at a terminal. This would provide meaningful assistance for prioritizing pipework inspection and maintenance based on Risk of failure of each pipeline. The model needs some extra professionalization steps in order for Vopak to adapt and implement it in the current (RBI) practice. These steps consist of, but are not limited to, model validation in time, optimization of the Hyper-parameter probability space and calibration of the Coefficient of Variation for different corrosive environments. In addition, the presented inspection coverage method for a sequential inspection strategy shows promising results for future research. Subject CorrosionTwo-Stage BayesianStochastic ProcessesRBIRCMPipeline reliabilityPipeline degradation To reference this document use: http://resolver.tudelft.nl/uuid:94ed78dd-cbbb-49c4-addc-2baf9e68b991 Embargo date 2027-06-02 Part of collection Student theses Document type master thesis Rights (c) 2017 Knijnenburg, L.J.