Assessment of Failure Frequency Methodology Applied to Dense Phase Carbon Dioxide Pipelines

Abstract

This paper determines the capability of current failure frequency methodology and models for assessing dense phase CO2 pipelines. An overview of these models is given in order to explain their applicability to CO2 pipelines, structure and to highlight the similarities and differences between each model. A pipeline must comply with the relevant design code in order to ensure conformity with pipeline safety regulations, for example in the UK it is PD-8010 (BSI, 2015). In this particular code the maximum allowable operating stress in the pipeline and the minimum allowable distance between the pipeline and any occupied buildings near to the pipeline route must be defined by the product being transported. The code was originally written to be applied to products which pose a thermal hazard and therefore there is currently no hazard category included in the code for dense phase CO2 and therefore cannot be applied to the design of dense phase CO2 pipelines. Where no design code for dense phase CO2 pipelines exists, to ensure the safe design, construction and operation, a quantitative risk assessment (QRA) approach is required (Barnett, 2014; Cooper, 2014). The purpose of a pipeline QRA is to determine the risks posed by the failure of the pipeline to people located nearby. The procedure involves the identification of hazard scenarios and considers both the probability of failure and the consequences of failure in order to calculate values for the individual and societal risk due to the pipeline. In general a QRA procedure covers the following steps: i) Identify hazards ii) Identify failure causes iii) Calculate the frequency of failure for each cause iv) Evaluate the consequences of failure v) Calculate the individual and societal risk a specific locations along the pipeline vi) Assess the tolerability of the calculated risks by comparison with recognised criteria In terms of failure causes, pipeline failure can occur due to numerous different mechanisms including, third party external interference, corrosion, material or construction defects, natural events and operational error; all of which must be considered as part of the assessment (Goodfellow, 2006). For oil and gas pipelines, the frequency of pipeline failure due to third party external interference has traditionally been calculated using models based upon probabilistic, structural reliability methods. Structural reliability methods are applied by combining the following: • Limit state functions, the mathematical models which define the conditions for failure; • Probability distributions based around selected random variables; • A mathematical technique to calculate the probability of failure (e.g. Numerical Integration, Monte Carlo, First Order Reliability Methods). For pipelines the limit state functions are based on semi-empirical fracture mechanics failure models; and the probability distributions are based on pipeline damage and derived from historical operational data. The failure probability is converted into a failure frequency to take into account the regularity of third party external interference damage. The various models currently in use within the oil and natural gas pipeline industry differ in their subtleties; however all are based upon a methodology originally developed in the 1980s. To calculate pipeline failure frequency for dense phase CO2 pipelines it would be desirable to extend the use of the current pipeline failure frequency methodology. The methodology has been employed for over 25 years, and as a result is tried, tested and well understood. The transportation of dense phase CO2 by pipeline however requires operational pressures in excess of the CO2 triple point; potentially up to 200bara. This high design pressure requirement necessitates the use of thick wall linepipe in pipeline construction, potentially with wall thickness dimensions outside of the limits of current operational experience. Consequently, the reliance of the failure frequency methodology on empirical data and semi-empirical relations needs to be examined. The requirement to develop a robust Quantitative Risk Assessment (QRA) methodology for high pressure CO2 pipelines has been recognised as critical to the implementation of CCS. Consequently, failure frequency models are required that are appropriate for high pressure CO2 pipelines. This paper addresses key components from the failure frequency for QRA methodology development.