Yi, Jiahuan; (2024) A Multi-Faceted Statistical Approach for Safety Analysis of Pressurised CO2 Transmission Pipelines. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text PhD thesis_Jiahuan Yi.pdf - Submitted Version Download (4MB) | Preview

Abstract

Global efforts to reach net zero emissions targets rely heavily on Carbon Capture Utilisation and Storage (CCUS) for decarbonising unabated coal power stations and industrial emissions sources such as refineries, cement and steel making industries. An essential element of the CCUS chain involves the large-scale transportation of the captured CO2 for permanent geological storage or as a feedstock for utilisation to produce chemicals or fuels. Pressurised pipelines are widely considered as the safest and most economical CO2 transport option. By 2050, the amount of captured CO2 is expected to increase significantly reaching ca. 7.6 Gt, requiring a vast network of 200,000 to 550,000 km of CO2 pipelines. Given that CO2 is increasingly toxic at concentrations over 7% vol/vol, and the large amounts involved, the failure of CO2 pipelines poses serious risks of fatalities, environmental damage, and economic losses. As such, ensuring the safe operation of such pipelines is of paramount importance to the public acceptability of CCUS as a viable means for tackling climate change. Central to the above is the reliable quantification of the risks posed by such pipelines in the event of an accidental failure. In essence the above involves three main steps namely, 1) modelling pipeline decompression to predict the outflow characteristics following failure, 2) performing quantitative risk assessment to evaluate the failure consequences, and 3) implementing emergency response planning strategies to mitigate the failure consequences to as low as reasonably practicable. This thesis presents the development and assessment of rigorous mathematical techniques for conducting such work. These include the development of a computationally efficient pressurised pipeline decompression model, an analytical approach for estimating pipe failure hole size distribution probability and a probabilistic Multi-Objective Optimisation (MOO) technique for optimising inline Emergency Shutdown Valve (ESDV) configuration. The computationally efficient pressurised pipeline decompression model is developed as a fundamental extension of a previously developed analytically based Vessel Blowdown Model (VBM) for simulating the transient outflow following the accidental failure of high-pressure pipelines. Based on the modification of the standard vessel discharge equations through incorporating additional inflow terms, the extended model addresses the fundamental limitations of VBM in handling un-isolated releases and fluid/wall heat exchanges. The new model is successfully tested against the results obtained using an extensively validated but computationally demanding numerical pipeline decompression model by simulating the failure of a hypothetical pressurised methane pipeline initially at 21 bar and 300 K. The verification tests include various feed flow rates (1 to 7.5 kg/s), pipe lengths (100 to 5,000 m) and puncture to pipe internal diameter ratios (0.2 to 0.8), producing a maximum disagreement of ca. ±7% between the two models’ predictions. The reliability of pipeline failure hole size probability distribution estimation heavily relies upon the availability of sufficiently large pool of historical data. Currently, this is an issue for CO2 pipelines given their relatively small number in operation. In this part of the thesis, the development of an analytical approach capable of addressing the above issue is presented. The procedure involves fitting statistical probability distributions to the historical failure hole size data using the maximum likelihood estimator, complemented by using bootstrapping method to improve the estimation confidence. The application of the above technique to both pressurised CO2 and hydrocarbon pipelines indicates that compared to the latter, CO2 pipelines, with at least 80% of their failures corresponding to punctures smaller than 50 mm, are more likely to experience smaller puncture failures, thus resulting in smaller magnitude but more prolonged releases. This directly impacts the preventive and emergency response planning as well as failure detection techniques required especially in the case of buried CO2 pipelines where small leaks can remain undetected for long periods. The final part of this thesis deals with the development and application of a probabilistic MOO technique for selecting the optimal inline ESDV configuration for pressurised CO2 pipelines by, for the first time, accounting for operational and failure uncertainties as probabilistic variables. Based on a case study for a 300 km, 309.6 mm internal diameter CO2 pipeline that operates at 129 bar and 307.24 K for a real CCS project, the MOO technique is applied and assessed to ascertain its effectiveness in curbing the risks identified from the case study while minimising ESDV costs. Starting with modelling the uncertainties in the important pipeline characteristics and operating conditions using standard Probability Density Functions (PDFs), a Monte Carlo simulation involving the random sampling from these PDFs is performed to obtain the probability distribution of the risk associated with pipeline failure. Based on the obtained probability distribution, the risks are mapped onto the objective function space to create a probabilistic solution plane for the decision makers to determine the optimal ESDV configuration. The efficacy of the proposed technique is demonstrated using a comparative study where the risk is treated deterministically as the worst-case scenario. The findings reveal that more cost-effective risk mitigation solutions can be attained when the risk is taken probabilistically, highlighting the importance of incorporating operational and failure uncertainties as part of the decision-making process when configuring in-line ESDVs for CO2 pipelines.

Download activity - last month

Export as CSVJSONXML

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item