Blowdown Simulation of CO2 Pipelines

Pipelines are the most practical option for transporting large volumes of captured CO2 to appropriate storage sites as part of the Carbon Capture and Storage (CCS) process. Proper maintenance, including periodic blowdown of pipelines or pipeline sections, is necessary for their safe operation, a pre-requisite for the public acceptance of CCS. Given the relatively high Joule-Thomson coefficient of CO2, blowdown can present significant risks to pipeline infrastructure. Depressurisation will result in rapid cooling of the inventory, potentially to below the CO2 triple point temperature (216 °K); and adjoining pipe wall, which may cool below its ductile to brittle transition temperature, resulting in a significant decrease in its resistance to brittle fracture. In this thesis a rigorous CFD model for pipeline outflow, based on the Euler equations, is coupled with a Finite Element model of heat conduction (referred to hereafter as FEM-O) in order to predict transient pipe wall temperatures during the depressurisation of CO2 pipelines. The Peng Robinson Equation of State (EoS) is selected from a range of EoS including the Soave-Redlich-Kwong, Span and Wagner and GERG 2008 for use with FEM-O. The selection was based on a review of the literature, the accepted computational efficiency of cubic EoS and a comparison of outflow predictions with large-scale experimental data generated by the UK National Grid. New formulations of two and three pipe junction boundary conditions are developed for FEM-O in order to model controlled venting of CO2 pipelines. FEM-O is validated against data gathered from various large-scale dense phase CO2 release experiments conducted by the UK National Grid. These included two full bore rupture experiments of a 144 m long, 0.15 m diameter shock tube, a pseudo-steady state release through two 0.05 m diameter pipes joined in series and the blowdown of a large CO2 pipe system through a 5.88 m long, 0.08 m diameter vertical vent pipe connected to a T-junction. One shock tube experiment utilised a binary mixture of dense phase CO2 with N2. The rest of the tests employed pure, dense phase CO2. Allowing for uncertainty in the experimental data, FEM-O predicted the range and rate of outer pipe wall cooling to ± 4 °C throughout each decompression test. Outer pipe wall temperatures were observed and predicted to fall from ambient temperatures to as low as 247 K over ca. 25 s. Fluid pressure and rapid transient predictions closely matched the experimental data. Fluid temperature was consistently under predicted by FEM-O. For the pseudo steady-state experiment, fluid pressure around the junction of the pipes was under predicted by ca. 5 bara (12 %) and fluid temperature predictions by less than 1 %. No experimental wall temperature data was recorded. For the venting of a pipeline system through a T-junction; FEM-O significantly over predicted fluid and pipe wall temperatures compared to the experimental data. This resulted from the assumption of isentropic fluid flow through the T-junction, which in this experiment caused the model to converge on an unrealistic solution for fluid entropy in the fitting. A verification study was also performed to investigate the performance of the FEM steady state pipe wall temperature calculation algorithm, the sensitivity of the pipe wall temperature predictions to the discretisation of the solution domain and to various different boundary conditions applied. Further, the performance of the newly formulated junction boundary conditions was verified. Lastly a large scale venting experiment was simulated to investigate flow regimes in the inventory. The results demonstrate the minimum requirements for the discretisation of the solution domain in order to maintain accuracy. The uninsulated boundary condition appears to under predict transient wall temperature while the insulated and buried boundary conditions display the expected performance. The new pipeline junction boundary conditions display the expected performance. The large scale venting simulation results suggest the inventory stratifies within seconds of the initiation of venting. The accuracy of FEM-O wall temperature predictions are shown to be dependent on the applicability of the fluid model to the blowdown scenario. For FBR scenarios transient pipe wall temperature predictions agree well with the available experimental data. However improvements cannot be claimed when simulating venting scenarios. The Finite Element computer code has been prepared in modular form and may be readily integrated with other blowdown models