Modelling the non-equilibrium two-phase flow during depressurisation of CO 2 pipelines

Abstract

a b s t r a c tThe development, testing and validation of a two-fluid transient flow model for simulating outflow following the failure of high pressure CO 2 pipelines is presented. Thermal and mechanical non-equilibrium effects during depressurisation are accounted for by utilising simple constitutive relations describing inter-phase mass, heat and momentum transfer in terms of relaxation to equilibrium. Pipe wall/fluid heat exchange on the other hand is modelled by coupling the fluid model with a finite difference transient heat conduction model. The two-fluid transient flow model's performance is tested by comparison of the predicted transient pressure and temperature profiles along the pipeline against those based on the simplified homogeneous equilibrium model (HEM) as well as real data captured during the full bore rupture of a 260 m long, 233 mm internal diameter pipeline containing CO 2 at 36 bara and 273 • C. The two-fluid model is found to produce a reasonably good degree of agreement with the experimental data throughout the depressurisation process. The HEM based flow model on the other hand performs well only near the rupture plane and during the early stages of the depressurisation process.

Rapid depressurization of a CO2 pipeline – an experimental study

“…The thermodynamic behavior of CO 2 during a rapid depressurization of a pipeline has been experimentally studied (de Koeijer et al, 2009(de Koeijer et al, , 2011Brown et al, 2014;Håvelsrud, 2012;Huh et al, 2014) and simulated using CFD tools (de Koeijer et al, 2009(de Koeijer et al, , 2011Brown et al, 2014;Håvelsrud, 2012;Huh et al, 2014;Witlox et al, 2009;Witlox et al, 2011;Dixon et al, 2012;Calay and Holdo, 2008;Wen et al, 2011;Jie et al, 2012;Munkejord et al, 2010;Lund et al, 2011;Clausen and Munkejord, 2012;Giljarhus et al, 2012;Lund and Aursand, 2012;Munkejord et al, 2013). In order to develop accurate models for predicting the expansion behavior of CO 2 during a rapid depressurization of a CO 2 pipeline, more experimental data is required (Lund and Aursand, 2012).…”

“…The experimental data were used to verify model predictions. Another experimental and modelling study (Brown et al, 2014) used experimental results of a full bore rupture of a 256 m, 233 mm inner diameter pipeline for testing a two-fluid transient flow model for non-equilibrium flow. An experimental study was conducted in the frame work of the Dutch national Carbon Capture and Storage (CCS) research program (CATO2) by studying the thermodynamic behavior of CO 2 during the release from a high pressure vessel through different size nozzles (Ahmad et al, 2013).…”

Multi‐phase decompression modeling of CO2 pipelines

“…Although results predicted by HEM were in reasonable agreement with the measurements, there are still relatively great discrepancies between simulation and measurements under some conditions. 10,31 The delayed nucleation probably affects the decompression characteristics of the fluid. 8,26 It was suspected that the discrepancies were introduced by the thermal equilibrium assumption in HEM.…”

“…8,26 It was suspected that the discrepancies were introduced by the thermal equilibrium assumption in HEM. 10,31,36,37 Brown et al 10,31 proposed a two-phase transient flow model using HRM for simulating the flow following failure of a high-pressure CO 2 pipeline. Some researchers have proposed a Homogenous Relaxation Model (HRM).…”

A geometrically based grid refinement technique for multiphase flows

“…Hence, to complement the modelling strategies described above, the use of Adaptive Mesh Refinement (AMR) techniques represents another possible tool for further efficiency. The application of AMR for CFD is widespread (Pelanti and LeVeque, 2006;Gourma et al, 2013;Brown et al, 2014), with various methodologies applied ranging from the popular hierarchical box-structured techniques, first described by Berger and Oliger (1984), to moving grid methods (see for example Tang and Tang, 2003;Kelling et al, 2014). The former begin with an underlying coarse grid and proceed to identify and refine areas to a predefined level where required; though successful, the hierarchical approach requires the implementation of complex data structures (Berger, 1991) and a means for dealing with the artificial internal boundaries at the refined regions (Berger and Oliger, 1984;Berger and Le Veque, 1998).…”