Slug Flow-Induced Vibration and Vortex-Induced Vibration of Catenary Risers

Abstract

Subsea risers are widely used in the offshore oil and gas industry for deep-water productions where hydrocarbon flows of multiple phases may appear in several flow patterns. Characterized by the alternating liquid- and gas-dominant regions, slug flows are practically problematic due to a fluctuation in flow momenta and pressure potentially causing slug flow-induced vibration (SIV). Compared with a single-phase flow-induced vibration, fundamental understanding of SIV phenomena and the associated effects on the dynamics of catenary risers are still lacking. Therefore, the present thesis aims to investigate SIV of flexible catenary risers subject to external excitations by the vortex-induced vibration (VIV). The literature of multiphase flows (gas-liquid flow patterns, flow maps) and flow-induced vibrations (SIV, VIV) is reviewed. A mechanistic steady-state slug model is considered to provide the slug flow-induced loads with a two-dimensional continuum riser model. A numerical study for fundamental planar dynamics of an inclined curved flexible riser carrying slug gas-liquid flows is conducted. Results demonstrate several SIV features and effects of slug flow characteristics (slug unit length, translational velocity and fluctuation frequencies) on the riser resonant dynamics. These fundamental observations enable a further study with combined VIV-SIV phenomena. To incorporate the VIV effect, a phenomenological model based on wake oscillators is used and first implemented for an elastically mounted rigid cylinder in uniform flows subject to cross-flow/in-line VIV. The dynamical system described by coupled nonlinear cylinder-wake oscillators is solved by a numerical-analytical approach. Several important VIV characteristics are captured, highlighting abilities of this semi-empirical model in VIV prediction. By applying the distributed van der Pol wake-oscillators, this low-order model is then extended to the analysis of cross-flow-only VIV of a catenary riser under uniform flows perpendicular to its initial curvature plane. Some experimentally observed VIV features of flexible cylinders are numerically predicted. By comparing VIV-only with VIV-SIV responses, VIV is found to prevail in the combined external-internal excitation cases. Slug flows generally result in a significant multi-mode VIV, which becomes more pronounced at higher external flow velocities: this is attributed to the effect of softening stiffness of the pipe caused by internal slug flow mass and pressure leading to a higher-mode VIV. To validate the numerical model of SIV, a small-scale flexible pipe conveying gas-liquid flows has been investigated experimentally for a catenary configuration. The experiment is carried out in an air-water test loop with a hanging tube section made of silica gel. Both slug flow-induced pipe motions and gas-liquid flow patterns are recorded simultaneously using non-intrusive high-speed cameras. Pressure variations are also measured at the pipe inlet and outlet by two pressure transducers. The flow-transporting system is tested by employing different gas-liquid flow rates. Occurrence of slug flows is captured at the relatively high ratios of the gas to liquid superficial velocities rendering a large-amplitude SIV of the flexible pipe. Under different flow conditions, slug flow characteristics are observed to vary significantly in terms of the travelling velocities, lengths and frequencies of different slug units. These entail the unsteady SIV with modulated amplitudes and frequencies. Some qualitative aspects of slug characteristics and SIV responses from numerical predictions are captured in the laboratory tests. Through comparisons of riser vibration responses and frequencies between numerical and experimental results, the prediction abilities of the mathematical model are recognized. In summary, this thesis has investigated fundamental and potential SIV effects on catenary risers. A significant role of the slug unit length and the slug translational velocity should be considered individually for determining large-amplitude SIV which is of practical importance from a design viewpoint. Possible amplifications of riser responses due to slug flows are remarked in combined VIV-SIV scenarios. This is meaningful for the riser stress, fatigue and failure assessment. Although the present numerical model is limited to steady-state slug flows, it paves a way for a future development of computationally efficient tools benefiting the screening analysis of subsea risers transporting gas-liquid slug flows.