The experimental study of counter-current two-phase flows in annuli

Abstract

Counter‑current two‑phase gas‑liquid flows in annuli occur in many industrial applications, including coal seam gas (CSG) production. In CSG wells, gas is typically produced through counter‑current gas‑water flow in an annulus. Yet, the models used by CSG companies were designed for conventional wells, wherein co‑current upward gas-liquid flow occurs in pipes. Accurately predicting the ensuing flow regimes and counter‑current flow limitation (CCFL) is important because they are intrinsically linked to flow characteristics, and dictate operating conditions. The CCFL describes the onset of co‑current upward or downward flow due to high gas (gas flooding) or liquid (liquid flooding) rates, respectively. There are currently no experimental publications on counter‑current gas‑liquid flow regimes in annuli, prohibiting predictive models from being validated. I collated 3,947 flow regime results from 35 experimental studies of counter‑current flows in pipes and co‑current flows in pipes and annuli, and confirmed that flow configuration, channel geometry, and fluid properties can significantly influence flow regimes (Chapter 2). This thesis aims to experimentally characterise flow regimes and the CCFL during counter-current two-phase flows in an annulus. These findings will allow predictive models of counter-current flow regimes in annuli to be validated.To study counter‑current flow regimes in an annulus (Chapter 3), I designed and built an experimental apparatus using acrylic pipes with annulus diameters of 170 mm and 70 mm (170/70 annulus). Flow regimes and the CCFL were documented for air-water at standard conditions using combinations of superficial gas velocities within 0.014‑5.794 m/s and superficial liquid velocities within 0.004‑0.240 m/s. Differential pressure data were recorded over 1 m and used to infer void fractions, and to quantitatively assess the flows using fast Fourier transform (FFT). High-speed video images were captured at 4,000 frames/s. An apparatus with annulus diameters of 44 mm and 19 mm was also assembled to investigate the impact of annulus size.My observations confirm the 170/70 annulus is a large annulus in which stable slug flow regime cannot develop due to Rayleigh‑Taylor instability. Subsequently, the flow regimes in this system are described as homogeneous and heterogeneous instead of bubble and slug‑churn regimes. Churn‑annular flow was observed, but, the CCFL was reached before a fully annular regime could develop. Existing empirical correlations for the CCFL only considered gas flooding and performed poorly when assessed against my experimental results. So, I developed an empirical correlation for the CCFL that also incorporated liquid flooding. Liquid flooding was detected using a novel experimental technique developed in Chapter 5. Essentially, a water barrel beneath the annulus, used for water re‑circulation, was dual‑purposed as a vertical separator. The proposed correlation for CCFL in an annulus is , where and are the dimensionless gas and liquid superficial velocities, respectively, and the annulus size is defined by its average circumference.In Chapter 4, I developed a new procedure to characterise flow regimes by applying four signal processing techniques (autocorrelation, power spectral density, Shannon entropy, and permutation entropy) to the pressure fluctuation signals associated with flow regimes. Autocorrelation could also identify the previously indistinguishable homogeneous‑heterogeneous transition zone at superficial gas velocities within 0.222‑0.539 m/s and superficial liquid velocities within 0.016‑0.064 m/s. My results indicate that churn‑annular flow is a transition towards annular regime rather than a distinct flow regime. Since annular regime was not observed, this transition is described as post‑heterogeneous transition to avoid confusion. Post‑heterogeneous transition could be classified using Shannon entropy or permutation entropy at superficial gas velocities within 3.550‑5.387 m/s and superficial liquid velocities within 0.016‑0.044 m/s.Signal analysis results and high‑speed video images from both Chapters 3 and 4 also provided insights into the physical mechanisms responsible for the development of counter‑current flow regimes and CCFL in annuli. These mechanisms were experimentally identified as: (i) homogeneous‑heterogeneous transition - agglomeration and coalescence of small bubbles; (ii) post‑heterogeneous transition - destruction of the liquid slug or wave that can be sustained as small droplets in a gas core; and (iii) gas flooding – formation and upward propagation of large liquid waves near the liquid inlet.I also investigated the effect of salinity on flow regimes (Chapter 6) using formation water extracted from a CSG well, and sodium chloride solutions with concentrations of 11,198±142 ppm and 2,863±13 ppm. My results confirmed that salinity inhibits bubble coalescence, skewing the bubble size distribution towards smaller bubbles and increasing void fraction by up to 6%. A stabilizing effect on the homogeneous regime was also observed, with the transition occurring at up to 9% (33→42%) greater void fraction and 0.111 m/s (0.200→0.311 m/s) greater superficial gas velocity compared with fresh water. Post-heterogeneous transition was largely unaffected as its physical mechanisms are not related to bubble coalescence.This study has provided valuable insights into counter-current two-phase flows in an annulus. Furthermore, the outcomes of this work make possible the validation of predictive flow models developed specifically for the CSG industry.