An investigation of gas void fraction and transition conditions for two-phase flow in an annular gap bubble column
2011-06-21T16:14:56Z (GMT) by
Gas-liquid flow may be characterised in terms of the gas void fraction, α. This is an important variable in two-phase flow, used in predicting the occurrence of flow regimes, and the associated pressure drop, and mass and heat transfer. The gas void fraction transitions in a two-phase flow system from uniform bubble flow (homogeneous) to churn-turbulent bubble flow (heterogeneous) in an open tube bubble column (OTBC) and an annular gap bubble column (AGBC) have been investigated using a vertical column with an internal diameter of 0.102 m, containing a range of concentric inner tubes which formed an annular gap; the inner tubes had diameter ratios from 0.25 - 0.69. Gas (air) superficial velocities in the range 0.014-0.200 m/s were studied. Tap water and aqueous solutions of ethanol and isopropanol, with concentrations in the range 8 - 300 ppm by mass, were used as the working liquids. Experimental results are presented to show that there are very significant differences in the mean gas void fractions measured in the OTBC and the AGBC, when operated at the same gas superficial velocity using a porous sparger. The mean gas void fraction decreases with increasing ratio of the inner to outer diameter of the annular gap column and the transition to heterogeneous flow occurs at lower gas superficial velocities and lower void fractions. Two reasons are proposed and validated by experimental investigations: (i) the presence of the inner tube causes large bubbles to form near the sparger, which destabilize the homogeneous bubbly flow and reduce the mean void fraction; this was confirmed by deliberately injecting large bubbles into a homogeneous dispersion of smaller bubbles. Moreover, (ii) the shape of the void fraction profiles changes with gap geometry, which affects the distribution parameter in the drift flux model. Radial profiles of the local void fraction were obtained using a two- and four-point conductivity probe, and were cross-sectionally averaged to give mean values that were within 12% of the volume-averaged gas void fractions obtained from changes in aerated level. The presence of alcohol inhibited the coalescence between the bubbles, and consequently increased the mean gas void fraction at a given gas superficial velocity in both the open tube and the annular gap bubble columns. This effect also extended the range of homogeneous bubbly flow and delayed the transition to heterogeneous flow. Moreover, isopropanol results gave slightly higher mean void fractions compared to those for ethanol at the same mass fraction, due to their increased carbon chain length. It was shown that the void fraction profiles in the annular gap bubble column were far from uniform, leading to lower mean void fractions than were obtained in an open tube for the same gas superficial velocity and liquid composition. The chord length measurements in the OTBC for both the tap water and alcohol solutions exhibited two trends with respect to increasing j_g: (i) at low j_g, in the homogeneous flow, an increasing function was obtained; and (ii) with further increase in j_g, a reduction in the chord length was observed. In the presence of the orifice, the results concerning mean chord lengths show a decreasing function of the bubble size with increasing j_g; this was visually demonstrated using photographs. For the AGBC, the chord lengths obtained from the conductivity probe offered evidence of the bubble size decreasing as j_g increased in the heterogeneous regime, which agreed with the findings of the OTBC. This was also confirmed using the results obtained from photographs. A novel approach for bubble size transformation was implemented to process the conductivity probe measurements. An analytical method was used as a forward transform to predict the chord length distribution from the bubble size distribution and an optimisation approach was applied as a backward transform method to obtain the bubble size distribution from the chord length distribution. The challenge was to consider a variable aspect ratio, φ, for the bubble shape, which depended on their size. The model gave excellent and reasonable predictions for the bubble sizes as their trends were identical to the trend of the chord length, and to the bubble size obtained from photographs.