2008 Spring Meeting & 4th Global Congress on Process Safety

(65b) Gas-Liquid Taylor Flow Characteristics in Straight and Meandering Rectangular Microchannels

Authors

Aloisiyus Y. Widianto - Presenter, Laboratoire de Génie Chimique UMR 5503 CNRS
Joelle Aubin - Presenter, Laboratoire de Génie Chimique UMR 5503 CNRS
Catherine Xuereb - Presenter, Laboratoire de Génie Chimique UMR 5503 CNRS
Norbert Völkel - Presenter, Laboratoire de Génie Chimique UMR 5503 CNRS


Taylor flow is a gas-liquid two-phase flow pattern which appears in microchannels and capillaries over a wide range of flow rates. It is characterized by elongated bubbles separated form each other by liquid slugs (Figure 1). The bubbles themselves are separated from the channel wall by a liquid film. A particular feature of
Taylor flow is the velocity field within the liquid slugs; two symmetrical vortices are created at both sides of the channel centre line. This behavior is considered as advantageous for mass transfer and turns the Taylor flow regime into an interesting operating condition for gas-liquid or gas-liquid-solid microreactors (solid catalyst deposited on the channel walls) [1].

Figure 1: Example image of
Taylor flow through a rectangular microchannels used in the frame of the present study (lS = slug length, lB = bubble length).

In order to provide sufficient residence time for a reactive system, typically long microchannels are needed. Since one of the design requirements of miniaturized devices is that they are compact, an obvious solution is to employ meandering channel geometries. In this case however, the channel bends induce a noticeable distortion of the recirculation motion within the liquid slugs [2]. The present work is part of a study, whose aim is to investigate if (and under which conditions) gas-liquid mass transfer in
Taylor flow is enhanced by perturbations of the liquid velocity field. For this purpose three different rectangular microchannel configurations are used; one straight channel and two meandering channels (turning angle 180°), the latter two differing from each other in their bend-design (Figure 2). The microchannels were manufactured by etching a silicon wafer from the top to the bottom and enclosing it within two pyrex glass plates by anodic bonding (Figure 2(d)).

Figure 2: Microchannel geometries used in this study with w = 1000 μm, h = 500 μm; gas-inlet channel with w = 525μm, h = 500 μm. (a) straight channel l = 70 mm, (b) meandering channel with curved bends, (c) meandering channel with sharp corner bends, (b) and (c) l ≈ 300 mm, (d) cross-sectional sketch of the microchannel plate.

For
Taylor flow, the contact time between the bubble and the liquid film plays an important role on the gas-liquid mass transfer. It depends on bubble length, slug length, bubble velocity and liquid film flow [3, 4]. For this reason, a preliminary study of
Taylor flow hydrodynamics was carried out for the three microchannels mentioned above. Air and ethanol were employed as working fluids at a temperature of 25°C (Bo = 0.16). The ranges of gas and liquid flow rates, as well as the characteristic values of the studied systems are given in Table 1. Movies of
Taylor flow were recorded at the fixed distance from the inlet section (centre point at x/dCh ~ 52) using a high speed camera.

Table 1: Ranges of volumetric flow rates with resulting superficial velocities used in this investigation and their corresponding superficial dimensionless numbers (working fluids are air and ethanol).

In Figure 3, the superficial two phase velocity (UTP) is plotted versus the bubble velocity (UB) data. The experimental results are compared with correlations given by former
Taylor flow studies in the literature. Firstly, it can be seen that the values measured in the straight channel are noticeably different from those found for the two meandering channels. This suggests that the bubble and slug lengths, as well as the bubble velocities are also a function of the two phase pressure drop. Based on a mass balance relationship derived for the
Taylor flow regime, it can be stated that this graphical domain is divided into two regions by a parity line, where either the relative velocity between the bubble and the liquid film is zero or the liquid film does not exist. In the region above the parity line, the slope is greater than 1 and the liquid film flows with a zero velocity or a velocity which is smaller than the bubble velocity. In the lower region of the domain, the liquid film velocity is greater than the bubble velocity and the slope of the linear plots is less than 1.

Figure 3: Superficial two-phase velocity (UTP = UGS+ULS) versus bubble velocities (the lines represent correlations proposed in the literature; the resulting range of Capillary numbers is 2.33•10-3 ≤Ca≤ 6.34·10-3).

Generally for horizontal
Taylor flow [5, 6, 7], the liquid film is assumed to be static, which suggests that the data should fall in the upper region of the domain. From Figure 3 it can be seen that the experimental results, as well as the literature correlation [10] for horizontal
Taylor flow fall in the lower region of the plot. Hence, this suggests that the liquid film appears not to be static.

Figure 4: Superficial phase velocity ration (UGS/ULS) versus (a) dimensionless bubble length (lB/dCH), (b) superficial phase velocity ratio (UGS/ULS) versus dimensionless slug length (lS/dCH), the drawn lines represent prediction criteria proposed by the literature, results obtained within a range of 2.33•10-3 ≤ Ca ≤ 6.34•10-3.

In Figure 4(a) it can be seen that the bubble length data for both meandering channels can be well predicted by the correlation given by Garstecki et al. [11], whilst the correlation proposed by Qian and Lawal [12] always over predicts the measured values. With respect to the slug length (Figure 4(b)), it can be seen that the experimental results generally do not follow the correlations [12, 13]. Furthermore, with an increasing superficial phase velocity ratio, the bubble length increases linearly, whereas the corresponding slug length decreases and tends towards an asymptotic limit. Since the cross-sectional bubble profile is assumed to remain fixed within the range of Capillary numbers studied here, a static behavior of the liquid film would also result in a linear relationship regarding the trend of the liquid slug lengths. Thus, this further suggests the possible existence of a liquid film flow.

In this paper we will use the findings of this hydrodynamic study to improve the existing models for the evaluation and estimation of gas-liquid mass transfer of
Taylor flow in horizontal microchannels. In particular, a non-static liquid film will be accounted for. The sensitivity of the gas-liquid mass transfer to the characteristics of the liquid film will be evaluated.

References:

[1] Kreutzer, M. T., Kapteijn, F., Moulijn, J. A., Heiszwolf, J.J., Chem. Eng. Sci., 60, 5895-5916, 2005.

[2] Günther, A., Khan, S. A., Thalmann, M., Trachsel, F., Jensen, K. F., Lab Chip, 4, 278-286, 2004.

[3] Vandu, C. O., Liu, H., Krishna, R., Chem. Eng. Sci., 60, 6430-6437, 2005.

[4] van Baten, J. M., Krishna, R., Chem. Eng. Sci., 59, 2535-2545, 2004.

[5] Warnier, M. J. F., Rebrov, E. V., de Croon, M. H. J. M., Hessel, V., Schouten, J. C., Chem. Eng. J., 2007. (in press)

[6] Suo, M., Griffith, P., ASME J. Basic
Eng., 86, 576-582, 1964.

[7] Thulasidas, T.C., Abraham, M. A., Cerro, R. L., Chem.
Eng.Sci., 50, 183-199, 1995.

[8] Liu, H., Vandu, C. O., Krishna, R., Ind. Eng. Chem. Res., 44, 4884-4897, 2005.

[9] Laborie, S., Cabassud, C., Durand-Bourlier, L., Laine, J. M., Chem. Eng. Sci., 54, 5723-5735, 1999.

[10] Chen, W. L., Twu, M. C., Pan, C., Int. J. Multiphase Flow, 28, 1235-1247, 2002.

[11] Garstecki, P., Fuerstman, M. J., Stone H. A., Whitesides, G. M., Lab Chip, 6, 437-446, 2006.

[12] Qian, D., Lawal, A, Chem.
Eng.Sci., 61, 7609-7625, 2006.

[13] Heiszwolf, J. J., Kreutzer, M. T., van den Eijnden, M. G., Kapteijn, F., Moulijn, J. A., Catal. Today, 69, 51-55, 2001.