| Popis: |
In this thesis slug flow in microchannels was investigated experimentally and theoretically with the aim to provide design equations enabling construction of multistep microfluidic networks. Two major aspects were studied: pressure drop in channels of various geometries and subsequent phase separation of generated droplets. Mathematical modeling of phenomena observed on microscale was validated by extensive experimental studies. Pressure drop of three-phase gas-liquid-liquid slug flow was investigated in capillaries of round cross section. Pressure losses were modeled combining frictional and interfacial components using Hagen-Poiseuille ∆P model for laminar flow and expression based on work of Bretherton to account for the presence of bubbles and droplets in the flow. Developed model is free of empirical parameters and only droplet frequency needs to be known a priori in order to calculate pressure losses. Experimental validation was performed using oleic acid-water-nitrogen and heptane-water-nitrogen slug flow in Teflon or Radel-R capillaries. Proposed model allowed for prediction of pressure drop with 20% accuracy. During the experiments influence of surface roughness on pressure drop and occurence of dry slug flow were also analysed, both induced unexpected pressure losses which were rarely quantified before. Based on the three-phase flow study, a more complex problem of slug flow in square channels was analysed. Liquid-liquid slug flow of toluene-water and silicone oil-water was studied in channels fabricated in silicon/glass. The same modeling approach was adopted with respective frictional and interfacial pressure drop correlations modified to account for the polygonal channel geometry. No empirical parameters were necessary but both droplet frequency and velocity had to be measured experimentally to enable the pressure drop prediction. Two models, one taking into account the presence of moving liquid film around the droplets and the other assuming lack thereof, were developed and subsequently enabled analysis of the influence of the lubricating film on pressure drop. While both correlations performed well for flows with similar viscosities of both phases, large discrepancies were discovered for liquid flows with viscosity ratio significantly different from one. First attempt has been made to use pressure drop predictions to calculate droplet frequency however it was found that an accuracy of about 10 % in pressure loss estimation is necessary to allow reliable fd determination. Finally separation of liquid-liquid slug flow in so-called capillary phase separators was studied. In those devices, dispersed phase is prevented from exiting through the capillary outlets by the balance between the capillary pressure and the pressure drop across narrow side-channels. Operating limits of the device were modeled with Young-Laplace equation to describe the upper pressure limit; a new model was developed for the lower pressure limit, based on the changing length of the slug and thus varying number of available capillaries as the slug is withdrawn through the separator. Operating boundaries were determined experimentally by studying water-toluene slug flow in glass/silicon devices, five different geometries were tested. Calculated operating diagrams were compared with experiments and showed better performance than the models used in literature. Prediction of the operating limits of the separator fully independent of experimental data was made possible by applying a modified scaling law to predict droplet and slug lengths formed in a T-junction. Developed models constitute a theoretical framework to design a network of microfluidic channels where various process units such as reactors, extractors, mixers, etc. may be implemented. Correct prediction of pressure drop and of the operating region of a capillary phase separator enables integration of multiphase flows into such setup. Thus it is a step towards a realisation of a fully versatile microfluidic platform for the applications in flow chemistry. |
