Multiphase flows occurring in pipelines with a serpentine configuration is an important phenomenon, which can be encountered in heat exchangers used in a variety of industrial processes. More specifically, in many industrial units such as a large cracking furnace in a refinery, the tubes are arranged in a serpentine manner and are relatively short. As flow negotiates round the 180o bend at the ends of the tubes, the generated centrifugal force could cause flow maldistribution creating local dry spots, where no steady liquid film is formed on the adjacent straight sections of the pipe. As a result, events including coking, cracking and overheating of heat transfer surfaces may occur and lead to frequent shutdown of the facilities. Consequently, this could increase operating costs and reduce production revenue. Thus, it is desirable to know the effect that the bends exert on the flow in the straight part of the pipe. Apart from this, knowledge of the bend effects on the flows in the pipeline could also be important for the design of other pipelines for gas/liquid transport, e.g. offshore gas and oil pipelines. Quite a large number of studies have been found in the literature. The majority of them were for two-phase flow with small diameter pipes (i.d. ≤ 50 mm). However, studies with large diameter pipes (i.d. ≥ 100 mm), have increasingly been considered in recent years as problems related to large diameter vertical pipes are being encountered more and more often in industrial situations. This thesis studies the effect of 180o bends on the characteristics and development of gas-liquid two-phase flows in large diameter downward and upward pipes. The study particularly focuses on the influence of serpentine configuration on flow structure, cross-sectional void distribution and circumferential liquid film profiles and their development along the downward and upward sections. It was found that both the top and bottom bends have considerable impacts on flow behaviour, although to varying degrees. These impacts were highly dependent on the air and water flow rates. For sufficient flow rates, the bends were observed to create flow maldistribution in the adjacent straight section, due to the effects of centrifugal force. The air moved towards the inner zone of the bend and the water towards the outer zone, while a lesser quantity of water was identified on the other surfaces of the pipe. Investigation of the film thickness development in the downward and upward sections showed that, the liquid film behaviour close to the bends was significantly different from those located further away. This can be attributed to the centrifugal force of the bends. Examination of the power spectral density (PSD) along the downward and upward sections showed that, the shape of PSD located in the adjacent section to the bends, was substantially different from those located further away. Furthermore, several flow regime maps were generated which showed that, in addition to bubbly, intermittent and annular flows, unstable flows existed along the upward section, particularly for low gas and water flow rates. In this study it was found that, the lower bend was periodically blocked by the liquid and then blown through by the accumulated air. The data obtained from this study were compared with different theoretical correlations found in the existing literature. Some discrepancy between the results of the current study and those of previous published materials was noted. Updated correlations were presented which provided well results when they applied for the data obtained from the current study and previous studies.

The works aimed at the further development and validation of models for CFD codes. For this reason, the new thermal-hydraulic test facility TOPFLOW was erected and equipped with wire-mesh sensors with high spatial and time resolution. Vertical test sections with nominal diameters of DN50 and DN200 operating with air-water as well as steam-water two-phase flows provided results on the evaluation of flow patterns, on the be¬haviour of the interfacial area as well as on interfacial momentum and heat transfer. The validation of the CFD-code for complex geometries was carried out using 3D void fraction and velocity distributions obtained in an experiment with an asymmetric obstacle in the large DN200 test section. With respect to free surface flows, stratified co- and counter-current flows as well as slug flows were studied in two horizontal test channels made from acrylic glass. Post-test calculations of these experiments succeeded in predicting the slug formation process. Corresponding to the main goal of the project, the experimental data was used for the model development. For vertical flows, the emphasis was put on lateral bubble forces (e.g. lift force). Different constitutive laws were tested using a Multi Bubble Size Class Test Solver that has been developed for this purpose. Basing on the results a generalized inhomogeneous Multiple Size Group (MUSIG) Model has been proposed and implemented into the CFD code CFX (ANSYS). Validation calculations with the new code resulted in the conclusion that particularly the models for bubble coalescence and fragmentation need further optimisation. Studies of single effects, like the assessment of turbulent dissipation in a bubbly flow and the analysis of trajectories of single bubbles near the wall, supplied other important results of the project.

Ziel der Arbeiten war die Weiterentwicklung und Validierung von Modellen in CFD-Codes. Hierzu wurde am FZD die thermohydraulische Versuchsanlage TOPFLOW aufgebaut und mit räumlich und zeitlich hochauflösenden Gittersensoren ausgestattet. Vertikale Teststrecken mit Nenndurchmessern von DN50 bzw. DN200 für Luft/Wasser- sowie Dampf/Wasser-Strömungen lieferten Ergebnisse zur Entwicklung von Strömungsformen, zum Verhalten der Zwischenphasengrenzfläche sowie zum Wärme- und Impulsaustausch zwischen den Phasen. Die Validierung des CFD-Codes in komplexen Geometrien erfolgte anhand von 3D Gasgehalts- und Geschwindigkeitsfeldern, die bei Umströmung eines asymmetrischen Hindernisses auftreten, das in der Teststrecke DN200 eingebaut war. Im Hinblick auf Strömungen mit freier Oberfläche untersuchte das FZD in zwei horizontalen Acrylglas-Kanälen geschichtete Zweiphasenströmungen im Gleich- bzw. Gegenstrom sowie Schwallströmungen. Bei den Nachrechnungen dieser Versuche gelang die Simulation der Schwallentstehung. Entsprechend des Projektziels wurden die experimentellen Ergebnisse zur Modellentwicklung genutzt. Bei vertikalen Strömungen stand die Wirkung der lateralen Blasenkräfte (z.B. Liftkraft) im Vordergrund. Zum Test unterschiedlicher Modellansätze wurde hierzu ein Mehrblasenklassen-Testsolver entwickelt und genutzt. Darauf aufbauend wurde ein neues Konzept für ein Mehrblasenklassenmodell, das Inhomogene MUSIG Modell erarbeitet und in den kommerziellen CFD Code CFX (ANSYS) implementiert. Bei Validierungsrechnungen zeigte sich, dass vor allem die Blasenkoaleszenz- und -zerfallsmodelle weiter optimiert werden müssen. Untersuchungen zu Einzeleffekten, wie z.B. die Abschätzung von Turbulenzkoeffizienten und die Analyse der Trajektoren von Einzelblasen in unmittelbarer Wandnähe, lieferten weitere wichtige Ergebnisse des Projekts.

Multiphase flows occurring in pipelines with a serpentine configuration is an important phenomenon, which can be encountered in heat exchangers used in a variety of industrial processes. More specifically, in many industrial units such as a large cracking furnace in a refinery, the tubes are arranged in a serpentine manner and are relatively short. As flow negotiates round the 180o bend at the ends of the tubes, the generated centrifugal force could cause flow maldistribution creating local dry spots, where no steady liquid film is formed on the adjacent straight sections of the pipe. As a result, events including coking, cracking and overheating of heat transfer surfaces may occur and lead to frequent shutdown of the facilities. Consequently, this could increase operating costs and reduce production revenue. Thus, it is desirable to know the effect that the bends exert on the flow in the straight part of the pipe. Apart from this, knowledge of the bend effects on the flows in the pipeline could also be important for the design of other pipelines for gas/liquid transport, e.g. offshore gas and oil pipelines. Quite a large number of studies have been found in the literature. The majority of them were for two-phase flow with small diameter pipes (i.d. ≤ 50 mm). However, studies with large diameter pipes (i.d. ≥ 100 mm), have increasingly been considered in recent years as problems related to large diameter vertical pipes are being encountered more and more often in industrial situations. This thesis studies the effect of 180o bends on the characteristics and development of gas-liquid two-phase flows in large diameter downward and upward pipes. The study particularly focuses on the influence of serpentine configuration on flow structure, cross-sectional void distribution and circumferential liquid film profiles and their development along the downward and upward sections. It was found that both the top and bottom bends have considerable impacts on flow behaviour, although to varying degrees. These impacts were highly dependent on the air and water flow rates. For sufficient flow rates, the bends were observed to create flow maldistribution in the adjacent straight section, due to the effects of centrifugal force. The air moved towards the inner zone of the bend and the water towards the outer zone, while a lesser quantity of water was identified on the other surfaces of the pipe. Investigation of the film thickness development in the downward and upward sections showed that, the liquid film behaviour close to the bends was significantly different from those located further away. This can be attributed to the centrifugal force of the bends. Examination of the power spectral density (PSD) along the downward and upward sections showed that, the shape of PSD located in the adjacent section to the bends, was substantially different from those located further away. Furthermore, several flow regime maps were generated which showed that, in addition to bubbly, intermittent and annular flows, unstable flows existed along the upward section, particularly for low gas and water flow rates. In this study it was found that, the lower bend was periodically blocked by the liquid and then blown through by the accumulated air. The data obtained from this study were compared with different theoretical correlations found in the existing literature. Some discrepancy between the results of the current study and those of previous published materials was noted. Updated correlations were presented which provided well results when they applied for the data obtained from the current study and previous studies.

The works aimed at the further development and validation of models for CFD codes. For this reason, the new thermal-hydraulic test facility TOPFLOW was erected and equipped with wire-mesh sensors with high spatial and time resolution. Vertical test sections with nominal diameters of DN50 and DN200 operating with air-water as well as steam-water two-phase flows provided results on the evaluation of flow patterns, on the be¬haviour of the interfacial area as well as on interfacial momentum and heat transfer. The validation of the CFD-code for complex geometries was carried out using 3D void fraction and velocity distributions obtained in an experiment with an asymmetric obstacle in the large DN200 test section. With respect to free surface flows, stratified co- and counter-current flows as well as slug flows were studied in two horizontal test channels made from acrylic glass. Post-test calculations of these experiments succeeded in predicting the slug formation process. Corresponding to the main goal of the project, the experimental data was used for the model development. For vertical flows, the emphasis was put on lateral bubble forces (e.g. lift force). Different constitutive laws were tested using a Multi Bubble Size Class Test Solver that has been developed for this purpose. Basing on the results a generalized inhomogeneous Multiple Size Group (MUSIG) Model has been proposed and implemented into the CFD code CFX (ANSYS). Validation calculations with the new code resulted in the conclusion that particularly the models for bubble coalescence and fragmentation need further optimisation. Studies of single effects, like the assessment of turbulent dissipation in a bubbly flow and the analysis of trajectories of single bubbles near the wall, supplied other important results of the project.

Ziel der Arbeiten war die Weiterentwicklung und Validierung von Modellen in CFD-Codes. Hierzu wurde am FZD die thermohydraulische Versuchsanlage TOPFLOW aufgebaut und mit räumlich und zeitlich hochauflösenden Gittersensoren ausgestattet. Vertikale Teststrecken mit Nenndurchmessern von DN50 bzw. DN200 für Luft/Wasser- sowie Dampf/Wasser-Strömungen lieferten Ergebnisse zur Entwicklung von Strömungsformen, zum Verhalten der Zwischenphasengrenzfläche sowie zum Wärme- und Impulsaustausch zwischen den Phasen. Die Validierung des CFD-Codes in komplexen Geometrien erfolgte anhand von 3D Gasgehalts- und Geschwindigkeitsfeldern, die bei Umströmung eines asymmetrischen Hindernisses auftreten, das in der Teststrecke DN200 eingebaut war. Im Hinblick auf Strömungen mit freier Oberfläche untersuchte das FZD in zwei horizontalen Acrylglas-Kanälen geschichtete Zweiphasenströmungen im Gleich- bzw. Gegenstrom sowie Schwallströmungen. Bei den Nachrechnungen dieser Versuche gelang die Simulation der Schwallentstehung. Entsprechend des Projektziels wurden die experimentellen Ergebnisse zur Modellentwicklung genutzt. Bei vertikalen Strömungen stand die Wirkung der lateralen Blasenkräfte (z.B. Liftkraft) im Vordergrund. Zum Test unterschiedlicher Modellansätze wurde hierzu ein Mehrblasenklassen-Testsolver entwickelt und genutzt. Darauf aufbauend wurde ein neues Konzept für ein Mehrblasenklassenmodell, das Inhomogene MUSIG Modell erarbeitet und in den kommerziellen CFD Code CFX (ANSYS) implementiert. Bei Validierungsrechnungen zeigte sich, dass vor allem die Blasenkoaleszenz- und -zerfallsmodelle weiter optimiert werden müssen. Untersuchungen zu Einzeleffekten, wie z.B. die Abschätzung von Turbulenzkoeffizienten und die Analyse der Trajektoren von Einzelblasen in unmittelbarer Wandnähe, lieferten weitere wichtige Ergebnisse des Projekts.

<div>With the rapid development of the advanced two-phase flow experimental technologies, more experimental databases with extended measurement ranges have been established to support the two-phase flow model development. The advantage of the Two Fluid model in modeling the complex two-phase flow phenomena over the mixture models stands out. One key aspect in the Two Fluid model development is the accurate modeling of the interfacial area between phases, which is strongly related to the interfacial mass, momentum, and energy transfer. As a closure relation of interfacial area concentration (interfacial area per unit volume) for the Two Fluid model, the Interfacial Area Transport Equation (IATE) provides dynamic predictions on the interfacial area change. It substantially solves the shortcoming of using flow-regime-dependent empirical correlations that can introduce numerical discontinuities between flow regimes. </div><div><br></div><div>The IATE has been extensively developed over the past twenty-five years. Many studies targeted on improving its prediction capability by developing bubble interaction source terms based on their experimental data. </div><div>The existing models are usually based on medium and large flow channels, yet the models may not be physically fit the small flow channels. The major reason is that the wall effect can have a larger influence on the two-phase flow in a small flow channel, as the surface area to volume ratio greatly increases. Therefore, the primary objectives of this study are to physically investigate the wall effect on two-phase flow and develop a generalized IATE by extending the application range of existing IATE from large and medium flow channels to small flow channel.</div><div><br></div><div>To achieve the objective, this study established a rigorous database of air-water two-phase flows in a small diameter pipe with its inner diameter of 12.7 mm, focusing on the bubbly-to-slug transition regime. The experimental analysis was performed on the pipe wall effect on the interfacial characteristics, based on the current experimental database and the existing experimental database collected on vertical pipes of different sizes. It is observed that 1) the pipe wall effect can alter the non-uniform radial two-phase distribution; 2) the bubbly-to-slug flow regime transition in a small diameter pipe happens in a smaller void fraction than in a large diameter pipe; 3) the bubble coalescence phenomenon can be more dominant for small pipe flow, and an intensive intergroup transfer can happen for the two-group interfacial area transport in two-phase flows. </div><div>As the interfacial area transport is directly related to the two-phase geometrical configuration, the two-phase geometrical parameters, void fraction and relative bubble size, are identified as the key parameters for modeling.</div><div><br></div><div>In the modeling of IATE source terms, the high geometrical scalability of the model is realized by properly including the wall effect into the modeling consideration. The following major improvements on the existing models are: 1) the inertia subrange assumption on the turbulent-driven interaction is properly improved; 2) the bubble-induced turbulent-driven interactions such as wake entrainment is revised by considering the wall effect on the wake region. In summary, models of bubble interaction due to random collision, wake entrainment, turbulent impact, and shearing-off are revised based on the existing studies on the IATE source terms development. The newly proposed interfacial area transport models are evaluated against an experimental database with 112 test conditions in total from a wide range of experimental pipe diameters from 12.7 mm to 304.8 mm. The new models can accurately capture the drastic intergroup transfer of void fraction and interfacial area concentration between two groups in transition flows. Overall, the relative error of void fraction and interfacial area concentration comparing with the experimental data are within ±15\% and ±10\%, respectively.</div>