Multiphase flow is defined as the simultaneous flow of different phases, such as liquids, gases or solids, and occurs in a variety of industrial and environmental processes. This type of flow poses special challenges for the operators and engineers due to the complex behavior of the phases, which is still subject to intensive research efforts. This applies in particular to gas-liquid pipe flows with its different flow patterns, i.e., categories of spatio-temporal phase distributions. The formation of these flow patterns depend on the fluid properties, the geometry of the pipes and the operating conditions. Several flow patterns with different properties have been identified in horizontal pipes, namely stratified flow, wavy flow, plug flow, slug flow, annular flow, bubbly flow, and mist flow. Due to their intermittent sequence of liquid blocks and large gas bubbles, the slug and plug flow pattern cause various disadvantages for the operating facilities, including higher measurement errors, larger pressure drops, undesired fluctuations in pressure and flow rates, as well as induced vibrations and the associated fatigue of the piping. These flows are encountered especially in the deep sea oil and gas production with its long transportation pipelines. Because of this, it is important to take these intermittent flow patterns into account for the design of multiphase transportation pipeline systems. Therefore, predictive models, numerical simulations, and empirical correlations have widely been used as a starting point for engineering designs. However, due to the complexity of these flow patterns, it is still subject to intensive research to find models, which reliably and accurately predict the flow patterns and their temporal and spatial characteristics for given flow rates, fluid properties, pipe diameters, and inclinations. For the development, calibration and validation of such predictive models, correlations and simulations, methods that provide precise descriptions of these flows are of special interest. This includes the complex intermittent flow patterns, such as slug flow, but also the separated flow patterns, such as stratified or wavy flow, since they often constitute an initial state in the development of the other, more complex flow patterns. In this thesis, quantitative methods and novel approaches are presented and discussed, which enable a precise temporal and spatial characterization of separated and intermittent flow patterns from spatio-temporal flow data, such as raw video data or snapshot sequences of phase fraction fields. The focus for these investigations is on the slug flow pattern, since it is similar but more complex compared to the plug flow pattern and also contains stratified and wavy elements. At first, a novel approach for the quantification of horizontal slug flow with snapshot Proper Orthogonal Decomposition (snapshot POD) is investigated in this thesis. This includes the aspect of the representation of the slugs in the POD -modes, the derivation of the spatial and temporal characteristics of slug flow from the POD-modes and the data requirements for this application. It was found that the slugs are represented in the dominant mode pair of snapshot POD analysis for the considered data sets. This includes a description of the gas-liquid interface dynamics in the temporal coefficients as well as the representation of the averaged slug in the spatial modes. It is shown that the conventional length and time scale for a characterization of the considered slug flows can be derived from the temporal and spatial information of the dominant mode pair. Hence, the snapshot POD allows a detailed temporal and spatial characterization of the slugs. Moreover, it was found that, the length of the observed pipe segment (region of interest) affects the accuracy of the derived temporal and spatial characterization. In the course of these investigations, an approximation of the liquid level time series as well as time scales of slug flow have been derived from video recordings. In a second step, this video-based temporal quantification of horizontal slug flow is assessed in a further examination, where parameters from an established tomography measurement technique are used as a reference. It was found that the approximation of liquid level time series from the video data shows a systematic overestimation compared to the tomography measurements. In contrast to this, the derived temporal characteristics of the liquid slugs from this approximation provide a reasonable and consistent temporal characterization of slug flow. In this context, a fast and reliable method for the approximation of the liquid level time series from video data can serve as a useful tool in industrial and academic operations, where a temporal quantification of separated and intermittent flow patterns, such as wavy or slug flow, is needed. In a third step, the accurate and efficient extraction of the liquid level time series from video observations of separated and intermittent flow patterns with a novel deep learning based image processing technique is investigated. It is shown that the presented method accurately extracts the liquid level time series from the considered video data. It can handle different types of data, even unseen data sets. Furthermore, it can overcome various noise effects, which are generally included in such image or video data. Once, the net is successfully trained, it predicts highly accurate segmentation maps in very short time. Hence, the liquid level time series can be extracted accurately and efficiently from the raw video data.
Quantitative methods for the characterization of separated and intermittent multiphase flow patterns
Quantitative Verfahren zur Charakterisierung von separierten und intermittierenden Mehrphasenströmungen
2022
Sonstige
Elektronische Ressource
Englisch
DDC: | 629 |
Rotary sleeve separated intermittent working inflating pump
Europäisches Patentamt | 2015
|Characterization of confidence in multiphase flow predictions
Tema Archiv | 2012
|Tema Archiv | 2012
|Characteristics of separated flow airfoil analysis methods
TIBKAT | 1983
|