The explosive growth of digital data collection and processing demands a new approach to the historical engineering methods of data correlation and model creation. A new prediction methodology based on high dimensional data has been developed. Since most high dimensional data resides on a low dimensional manifold, the new prediction methodology is one of dimensional reduction with embedding into a diffusion space that allows optimal distribution along the manifold. The resulting data manifold space is then used to produce a probability density function which uses spatial weighting to influence predictions i.e. data nearer the query have greater importance than data further away. The methodology also allows data of differing phenomenology e.g. color, shape, temperature, etc to be handled by regression or clustering classification. The new methodology is first developed, validated, then applied to common engineering situations, such as critical heat flux prediction and shuttle pitch angle determination. A number of illustrative examples are given with a significant focus placed on the objective identification of two-phase flow regimes. It is shown that the new methodology is robust through accurate predictions with even a small number of data points in the diffusion space as well as flexible in the ability to handle a wide range of engineering problems.

<p dir="ltr">The accurate identification of flow regimes in multiphase flow systems is of paramount importance in many engineering applications. This thesis explores the significance of flow regime identification using neural networks, specifically employing a self-organizing map (SOM) algorithm. The focus of this research is on the determination of bubble void fraction probability density function (PDF) using local conductivity probe measurements. The thesis begins by providing an overview of the importance of flow regime identification in understanding and predicting the behavior of multiphase flows. Various flow regimes such as bubbly flow, slug flow, annular flow, and others, are discussed highlighting their distinct characteristics and implications for system performance. The self-organizing map is introduced as a powerful neural network technique capable of identifying and classifying different flow regimes based on input parameters obtained from local conductivity probe measurements. The SOM algorithm is explained in detail, emphasizing its ability to learn and adapt to complex patterns in the data. To validate the effectiveness of the proposed approach, experimental measurements of local conductivity probe signals were conducted in a multiphase flow system. The obtained data was used to train and optimize a self-organizing map for flow regime identification. The bubble void fraction probability density function was calculated based on the local time-averaged void fraction measurements from the droplet-capable conductivity probe (DCCP-4). The analysis of the PDF provides valuable insights into the distribution and characteristics of bubbles within the multiphase flow system. These insights can enhance the understanding of bubble behavior, droplet behavior, interfacial phenomena and overall system performance. The thesis concludes with the classification results along with an error analysis conducted to highlight potential discrepancies in the tested results. Additionally, future research directions and potential improvements in the flow regime identification methodology are outlined.</p>

Flow regime identification in industrial systems that rely on complex multi-phase flows is crucial for their safety, control, diagnostics, and operation. The objective of this investigation was to develop and demonstrate objective and minimally-intrusive flow regime classification methods for gas/water/paper pulp three-phase slurries, based on artificial neural network-assisted recognition of patterns in the statistical characteristics of pressure fluctuations. Experiments were performed in an instrumented three-phase bubble column featuring vertical, upward flow. The hydrodynamics of low consistency (LC) gas-liquid-fiber mixtures, over a wide range of superficial phase velocities, were investigated. Flow regimes were identified, gas holdup (void fraction) was measured, and near-wall pressure fluctuations were recorded using high-sensitivity pressure sensors. Artificial neural networks of various configurations were designed, trained and tested for the classification of flow regimes based on the recorded pressure fluctuation statistics. The feasibility of flow regime identification based on statistical properties of signals recorded by a single sensor was thereby demonstrated. The transportability of the developed method, whereby an artificial neural network trained and tested with a set of data is manipulated and used for the characterization of an unseen and different but plausibly similar data set, was also examined. An artificial neural network-based method was developed that used the power spectral characteristics of the normal pressure fluctuations as input, and its transportability between separate but in principle similar sensors was successfully demonstrated. An artificial neural network-based method was furthermore developed that enhances the transportability of the aforementioned artificial neural networks that were trained for flow pattern recognition. While a redundant system with multiple sensors is an obvious target application, such robustness of algorithms that provides transportability will also contribute to performance with a single sensor, shielding effects of calibration changes or sensor replacements.

New decline curve models have been developed to overcome the boundary-dominated flow assumption of the basic Arps’ models, which restricts their application in ultra-low permeability reservoirs exhibiting long-duration transient flow regimes. However, these new decline curve analysis (DCA) methods are still based only on production rate data, relying on the assumption of stable flowing pressure. Since this stabilized state is not reached rapidly in most cases, the applicability of these methods and the reliability of their solutions may be compromised. In addition, production performance predictions cannot be disassociated from the existing operation constraints under which production history was developed. On the other hand, DCA is often carried out without a proper identification of flow regimes. The arbitrary application of DCA models regardless of existing flow regimes may produce unrealistic production forecasts, because these models have been designed assuming specific flow regimes. The main purpose of this study was to evaluate the possible benefits provided by including flowing pressures in production decline analysis. As a result, it have been demonstrated that decline curve analysis based on pressure-normalized rates can be used as a reliable production forecasting technique suited to interpret unconventional wells in specific situations such as unstable operating conditions, limited availability of production data (short production history) and high-pressure, rate-restricted wells. In addition, pressure-normalized DCA techniques proved to have the special ability of dissociating the estimation of future production performance from the existing operation constraints under which production history was developed. On the other hand, it was also observed than more consistent and representative flow regime interpretations may be obtained as diagnostic plots are improved by including MBT, pseudovariables (for gas wells) and pressure-normalized rates. This means that misinterpretations may occur if diagnostic plots are not applied correctly. In general, an improved forecasting ability implies greater accuracy in the production performance forecasts and more reliable reserve estimations. The petroleum industry may become more confident in reserves estimates, which are the basis for the design of development plans, investment decisions, and valuation of companies’ assets.