Hassan Shaban will present his PhD seminar talk this Friday on “Application of Machine Learning Techniques to Measurements in Gas-liquid Flows”. The abstract of the talk is below. Hassan is working on his PhD under the supervision of Dr. Stavros Tavoularis.
Date: Friday November 14th
Room: CBY B205.
Application of Machine Learning Techniques to Measurements in Gas-liquid Flows
Gas-liquid flows are found in the environment, as for example in clouds and breaking water waves, as well as in many industrial settings, as for example in heat exchangers containing boiling liquids and in petroleum pipelines in which oil flows mixed with gases. In this seminar, novel experimental methods will be presented for the identification of the flow regime and measurement of the flow rates of both phases in gas-liquid pipe flow using simple instrumentation. The differential pressure was measured in vertical upward air-water pipe flow, and the probability density function (PDF) of normalized differential pressure, was found to be indicative of the flow regime. The Elastic Maps algorithm was used to project this PDF onto a two-dimensional map, which when appropriately calibrated, allowed a user to identify the flow regime from the co-ordinates of the projected point. Also, the flow rates of the two phases in vertical upward air-water flow were found to have consistent effects on the PDF and power spectral density of normalized differential pressure. Artificial neural networks were used to correlate these features extracted from the differential pressure signal to the flow rates of both phases. Compared to other methods of flow regime identification and two-phase flow rate measurement, the developed methods had the advantages of being automated, non-intrusive and economical, such that their use was feasible in industrial as well as laboratory settings. To conclude the talk, other examples of the integration of machine learning with fluid flow instrumentation, as well as a practical application of similar methods in a header-feeder system will be briefly described.