The purpose of this project is to determine an optimum Electrical Capacitance Tomography (ECT) sensor design for oil-fraction estimation using a trained Artificial Neural Network (ANN) system. This involves investigation on the optimum ECT parameters that suit the trained ANN for the mentioned purpose. An ECT can be used to obtain information on the distribution of flow materials within a pipe cross-section. Five types of flow patterns have been investigated. They are Stratified Flow, Annular Flow, Bubble, Full Pipe and Empty Pipe. These five types of flow geometries were constructed and the ECT data corresponding to the flow patterns were simulated. This project is important for the chemical industries regarding safety, environmental protection, energy conservation, and quality assurance. The ECT measurement technique is essential for the design and control of chemical plants and transportation systems. The common used for the ECT measurement technique in for the petroleum industry or petroleum mining. The ECT measurement technique is normally used to determine the components concentration of two-phase flows in industrial process. The steady-state capacitance transducer systems work on the principle of measuring the difference in the mean capacitance value of sensing electrodes mounted on a section of a flow pipeline. For this project, the two flow components, which are gas and oil, have been different permittivities. So, a direct measurement of component volumetric concentration can be made. These are the actual capacitance measurements. The Artificial Neural Network (ANN), which has been intelligently trained according to certain parameters value, was used to estimate the oil fraction based on the capacitance measurements. The last part of this project is to compare the two values of estimated and calculated mean error for each type of the flow pattern. Finally, graphs are plotted for analysis to determine the optimum ECT parameters.
Optimum ect sensor for oil-fraction_C'hng Siew Peng_E3_2005_NI