Mass flow measurement of multi-phase mixtures by means of tomographic techniques
Doctoral Thesis
2002
Permanent link to this Item
Authors
Supervisors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
University of Cape Town
Department
License
Series
Abstract
This thesis investigates the use of a dual-plane impedance tomography system to calculate the individual mass flow rates of the components in an air-gravel-seawater mixture. The long-term goal of this research is to develop a multi-phase flowmeter for the on-line monitoring of an airlift used in an offshore mining application. This requires the measurement of both the individual component volume fractions and their velocities. Tomography provides a convenient non-intrusive technique to obtain this information. Capacitance tomography is used to reconstruct the dielectric distribution of the material within a pipeline. It is based on the concept that the capacitance of a pair of electrodes depends on the dielectric distribution of the material between the electrodes. By mounting a number of electrodes around the periphery of the pipeline, and measuring the capacitances of the different electrode combinations, it is possible to reconstruct the distribution of the phases within the pipeline, provided the phases have different dielectric constants. Resistance tomography is used to reconstruct the resistivity distribution within the cross-section of the pipeline and operates in a similar way to capacitance tomography. Impedance tomography can be described as a dual-modal approach since both the capacitance and conductance of the different electrode combinations are measured to reconstruct the omplex impedance of the material distribution. Previous research has shown that impedance tomography can be used to reconstruct a three-phase air-gravelwater mixture [3,4]. In addition, it has been shown that neural networks can be used to perform this reconstruction task [3,4]. In particular, a single-layer feed-forward neural network with a 1-of-C output encoding can be trained to perform a three-phase image reconstruction. Further, a double-layer feed-forward neural network can be trained to predict the volume fractions of the three phases within the flow directly, based on the capacitance and conductance readings obtained from the data acquisition system. However, these tests were only for static configurations. This thesis will readdress this problem from the dynamic viewpoint. In addition, the individual component velocities will be calculated using the cross-correlation of the volume fraction predictions from two impedance tomography systems spaced a certain distance apart.
Description
Includes bibliographical references.
Keywords
Reference:
Teague, G. 2002. Mass flow measurement of multi-phase mixtures by means of tomographic techniques. University of Cape Town.