Typical industrial process analysis techniques require an optical path to exist between the measurement sensor and the process to acquire data used to optimise and control an industrial process. Ultrasonic sensing is a well-established method to measure into optically opaque structures and highly focussed images can be generated using multiple element transducer arrays. In this Thesis, such arrays are explored as a real-time imaging tool for industrial process analysis. A novel methodology is proposed to characterise the variation between consecutive ultrasonic data sets deriving from the ultrasonic hardware. The pulse-echo response corresponding to a planar back wall acoustic interface is used to infer the bandwidth, pulse length and sensitivity of each array element. This led to the development of a calibration methodology to enhance the accuracy of experimentally generated ultrasonic images. An algorithm enabling non-invasive through-steel imaging of an industrial process is demonstrated using a simulated data set. Using principal component analysis, signals corresponding to reverberations in the steel vessel wall are identified and deselected from the ultrasonic data set prior to image construction. This facilitates the quantification of process information from the image. An image processing and object tracking algorithm are presented to quantify the bubble size distribution (BSD) and bubble velocity from ultrasonic images. When tested under controlled dynamic conditions, the mean value of the BSD was predicted within 50% at 100 mms-1 and the velocity could be predicted within 30% at 100 mms-1. However, these algorithms were sensitive to the quality of the input image to represent the true bubble shape. The consolidation of these techniques demonstrates successful application of ultrasonic phased array imaging, both invasively and noninvasively, to a dynamic process stream. Key to industrial uptake of the technology are data throughput and processing, which currently limit its applicability to real-time process analysis, and low sensitivity for some non-invasive applications.
|Date of Award||17 May 2019|
- University Of Strathclyde
|Sponsors||EPSRC (Engineering and Physical Sciences Research Council)|
|Supervisor||Anthony Gachagan (Supervisor) & Alison Nordon (Supervisor)|