Applications of image processing for performance assessment of wastewater treatment plants
MetadataShow full item record
This item's downloads: 150 (view details)
This thesis is concerned with the development of methodologies for automatically assessing fluid properties in wastewater treatment plants, and to monitor plant performance and efficiency. Reliable performance metrics are important for optimising the operation of treatment plants. While sensor technologies are not new to the wastewater treatment field, the application of imaging and image processing systems has seen little development in recent years. Currently, there are a multitude of sensor technologies to monitor a variety of performance parameters. However, research has found that, in practice, the results from these sensors are rarely relied upon due to inconsistent calibration schedules, compounded by a lack of full-time maintenance staff. Instead, these sensors are frequently bypassed in favour of more subjective manual estimates. In this thesis, the problems with current monitoring practices for wastewater treatment plants are considered. Two areas of interest are highlighted; (i) sludge monitoring, using Sludge Volume Index as a performance metric and (ii) effluent monitoring, using turbidity as a quality metric. A review of the guideline sludge monitoring procedures is presented as well as a discussion of alternative on-site monitoring practices. This review highlights the need for an automated system for performing sludge monitoring, as per the guideline procedures. Subsequently, an image processing system for settled sludge volume measurement is proposed and tested. Effluent turbidity is primarily affected by a high number of colloidal particles in the particle size distribution. Current effluent turbidity monitoring practices include submerged turbidimeters, that require regular calibration. In practice, it was found that subjective manual estimation of turbidity was often being conducted ”by eye”. A study was devised to characterise the accuracy of the subjective estimation. Firstly, an imaging methodology was designed to capture effluent images illustrating light decay as a function of increasing fluid depth. These images were then presented to persons with no knowledge of wastewater treatment monitor and subjectively rated on turbidity. The results of the subjective test were then compared to established laboratory-based turbidity measurement, and a clear correlation was found between the two. Subsequently, an image processing system was designed to replace the observer and to objectively characterise the light decay as a function of fluid depth. Once again, the results from this image processing system were compared to the established laboratory-based measurements and an improved correlation over the subjective comparison was found. Finally, the implications of the deployment of a combined monitoring system is discussed, along with the benefits to current monitoring practices.