Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.577679
Title: Artificial intelligence control of pumping in sewer networks
Author: Ostojin, Sonja
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2012
Availability of Full Text:
Access through EThOS:
Abstract:
The control strategy for sewer pumping stations currently used by Anglian Water, and most other UK water service providers, is classical on/off switching. This is based simply on the fluid level in the inlet wet well. Such local control can lead to poor performance across a variety of indicators, including energy costs, hydraulic performance and efficiency. Despite the listed issues, the Anglian Water sewerage system is operating at a satisfactory level in terms of conveying flow to treatment. In clean water systems, artificial intelligence and other techniques have been used to optimise pump scheduling and research is now needed to provide similar efficiency gains for sewerage systems. Artificial intelligence is widely used for different control and automation applications in many areas. There is a limited use of optimal control in clean water systems but little or no research has been done to apply these techniques to waste water. To the best of author's knowledge, this research is the first time that within the water sector that Fuzzy logic (FL) has been applied specifically to the control of pumping in sewer networks. This thesis presents details of a FL system developed in dry weather flow (DWF) for control of sewer pumping stations for energy and costs savings. This research was a collaborative project between Anglian Water and the University of Sheffield. The FL controller outperformed the Base case representing current practice in terms of increased energy and costs savings and a lower number of pump s'Aitches. The functionalities of the FL controller were achieved by: • A Genetic Algorithm was used to tune the FL controller by determining the base lengths of the Membership Functions and the locations of the peaks (vertices of the membership functions' two non-base edges). • The Rule Base impact on level control was investigated and representative Rule base designs were chosen for each of the electrical tariffs. • The FIS controller was extended to include pump efficiency. Pump efficiency was then taken into consideration by running close to the optimal pump operating point, not taking into account long term pump degradation. • An ANN prediction model was developed for DWF inflow prediction which was trained, tested and validated on flow data obtained from survey. The ANN model has just two inputs (population and trade flow) and thus a flow survey is no longer required in order to tune the FL controller. • The FL controller represents a general solution which can be successfully transferred to another pumping station with different characteristics: wet well dimensions, min/max levels and different pumps. Customarisation is needed in terms of physical characteristics of pump station. Simulation results were confinned by the live trial. • The FL system was tested for its robustness using a wet weather event flow profiles. It continued to demonstrate an ability to supply energy cost savings under wet weather conditions albeit not as optimally as for DWF. Wet weather flow (WWF) was out of a scope of this research. FL control system would switch to fail-safe operating mode during WWF event (on/off control timer controlled mode with the pump switch on point sets to 90% of maximal level in the wet well). In summary, Anglian Water currently spends £13 million annually on energy for wastewater pumping. Around 10% of pumping stations within Anglian Water could eventually benefit from intelligent control. Energy savings from intelligent control are expected to deliver 5% or better, based on the results of this PhD research project. This would reduce the annual energy bill of the company by a third of a million pounds.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.577679  DOI: Not available
Share: