A Feasible Data-Driven Mining System to Optimize Wastewater Treatment Process Design and Operation

Abstract

Achieving low costs and high efficiency in wastewater treatment plants (WWTPs) is a common challenge in developing countries, although many optimizing tools on process design and operation have been well established. A data-driven optimal strategy without the prerequisite of expensive instruments and skilled engineers is thus attractive in practice. In this study, a data mining system was implemented to optimize the process design and operation in WWTPs in China, following an integral procedure including data collection and cleaning, data warehouse, data mining, and web user interface. A data warehouse was demonstrated and analyzed using one-year process data in 30 WWTPs in China. Six sludge removal loading rates on water quality indices, such as chemical oxygen demand (COD), total nitrogen (TN), and total phosphorous (TP), were calculated as derived parameters and organized into fact sheets. A searching algorithm was programmed to find out the five records most similar to the target scenario. A web interface was developed for users to input scenarios, view outputs, and update the database. Two case WWTPs were investigated to verify the data mining system. The results indicated that effluent quality of Case-1 WWTP was improved to meet the discharging criteria through optimal operations, and the process design of Case-2 WWTP could be refined in a feedback loop. A discussion on the gaps, potential, and challenges of data mining in practice was provided. The data mining system in this study is a good candidate for engineers to understand and control their processes in WWTPs.