Advanced Process Monitoring for Wastewater Treatment Works
O’ Brien, M., McEwan, M. and Mazier, S.
Perceptive Engineering Ltd
(free)
Widespread use of online instrumentation in the wastewater industry in recent years has provided a large wealth of process information. For multivariable and large scale processes such as wastewater treatment works, discerning and highlighting the appropriate information is key. Advanced process monitoring aims to decrease the volume of process alarms provided to operators, whilst increasing the quality of information conveyed. This paper considers the application of process monitoring approaches, such as Principal Component Analysis, in the detection of abnormal process operation and the provision of appropriate operator alarms for the secondary treatment stage at a wastewater works.
Keywords
Process Monitoring, Principal Component Analysis, Intelligent Instrument Monitors
Introduction
With the increasingly common usage of online analytical instruments in wastewater treatment works there is a wealth of data available. Coupled with this increased availability of instrumentation is a need for reliable data screening and fault detection. The robustness of the data impacts directly upon any control decisions made or process information interpreted from the data. This is of particular importance in the wastewater industry, where the harsh operating environment takes its toll upon instruments – particularly those situated at the plant influent.
One of the key factors in decision-making, from plant design to process control, is the treatment plant performance and the reliability of the plant with respect to consent limits. Given the dynamic and variable nature of the wastewater treatment plant, plant performance also varies – diurnally, seasonally and according to operating mode. Standard univariate process monitoring approaches are not always suitable for monitoring of wastewater treatment processes with respect to water quality, because these variables are not typically normally distributed. Monitoring of a large dynamic process with many coupled variables, such as this, requires an advanced approach capable of considering the multivariable interactions in the process and utilising this knowledge to identify abnormal operating conditions.
The recent shift towards online instrumentation for water quality monitoring is welcome – providing more information on the dynamics of the treatment process than the traditional approach of sampling and laboratory analysis. Combining both online monitoring and the traditional sample-based approach can increase the reliability of online instruments – by allowing advanced monitoring systems to consider the accuracy of the online instrument. Calibration monitoring is possible, monitoring instrument measurement drift (based on comparison with laboratory samples) and providing a method by which preventative maintenance may be implemented.
