Abstract
Maintaining consistent plant operation is dependent on timely detection of process issues, rapidly diagnosing their cause, and accurately determining the nature and resolution of the problem. Many of the issues that occur on wastewater plants require a human intervention to intuitively determine both the cause and the relative importance of each problem. A real-time maintenance diagnosis system is able to monitor multiple variables as well as the correlation between different process measurements. Current decision support systems (and operator responses) work backwards from the event, using available data to try to identify cause. Predictive diagnostics, based on multivariable monitoring, provide early detection of unfolding events, enabling intervention to be planned or scheduled, rather than be performed after the fact. This ‘Operator 2.0’ system guides the user to make the appropriate corrective and pre-emptive actions, based on both captured process knowledge and industry best practise.
Keywords Predictive, Advanced, Multivariable, Preventative, Diagnostics, Maintenance.
Introduction
Event Monitoring is an umbrella term for any technology which utilises real-time information to detect events as they occur, or predicts their likely occurrence. Event monitoring is used in monitoring processes and sub-processes, predicting end results (or final quality), optimising systems in real-time, and providing users and operators with a deeper understanding of their process.
It is important not only to measure and record process information, but to fully exploit the available information – interpreting what can be complex and widespread process data and providing a more holistic analysis of the process. In other words, transform data into information, knowledge and action.
During the last AMP cycle, the wastewater industry saw a rapid increase in instrumentation within key processes, in part due to increased reliability of process instruments. Coupled with an industry-wide focus on collecting and storing key data, large investment in telemetry, and increasingly stringent regulatory and energy demands, the advent of data-based event monitoring in the wastewater industry has arrived.
Such monitoring must, of course, focus on consent requirements as its primary driver – with energy demands, operational cost, reduced process down-time, and increased reliability and stability of operation as other key factors. The current limitations on event monitoring for wastewater processes are the reliability of instrumentation and the presence of complex, interactive processes. Faced with a hostile operational environment, degradation and fouling of instruments is almost inevitable. Any robust monitoring system must provide a reliable means of monitoring, even in the face of reduced ‘visibility’ due to sensor degradation.
Wastewater treatment relies on some of the most complex processes used in any industry, due to the presence of physical, chemical and biological activity all within the same plant, as well as multiple parallel and interactive streams and highly variable incoming loads. Today’s monitoring systems must consider the unique aspects of wastewater treatment, and exploit those in providing an appropriate and rigorous approach to supporting operation. These should include:
- Diurnal pattern recognition: The vast majority of plants experience an identifiably diurnal variation in the characteristics of the incoming wastewater. This is a variable but typically predictable mode of operation. Detecting excursions from this ‘norm’ can provide advanced warning of potential upsets, such as septicity, solids washout or toxic events.
- Monitoring and Classification: Identifying abnormal operation is paramount, but to assist site operators with resolving problems and mitigating risks, being able to classify multiple ‘symptoms’ into one event is helpful. The development of ‘intelligent’ alarms – which are classified not according to type (e.g. pre-assigned priorities) but according to events – will enable all available information to be more fully exploited.
- Event Detection and Root-Cause Analysis: Highlighting events to operators is the first step in a monitoring scheme. For the information to become immediately usable, a ‘root cause’ analysis must provide the operator with an indication of the most likely root cause(s) of the process event, to enable early action to prevent or mitigate.
