Don’t Just Control The Process – Control The Risk
McEwan, M. – Perspective Engineering Ltd
(free)
The risk of a process moving outside quality limits is dependent on two factors: First, the observability of that process: how much relevant information is being gathered, how trustworthy is that data, and is that data being correctly interpreted and acted upon? If the data cannot be trusted, the process cannot be confidently controlled, so compromise has to be accepted. Second, controllability: wastewater treatment involves complex processes, multiple integrated assets, long time delays between interactions, and huge variability in inlet quality. A standard PLC-based control scheme cannot easily be re-tuned for each set of new conditions, so a ‘best fit’ approach is adopted, where control is never optimal but is ‘as good as possible’.
Model-based predictive control takes a different approach. The prediction of future operation (and, hence, future process quality) is made within ‘confidence limits’ that are derived from process data. If the data is statistically proven to be reliable, the calculated confidence limits are very narrow, enabling much tighter control to be exercised. This improvement in process capability can be translated into greater energy efficiency, while simultaneously minimising the risk of non-compliance. This can be fully –
and automatically – exploited if seasonal consenting is in place, or during TRIAD periods, or when consent limits are varied based on the dilution effect of receiving waters, in trials being run by the EA.
When observability is compromised – when signals are lost or sensors fouled – those confidence limits expand. To maintain the same lowered cost of operation would be to accept a higher risk of non-conformance. Alternatively, the risk of non-compliant operation can be minimised if a compromise is made on the cost of operation. The additional cost – energy or chemical – associated with loss of critical information can be calculated in real time as a ‘Lost Opportunity’ metric. Perceptive’s system now identifies and displays onto the control room SCADA the cost of sensor failure, enabling maintenance tasks to be prioritised based on validated intelligence.
