subscribe
 

Model-based monitoring and control enhances drying systems

1st April 2013


With energy prices continuing to rise, everyone in production is beginning to feel the pinch – and looking for ways to reduce operational costs whenever possible. Evidence from European studies suggests that up to 50 per cent energy savings are achievable through Best Practice initiatives in drying. One of these initiatives, model based engineering, is explored briefly in this article; the techniques described have been developed and commissioned on a variety of drying systems throughout Europe, producing valuable energy savings in starch, milk powder, and casein production.
The concept behind model-based systems is simple: a model is used to predict the impact that known disturbances will have on your operation in the near future. Using this predictive information and knowledge of operational constraints, the system is able to effectively co-ordinate the available actuators and take appropriate action to minimise the effect of such disturbances at the earliest opportunity. The techniques frequently use a data driven model of the process, empirically capturing the process behaviour, to be used at the core of a model predictive control (MPC) engine. The MPC is primarily used to reduce the variability in the drying process, delivering powder with tightly controlled moisture and bulk density. The model predictive controller is usually combined with an Optimisation engine that calculates a steady state optimum position for the dryer for whatever prevailing conditions exist; which may include changing air humidity, varying powder characteristics, as well as changing throughput and emission constraints.
These model-based control systems have become commonplace over the past five years, however they still present interesting control challenges due to the large amount of unmeasured process disturbances, unreliability of instrumentation and a greater operational flexibility required from each drying unit.
One new development is the combination of statistical monitoring techniques with the model-based engines to provide some notable advantages:
• Early detection of process faults, such as dryer and cyclone blockages.
• Information describing the current state of the dryer, eg fouling, powder deposition, sensor drift.
• An ability to accurately relate lab results to on-line process measurements – providing immediate feedback to the operator if the ‘Product Quality’ envelope is breached.
• Dynamic Alarm Limits that profile ‘good’ start-up – leading to a clean and extended production run.
These techniques offer non-invasive additions to the control system, yet dramatically aid process operability and quality control management using only existing process measurements. It seems a waste to set up and store vast amounts of process data only for historical use when it could be continuously and actively used to enhance operability of the process plant.
The integration of these technologies provides significant reduction in dryer blockages, increased throughput and yield. Examples have shown that product quality variability is typically reduced by 40–60 per cent, blockages reduced by 75 per cent and significant savings are available even on newly installed equipment. 

Juliette Cameron is with Perceptive Engineering Ltd, Lymm, Cheshire, UK. www.perceptive-engineering.co.uk





Subscribe

Subscribe



Newsbrief

FREE NEWSBRIEF SUBSCRIPTION

To receive the Scientist Live weekly email NewsBrief please enter your details below

Twitter Icon © Setform Limited
subscribe