What is control benchmarking?
When applied to a process, controls become part of a changing environment. Over a period of time, processes are typically subject to:
- Product changes (new operating points etc.)
- Raw material source changes
- Wear and tear of actuators (valves, drives etc.)
- Degradation of measurement sensors
- Plant item (non-actuator) wear and tear
These changes often degrade control and plant financial performance. Ideally, checks by instrument/control personnel identify and address such effects. However, if the root cause is not clearly established, control re-tuning activity can make matters worse. In practice, diagnosis tends to depend on human experience. This is against a background of increasing difficulty in recruiting experienced control personnel.
To remove this dependency, it is necessary to utilise process diagnostic tools. Typically these techniques either require 'expert' users or are relatively crude and ineffective.
The challenge is to compare actual loop tuning against the optimum possible. This is defined as control benchmarking.
- Generalised Minimum Variance
- Restricted Structure Linear Quadratic Gaussian
- Linear Quadratic Gaussian Predictive Control
- Selection of benchmark weighting (PDF)
- Controller benchmarking: from single loops to plant-wide economic assessment (PDF)
- New developments in performance assessment and benchmarking (PDF).
- Structure and function of MIMO benchmarking software (PDF)
- Performance assessment and benchmarking in applications (PDF)
- New ideas in performance and benchmarking of nonlinear systems (PDF)
- New approaches to condition monitoring (PDF)