Top Ten Limitations – Value Analysis

Automation engineers don’t get anywhere near the credit they deserve for increases in process performance. The value would be obvious if the automation system could be turned off but no one would allow this capture of the “before” and “after”. We need to do a better job of documenting where we are and could be. The best way Monsanto-Solutia found was a gap analysis. Much of what was done manually from cost sheets could be automated by configuration and historization. The following guide is a synthesis of what was done and could be done. Initially this process control improvement may need to be an extracurricular effort. Once you have evidence of a good “before” and “after” case, doors will open including managers and directors. I would enlist the participation of an open minded process engineer. Be prepared to deal with noise and the grab of benefits by operations, maintenance, and process design. To get credit you need to document performance immediately “before” and “after” the improvements. Look for quick fixes that just involve parameter, setpoint, and configuration changes. If you can use a plant spare to improve measurement or valve performance, go for it. For more information checkout the Control Talk column Show Me the Money – Part 1 and entries on this website, such as Interactive Opportunity Assessment – Introduction and Exceptional Opportunities in Process Control – Online Metrics 

Guide to Process Control Improvement 

(1)   Totalize all important utility and process stream flows.

(2)   Compute ratios of off-spec, utility, recycle, and raw material to product flows.

(3)   Find when batch actually reaches endpoint by noting when fed-batch, vent, and utility flows manipulated by process controllers nearly go to zero. Even better use a neural network or partial least squares estimator to predict batch end point corrected by lab analysis. The ultimate way uses at-line or online analyzers.

(4)   Find rise times for major changes in batch setpoint

(5)   Find weeks with best and worst ratio, total production, and earliest endpoint.

(6)   Compute average of ratios, production, and time to endpoint.

(7)   Multiply by dollars for each flow

(8)   Compute gap between best and average for each unit operation.

(9)   For best and worst weeks look for utility and process disturbances, raw material delivery, changes in raw material sources, changes in recycle, changes in setpoint, changes in configuration, manual actions, maintenance, tuning, holds, shutdown and startup time, changes in weather, changes in shift, and valves running near output limits.  

(10)                       Propose quick fixes for problems found.


Quick Fixes

(1) Tune digital positioners

(2) Tune loops for less oscillation, better disturbance rejection, and faster rise time

(3) Eliminate manual actions by smarter PID loops and automated startups

(4) Improve setpoints

(5) Add feedforward control to compensate for measured distrubances

(6) Add valve position control to push process

(7) Eliminate batch hold and wait times

(8) Use flexible batch time to take advantage of early endpoint

(9) Stop chasing noise

(10) Eliminate oscillations from limit cycles, interactions, and analyzers by use of an enhanced PID (PIDPlus)