Mass Analytics Improve Process Control Performance

  • May 27, 2011
  • Feature
May 2011
By Bill Lydon, Editor
ExperTune User Conference - May 10-12, 2011, Milwaukee, Wisconsin
The PlantTriage “true believers” gathered at the ExperTune User Conference from around the world to improve their skills and share their project successes. The conference focused on ways to improve process control performance with training, access to industry experts, and case studies presented by customers.
Founded by John Gerry, ExperTune has been providing tools for improving process control performance since 1986. Gerry is multiple chapter author for the "Instrumentation Engineer's Handbook, 4th Edition." Gerry is also the recipient of ISA's 2003 Douglas H. Annin Award, recognizing outstanding achievement in the design or development of components of an automatic control system. 
Dinner at the Harley-Davidson Museum
Different Perspective
The applications and success stories about using the ExperTune PlantTriage® Control Performance Monitor software was the major focal point at the conference. 
I reviewed PlantTriage when it first was introduced in 2003 and the product has improved significantly to be a highly refined software suite designed to improve the operations of any process plant.  PlantTriage takes advantage of low cost high performance computing to perform a high volume analysis of all the process control loops in a plant. It is a different approach to analyzing process operations by correlating all the process loops in a plant to identify relationships between control loops. 
It took me awhile to shift my thinking and understand how this software is different. Let me emphasize that PlantTriage performs a massive mathematical cross correlation analysis of ALL process loops in the plant, looking for patterns of data that correlate.   This is done on a continuous basis.
Users Presented Applications and Results
This is a bit counter intuitive. Why would you look for correlations between loops that have seemingly no process relationship to each other?  After presentations from numerous users at the conference, I began to understand that loops can be related in ways that are not always obvious. Another factor is that process plants are not static but dynamic, and as process plants change and evolve, control loop relationships may change. The changes may be subtle and have unintended consequences.
For example, if you have a quality problem you can use PlantTriage to compare the pattern behavior of the quality loop abnormality to all other loops. PlantTriage will correlate the behavior of the quality loop to all other loops seeking to find a data relationship to this pattern.    PlantTriage does in a modern way what engineers did in a limited way in the past by comparing strip chart recorders, data logs, or graphs to visually inspect data looking of correlations or patterns to problem loops.  In addition, the software provides tools including loop tuning, valve diagnosis, control analysis, indentifying over-control, and interaction analysis.
Highly Visual
Engineers by nature are visual and the PlantTriage product provides powerful visual tools that provide an efficient interface.
The Interaction Hot Spots screen illustrates interactions across an entire plant facility using color coding.  Engineers can quickly see previously unknown relationships between loops.  Strong interactions are shown in red, while green indicates lack of interaction.  This information can be useful for process understanding, design of control strategy, and to eliminate possible causes of problems.
Interaction Hot Spots Screen
Using the Process Interaction Map, engineers can visually determine the root cause of upsets using metrics they choose including quality, process, or cost variables. Strong colors indicate strong interactions – the x axis indicates time shifts.  PlantTriage users can pinpoint the source of upsets using this tool.  Presentations by users during the conference illustrated that quite often the root cause of upsets are found in other parts of a plant such as utilities or subsystems.
Process Interaction Map Screen
User Experience
The ExperTune user conference featured presentations from a number of people that reported serious improvements and economic benefits by applying PlantTriage in process plants. These users were enthusiastic about the results they achieved and prospects for further improvements of plant processes. Application presentations illustrating increased efficiency included:
  • Alkylation Plant Energy Quality & Yield Improvements
  • Chemical Plant Fuel Gas System Flare Minimization
  • Paper Plant Basis Weight Problem Resolved
  • Power Plant Operations Improvements
  • Chemical Plant Distillation Control Improvements
  • Ammonia Plant Performance Improvements
  • Iron Ore Concentration Plants Process Improvements
Companies represented included Advanced Process Solutions & Technology (Korea), Invensys Operations Management, Control Specialists Ltd, (United Kingdom), Lyondell Chemical Company, and Accenture.
Thoughts & Observations
PlantTriage is a high level analytical tool made possible with low cost high performance computing now available to improve process control operations.
Users commented that they found numerous plant problems using the software that were not previously obvious.
Another comment was that using the software to do Advanced Process Control (APC) and Model based control yielded significantly better results.
After listening to the user case studies and thinking about my experience in plants, it occurred to me that management and many plant operations people think of a process plant as a static, predictable process. This is not an accurate way of thinking about a plant since they change over time. 
Business people always use cost accounting audits and controls to properly manage a business. I think a key function of PlantTriage is analogous to this, plus it gives engineers powerful tools to do root cause analysis and fix problems. For more information:

Did you enjoy this great article?

Check out our free e-newsletters to read more great articles..