Operator Performance Research Put into Practice

This week, the Center of Operator Performance (COP) is meeting here in Austin, Texas. I have the opportunity to listen in, and share a few observations with you. You can follow these on my Twitter posts tagged #COPmtg.

For those not familiar with the work of this consortium of process manufacturers, automation suppliers, engineering organizations, and academia, their website summarizes the mission:

…addressing human capabilities and limitations with research, collaboration, and human factors engineering. Our mission to raise the performance level of our operators and improve Health, Safety, and Environmental effectiveness is accomplished through:

  • Openly sharing knowledge and ideas
  • Putting collaboration ahead of competition
  • Including vendors in research decisions
  • Teaming with leading human factors researchers and universities

Current members in the consortium include Wright State University, Beville Engineering, BP, Chevron, Flint Hills Resources, Marathon Petroleum Company, Suncor Energy, ABB, and Emerson Process Management.

Some of the current research projects underway include:

  • Human Factors Performance Metrics
  • Knowledge Management – Automated Procedures
  • Data Mining of Near Miss Incidents
  • Mental Models of Control Operators
  • Perceptual Cues Used by Expert Control Operators
  • Follow on to Event Prediction & Mitigation

You can see some of the pending and completed research on this page as well.

At the recent Emerson Exchange, Emerson’s Cindy Scott, Mark Nixon, and Wright State University’s Dr. Jennie J. Gallimore presented You Want What in Your Operator Displays? – Methods and Best Practices to Map Displays to Decisions. Based upon COP research underway in display content, organization and format with the goal of improving operator performance, the presenters raised the questions:

  • Do your displays present the right information to operators at the right time?
  • How do you know?

The current approach is typically to design the operator displays based on P & ID drawings because it’s easy to estimate and implement and is the lowest cost approach to display design. The question is—is this sufficient?

The presenters explored display designs based on content, organization, and format. Ultimately, these operator displays are used to provide information to support decisions to operate the plant optimally and return the plant from abnormal situations back to normal conditions. The data on these operator displays starts with analyzing the decisions operators need to make sure the information is present, is not obscured through information overload, and has intuitive links to related information.

Next is how to organize the information based on equipment groupings and operator display hierarchies. The presented shared 4 levels where the levels and number per level depends on complexity:

  • Level 1—high level overviews & alarms
  • Level 2—primary operation (unit wide operation)
  • Level 3—secondary operation (task oriented operation)
  • Level 4—process detail or support graphic

The research applied cluster analysis of operational decisions to define display information and organization. It is used to find groups of objects, where the objects in a group are similar (or related) to one another and different from (or unrelated to) the objects in other groups. In this case, operator decisions are mapped to operator displays to help identify the key operator decisions for each major section of the process (e.g. area, unit) coupled with having experienced operators define the importance of the data for each decision. Cluster analysis is done on the data, based on importance to determine cluster groups. The number of groups can be used to define the display hierarchy needed to move from high-level situational awareness across multiple decisions down to individual data points on mimic displays.

This method begins with the specific goal(s) for the display(s) to be designed. Next, the decisions that the operator must make and the information that is needed to make these decisions is required. Then, using rating and cluster techniques, the information to the decisions is mapped. Finally, rules are determined for separating information for the different display levels.

By combining the perspectives of process manufacturers, engineering organizations, automation suppliers, and academia, this practical research can be put into application to improve operator performance and overall plant efficiency.

Leave a Reply