Technique can be used to:
- Support display design for decision making.
- Support organization of information into hierarchies.
- Support investigating differences among operators (e.g. novice vs experts) to support identification of knowledge gaps and training.
- Be combined with other data gathering methods for more robust information, including data related to actual screen usage and eye tracking patterns.
Dr. Jennie Gallimore of Wright State University adapted a method to determine display content using cluster analysis of ranking for the importance of data on different decisions. Operators from two refineries rated the importance of key process variables in answering questions console operators might have about their process (e.g., why am I venting so much?). The results were to be tested on an actual plant to determine how it would alter the existing display structure, but that was not conducted. Despite being only in Excel, the approach has been used successfully by at least three member companies.
The purpose of this project was to develop a technique in order to map data and information to the decisions that chemical process operators are making while supervising a process. Doing so improves the process of display design so current display designers within the operating plants can determine what information should be used and how it should be organized with respect to high level information down to low-level information.
The important question is whether current SCADA display techniques are appropriate for the types of decisions that are made and what information is actionable. If presentation of information to support SA and decision-making is not appropriate, color will make little or no difference. We need to ask the questions: Do current graphics/displays provide information in the best manner for assessment of the health/status for the operator’s span of control? What are the benefits of current representations? What are the limitations? What is the best way to combine the information so that it is easily available given the decisions being made? How do operator strategies affect how information should be displayed? What information do expert systems controllers look for and focus on throughout the day under normal and abnormal circumstances? Is this different for experts versus novices?
The specific method, Mapping Information to Decisions and Displays (MIDD), included the following basic steps: 1) determining what information is needed for specific operator decisions (accomplished using survey techniques and subject matter experts), 2) using statistical clustering techniques to determine how data are related across decisions, and 3) determining how the cluster output of information should be grouped. The last item showing the number of hierarchical levels needed to move from high level situation awareness across multiple decisions down to individual data points already available on the mimic displays.
- Methodology for display assessment and development (cluster analysis);
- Outcome of cluster analysis;
- Recommendations for possible display alternatives for unit(s) studied
- Project raw data files are available for use by FHR and Marathon. Contact Lisa Via for assistance.