The work performed can be used to create a structure for operator decision making that would accommodate any process unit. However, the work is not in a state for easy application. Further refinement is necessary.
Dr. Chris Hale of Georgia Tech Research Institute was to further develop the cluster analysis technique. The project instead attempted to develop new mathematical models for presenting process data. Project Champion made decision to end project without completing all RFP requirements when it became clear that researcher could not deliver.
- Validate the use of mapping decisions to data as a means to identify content and hierarchy requirements for displays
- Develop a methodology for creation of decisions used in mapping
- Test results against current display system
Important Note About Project Objective, Method and Deliverables:
The intent of the RFP was to address concerns with Display Content I on how to determine the operator decisions used in the analysis and apply the technique to determine what, if any, differences it would produce in display structure. The researcher focused on decisions the operators make, but at too high of a level. An effort was made to have the researcher focus on the sort of decisions used in the Display Content I analysis, but it became clear that they were not capable of producing the sort of system that would address the need underlying the RFP request. Project Champion made decision to end project without completing all RFP requirements.
- Does cluster analysis of the data which are needed to make key decisions result in a systematic method to help plants determine the content and hierarchy of displays?
- How would a plant go about creating the decisions to be evaluated?
- How different would a display system be using this method than the current?
This project continues research begun by Dr. Jennie Gallimore of Wright State University. In Phase I, cluster analysis was applied to operators ranking the value of different parameters in making key decisions. The results indicated that this may be a way for operators to effectively and easily identify the content of their display system and the automatic creation of a display hierarchy. One concern was how a plant would create the decisions used in the ranking process.
See Note above under Objectives. Not all of the deliverables were met.
- Critical decisions in refinery processes and a description of the methods by which the decisions are identified.
- Information required for each critical decision and its hierarchical clustering.
- A description of the methods used to identify information requirements and their organization, algorithms used in analysis, and any software .
- The decision taxonomy applied to critical decisions; along with a description of the analysis used in defining the taxonomy, algorithms used in the analysis, and any software code uniquely developed for the analysis.
- Rules/examples used to define display formats and behaviors for each critical decision.