This project provides data on the improvement in both operator and plant performance with upset automation (safe-park) and state-based alarming
The study evaluated the effectiveness of Alarm automation, Alarm rationalization, and the use of a Safe Park application. This was completed by Dr. Beth Blickensderfer and Dr. Joe Keebler.
Eleven Operators completed six scenarios with different malfunctions during a Propylene Refrigeration compressor trip on the Finishing side of an Ethylene facility. A high-fidelity simulator was used at the NOVA Chemicals Joffre site.
The study found the operators were able to detect the malfunction with less time using Smart alarming and using the safe park application. This resulted in less flaring and a lower perceived workload for the operators.
The rationalized alarm used in this study does not match the expected results formed by way of prior evidence from previous research. A difficulty in designing six scenarios of equal complexity may have contributed to this. As well, Alarm rationalization was developed with steady state conditions in mind and not specifically for a plant outage. Further investigations into the effects of rationalized alarms on operator performance are recommended.
It is generally an article of faith that improved alarm system performance and increased automation will improve system performance. However, little quantifiable data exists to support what degree of improvement, if any, that alarm improvements and increased automation provide. Alarm rationalizations typically cost several hundred thousand dollars to conduct and implement, plus they require the use of experienced console operators, a valuable resource. Automation of upset response can have similar costs. Do the benefits exceed the costs, and if so, by how much?
- How much improvement does proper alarm rationalization provide in upset management?
- How much improvement does automation of upset response provide?
Human performance theory argues that overall system performance is better if the operator (1) receives only useful warnings (no false alarms) and (2) is not burdened with rote control actions during an upset, those actions being delegated to the control system. Alarm management efforts have been undertaken industry wide to achieve the first item. Standards committees are underway attempting to aid in the implementation of the second item. However, how much benefit do these achieve?
COP members and staff at the NOVA Chemicals facility will -
- Modify simulator as necessary to support experimental design
- Define alarms to be adjusted for differing levels of rationalization
- Conduct testing of operators
- Gather results data from the testing
An independent researcher will -
- Develop the experimental design/protocol, special consideration needs to be given to the limited number of sample subjects (n=12) and the potential for learning effects,
- Conduct statistical analyses of the results, and
- Document results of above