Changes In Operator Activities Due To Upsets; When the ---- Hits The Fan

NEWSLETTER ARTICLE

Just how bad do things get when problems arise on processing units? How much does a control board operator's tempo of activity increase when responding to an upset? Knowing the nature and magnitude of the changes can help in formulating strategies to aid the operators' upset response.

In our last Newsletter, an analysis was done on the top quartile of control board operator samples from our job sampling database. The analysis classified the samples in relation to the events causing high loading. Results of the analysis indicated that unplanned abnormal upset events only accounted for about 4% of the samples. The next logical step was to evaluate just how much the events added to the operators' loading. This study is the follow-up that examines the workload differences between the steady state and upset conditions.

For a few of the upset samples the exact causes of the upsets could not be clearly identified before the sample had ended. As a result, a couple of the samples from the database were not included in the present study.

In the figure below, the percentage increase in activity between steady state and upset is listed. As expected, the alarm rate, the first parameter compared, increased substantially during upsets. The range of change was from 150% to 400%. Although 400% sounds like a huge increase, at one FCC that was having differential pressure problems the alarm rate increased some 1500%.

Control changes increased by an average amount of 366%. The amount of advanced control likely increased the variance between the samples. The highest control change increase was on a process that had probably the most amount of advanced control. Operators made roughly the same number of control moves responding to upsets on all the processes. But, advanced control had the effect of reducing the number of control adjustments during steady state, making the change between steady state and upset more pronounced. This highlights the biggest problem of advanced control and operator performance. During steady state, advanced control robs the operator of the opportunity to adjust the process and learn it's control characteristics. During upsets, the unpracticed operator must figure out what the advanced control was doing, retake control and stabilize the process.

Another item to notice in the table is that the increases in alarms and control changes do not track one another. A high number of alarms did not necessarily result in a high number of control changes.

Communications had some interesting results that were related to the nature of the upset. Communications to outside operators increased roughly 200% during all of the upsets. Communications to other units increased on those upsets that involved other units, but decreased on the upset that did not involve other units, the desalter line pluggage. The desalter line pluggage control operator was also in a centralized control center. An argument can also be made that the central control center allowed other interacting controllers to notice the controller's problems without having to verbally communicate.

The data clearly illustrate the challenges faced by operators when trying to handle upset conditions. Managers can expect the level of activities to more than double during upsets. The large increase illustrates why control operators need a significant workload reserve to handle upset conditions with current control system configurations.

The data also illustrate that upsets cut across all aspects of the operators' job, it is not just a control system problem. Focusing on the control system alone will not solve all of the problems of an upset. Other aspects of the operators' job, such as communications, also need to be addressed.

The data show that more resources need to be spent on moderating activity swings between steady state and upset conditions. Clumsy automation has made many control positions a bore during steady state operations and a terror during off-normal upset conditions.

Copyright 1997 Beville Operator Performance Specialists, Inc., All Rights Reserved