Redesigning Process Plant Jobs Through Job Task Inventories? Be Careful!

NEWSLETTER ARTICLE

Over the past couple of years, Beville has encountered a number of chemical process plants that were redesigning hourly jobs and had used job task inventories to model operator activity. Managers wanted to know if operators had spare capacity to absorb new tasks and if some positions could be combined.

To answer the questions, just like the textbooks said to do, plant personnel carefully compiled detailed lists of tasks operators complete while on shift and assigned estimated task times for each task. The lists included three items: (1) the list of tasks, (2) the task duration for each task, and (3) the number of times the task was done over the course of the shift. Unfortunately, when the task times were summed, the task times exceeded the number of hours in a shift. At one plant, the reported total exceeded available time by 20% while at another plant, the excess was nearly 50%. What went wrong; why didn't the technique work?

There are several reasons the task list techniques didn't work. The underlying reason is because users didn't recognize the inherent limitations of the techniques they used. Most industrial engineering techniques have been developed for one type of work: piecemeal, assembly line discrete production. These jobs have specific lists of tasks that are completed in the same order time after time with little variation. For this type of work, the job task inventories with self-reported task times are easy to do and provide reasonable estimates.

Unfortunately, most chemical process plant jobs don't fit into this piecemeal work category. Unlike assembly line work, process plant output is not tightly coupled to worker output. Continuous actions by the operator don't directly result in related amounts of product produced.

Another reason why the task list approach didn't work is due to a higher degree of variability and flexibility in process plant operations. This variability comes in the form of maintenance tasks, equipment problems, etc., causing spurious peaks in workload.

Flexibility in process plant jobs relates to how some tasks can be done in parallel. Assembly line/piecemeal work does not lend itself to parallel task completion; there are specific steps that have to be taken in set orders. In contrast, process plant jobs have a certain degree of flexibility, and the tasks can be done in different orders or combinations. For example, during outside equipment inspection rounds, operators often collect samples or perform other tasks in parallel with their rounds. Not accounting for parallel tasks leads to overestimation of task times.

Because of these issues, standard industrial engineering techniques that are applied without prudence to process plants are almost certain to give misleading answers. For process plants, techniques beyond the standard task lists should be used for modeling operator's activity and estimating operator workload. And rather than focus on a single metric of job performance, multiple indicators should be used. The best way to capture the different facets of process plant operator workload is to create profiles of the operator's job activity.

To derive a valid assessment of operator activities, two distinct phases of operations need to be addressed: (1) steady state operations and (2) plant upsets.

For steady state operations, the time-tested method of direct observations provides the most accurate task data. Variations of the classical Taylor time-and-motion studies can be used to capture the highest steady state workload periods. Task lists provide expectations of what they should be doing; observations provide data on what they are actually doing.

Plant upsets are best evaluated through a task analysis based method looking at worst-case upsets. Rather than focusing strictly on time, the risk of not completing the task and effort needed to complete the task can be defined and rated. The ratings can be benchmarked and cross-compared.

Accurately estimating process plant operator workload is far more complex than compiling task lists and assigning task times. If you would like more information on estimating operator workload and job design, please visit our web site at www.beville.com.

Copyright 2001 Beville Operator Performance Specialists, Inc., All Rights Reserved