How does automation shape operator performance?

Performance-Shaping Factors Series

NEWSLETTER ARTICLE

    
 

Lesson 3 – Automation

Since the Industrial Revolution, machines have been replacing human labor. It's seen as a struggle for control over means of production—man versus machine or, as we say today, person versus system. In human factors, however, this saying refers to one of the first issues tackled by the discipline: Which tasks should be allocated to humans and which to machines? Dr. Paul Fitts of Ohio State University worked with the US Air Force to arrive at a set of criteria separating what people do well and what machines, i.e., automation, do well (Figure 2), although some changes have occurred since first proposed in 1951. 

The "too much to do" dilemma . . .

But what happens when the primary reason that the person is present is to carry out tasks that could otherwise be automated? We refer to these as staffing bottlenecks, a condition that requires an operator be present that generally would not otherwise be needed for safe operation. If an operator needs to turn a valve to prevent some catastrophic situation, then that operator must be present even if they have nothing else to do. Analogous, though less dramatic, situations are seen in process plants all the time. Both console and field operators who have little to do on a routine basis or during certain periods of time (e.g., nights when maintenance is absent) must be present in case an upset should occur. They must be present because there’s “too much to do” for there to be one less operator.

Breaking bottlenecks . . .

Automation can often remove the staffing bottlenecks, the critical task(s) that currently must be performed by an operator. In the field, manipulation of equipment has been done by control valves for decades. Isolation valves are often used where the consequences of failing to isolate a line can be catastrophic. Those same valves can be used for situations where the consequence may not be catastrophic but necessitates the presence of an otherwise underutilized operator. Remote start/stops on pumps are common in pipelines, but process plants often feel a person must be present, at least for the start of a pump. Automation can start the pump faster and more reliably than a person.

Automation can drastically reduce workload and improve reliability during process upsets. Consider a process where during abnormal situations the operator must remove heat (energy) from the system. Now that often entails the operator going to multiple displays, taking manual control of a valve, closing it, and then proceeding to the next one. It is not uncommon for the operator to make these adjustments on a dozen valves. How long would it take an operator to perform these actions (assuming they didn’t get interrupted in the middle and forget to come back to this task)? One-half a minute? A minute? Two minutes? Five minutes? Automation can do all dozen in a few seconds upon command from the operator. Obviously, an operator will give a little more thought to initiating this automation then they would for taking manual control of any one valve. However, human performance research has shown that even with the extra time deciding to initiate the automation it is far superior to doing each step by itself. This is called the decision complexity advantage; one complex decision is superior to many simple decisions.

Staying in your lane . . .

There are cases where automation should function without input from the operator, to ensure that certain actions take place. These are the domain of the safety instrumented systems, preventing reactor runaways or firebox explosions. They should define the boundary of which the process must not exceed, or a serious safety condition could occur. However, that boundary must be indicative of an unsafe condition that will occur imminently. If it’s not an imminent threat, then the automation has “cried wolf”.

Is automation infallible and incapable of failure? Of course not. In fact, its failure probability is generally known. That the automation will fail is why a person is always needed to assess the impact and CREATE the proper response. The US Navy attempted several years ago to create a totally automated ship. It wouldn’t sail. De-automation was necessary for it to become functional. As Paul Fitts noted, the goal is not “all person” or “all automation” but the balance that has each doing what they do best.

Like one of the team . . .

Where automation is independent of the operator, one of the major failure modes is the operator. Automation, in other words, should be thought of as another operator. The US military often calls it the electronic crew member. If the operator disagrees with what the automation is doing, they have been known to bypass the automatic system. The operators at Three Mile Island shut down the automatic emergency cooling because they failed to understand the true event. Three operators on a 200,000 bbl/day crude unit introduced fuel gas into a hot, tripped heater because they misunderstood what the automatic system was doing, resulting in a firebox explosion. Prevention of these instances requires that, like all good teams, everyone is on the same page, using the same script. The automation cannot be hidden from the operator, its actions and algorithms must be easily understood so the operator can integrate them with the actions of the team, just like you do for the human crew members.

Automation is one of the key tools in the human factors engineer’s toolkit. Proper automation can radically improve operator performance. Improper automation can result in failure of the automation at the worst possible time. The key is the proper balance and integration of the automation and operator.

Copyright 2023 Beville Operator Performance Specialists, Inc., All Rights Reserved