In today’s competitive environment, all plants are interested in reducing costs to maximize profitability. One fixed-expense area most frequently targeted is hourly staffing level. If one duty station can be eliminated in a plant that operates 24-hours-a-day, head count can be reduced by at least four full-time employees. Taking into account salary, benefits, overtime and other related expenses, this typically represents a savings of $500,000 per year. However, one major incident caused by inadequate staffing can wipe out these savings for many years.
Management argues for fewer positions, whereas the workers, who are frequently unionized, always want more positions. A question that always comes up is, “What’s the right number of operators for our facility?” This question does not have an easy answer, as German chemical manufacturer Hoechst discovered a few years ago. After a series of accidents, the German Government and Labor Unions insisted that Hoechst rehire recently laid-off employees.
Due to similar incidents in the industry, different local, state and federal organizations are getting more and more involved in setting criteria that must be addressed before staffing reductions can be implemented.
Until recently, the method most often employed to determine staffing reductions was management edict. Some manager would determine that a consolidation would take place and it was imposed upon the workforce. Unfortunately, many managers, although well intentioned, don’t have the skills or experience to understand the full scope of the change because their previous education has been predominantly engineering and accounting, with little in the way of human factors and personnel development.
The task analysis method, or time-in-motion study, is still in use today, but has several shortcomings. As prescribed by this method, an observer follows a few operators and makes note of the amount of time they spend doing certain tasks. The most obvious problem with this method is that even the most dedicated operator will change his behavior while being watched. Likewise, any unusual conditions that occur during the sample period, such as a large amount of maintenance activities, will cause the job to look busier than it really is. Further, this method does not attempt to address any management-system issues. Although this is an improvement over the management edict, the method is still sorely lacking.
Many companies have done one task analysis, and few have followed its recommendations.
There are other methods that better account for the issues we will discuss. One such method that is factual, uses empirical data, and has a qualitative risk assessment at its conclusion is outlined below.
Assess several factors
Some time ago, a guideline was developed and generally accepted that a console operator could handle about 200 control loops, with an upper limit of around 280 loops. Interestingly, it’s difficult to find the origin of this rule. A literature search yields no studies or published articles rationalizing the estimate. However, like a good urban legend, it’s widely believed and still used today to determine console loading at many facilities.
But what really impacts the console operator’s work load? Is it really as simple as just the number of loops?
Loop count is a measure of control span and it is certainly a factor in console operator workload. Loop count gives us an idea of how many things the operator needs to keep an eye on at any one time. But, does storage-tank-level control create the same workload as reactor-temperature control? Is a small wash-water flow adding as much to the workload as a furnace fuel-gas-flow controller? I think we can all agree that the answer to both questions is no. It’s not just the number of loops, but the type of equipment and the complexity of the process that contributes to the workload.
Anyone who has spent any time in a control room knows that a large portion of an operator’s work can be caused by upstream or downstream disturbances. A unit drawing feed from tankage and sending its products to storage is typically much easier to run than a highly integrated unit that takes hot feed from other processes and directly feeds another unit(s). These interactions can have a huge impact on job complexity and must be considered. Likewise, it requires more effort for a console operator to communicate with a large number of closely linked units than if the feed and products are isolated by tankage.
How about the physical situation of the operator? Can this impact the operator’s workload? An operator’s duties and his surroundings vary from working out of a local shelter and having both field and control responsibilities, to dedicated console operators based in a centralized control room. Might we reasonably expect the dedicated operator to handle more control work than an inside/outside operator?
What about the type of control hardware and the level of automation? A variety of control hardware is in use today, from pneumatic field controllers to distributed control systems (DCSs), and everything in between. To complicate the situation further, some plants have advanced dynamic control and online optimization, whereas others have poorly tuned loops with a large percentage not running in their optimum mode. It seems reasonable that the operator with the DCS and advanced control can handle more responsibility than the operator with poorly tuned loops who is forced to make frequent, manual adjustments.
Let’s look further at the DCS, since such systems are frequently used to justify staffing reductions. Although the DCS can be a huge improvement compared to the old panel boards, it can also introduce its own problems. Poorly laid-out graphics can make important information easy to miss in the clutter. Poor navigation schemes can cause operators to waste time during an emergency while they move between screens trying to find critical information. Displays that show a small portion of the plant in great detail can create a situation where the operator is looking at the plant through a keyhole. Poorly designed alarm systems can generate thousands of alarms in minutes during an upset so that operators miss crucial alarms (see CP, January, p. 28, for more information about alarm overload and how to deal with it).