The Trick To Applying OEE Across Several Plants

Be sure you’re comparing apples to apples when using overall equipment effectiveness tools.

Aw 4831 Oee15

Motivating others to squeeze more from the capital equipment in your company’s plants is not easy, especially if you’re not always there to do it. But that doesn’t stop Stephen Sefton, the manager of enabling technologies in the United States for Rexam PLC, a London-based producer of beverage cans. Like a growing number of executives in his situation, he has found that using measures such as overall equipment effectiveness (OEE) can give him the tools that he needs to drive continuous improvement—and greater profits—at his company’s U.S. facilities.

OEE is a dimensionless efficiency index calculated from the equipment’s availability, production rates and quality. “Managers can use it to benchmark or analyze operational metrics across similar plants to identify best practices,” explains David Brochu, executive vice president, Acumence Inc., the Chicago-based company that helped Rexam to automate OEE computations and reporting. By comparing the historical and current index against these benchmarks, managers can gain valuable insights into the effectiveness of their capital assets, identify problem areas and make decisions about investment.

Although plant managers have been using OEE for a while now to make similar decisions at the plant level, corporate executives are grappling with deciding what role the metric should play in evaluating a group of plants. In the past, the corporate level of management relied almost exclusively on the financial statements for evaluating performance and making decisions about the business. Using money as the universal measure orients all evaluations automatically toward the ultimate goal of the business—to turn a healthy profit. Because metrics such as OEE are not quite as clear-cut as a financial statement, executives tended to relegate them to the background as supplemental data.

“That is changing, though,” notes Darren Riley, Manufacturing Execution Systems (MES) business development manager for the electronics industry for Milwaukee-based Rockwell Automation Inc. “There has been too much lag between the operations and administrative sides of the business.” Besides being slow, the financials also lack the detail and accuracy often found in the data used to calculate OEE. Consequently, many executives are turning to OEE to extract the current data that they need from their automated equipment to accommodate rapidly changing business conditions.

Such is the case at Rexam. The software supplied by Acumence extracts details about the operation from the controllers overseeing the can manufacturer’s machinery, calculates the efficiency metric, and reports it to management over the company’s intranet. “[Rexam manager] Sefton can look at how well the lines are doing on his screen in real time,” says Brochu. “If that number starts to look bad, you know you’re going to get a phone call.” Knowing that corporate management takes these numbers seriously motivates plant management not only to keep them high but also to nudge them ever higher.

 

A Controversial Number

Although few dispute that OEE is a great benchmark for measuring the progress of a process, just how much one can use the metric to compare operations across several plants is a matter of controversy. OEE has some profound limitations, and managers must know what they are in order to keep the numbers in perspective.

To illustrate the point, Brochu explains how differences in product mix can skew the numbers. Consider a plant dedicated solely to making cans for Coke Classic, for example. It will be much more efficient than one that makes cans for several Coke products, such as Diet Coke and Diet Coke with Lime, because it will have less downtime for changeovers than the multi-line plant. Consequently, if the dedicated plant’s efficiency were 90 percent, it would appear to outperform the other plant running at, say, 80 percent. But the lack of changeovers would mean that the plant is capable of running at a higher OEE, perhaps 95 percent; whereas the plant with more changeovers might be capable of only 82 percent. In this case, the plant running at 80 percent would actually be more efficient, as it was running closer to its maximum availability.

It’s also important to recognize that efficiency is not the same thing as profitability. “OEE is a proxy for profitability, but not a perfect one,” says Brochu. If a highly profitable product has a measure of complexity that requires a lengthy setup and frequent adjustments to the machinery, then the process for producing it will have a lower OEE than a less profitable product that is much easier to make. So Brochu thinks that executives should really consider each plant’s profit contribution to overall business performance when making investment decisions.

Most other vendors concede the point and agree that OEE cannot be the sole measure of an individual plant’s value to the business. They argue, though, that OEE can help executives to compare the value of manufacturing assets across several plants if the calculations and comparisons are done correctly. “What is required is agreement among plant managers on how they measure the three factors for calculating OEE—availability, rate, and quality,” says Magnus Pousette, vice president, ABB Reliability Services, North America.

If management defines the factors accurately, he believes that OEE will be a good indicator of how well a plant’s assets are being used, regardless of the actual volume produced at the plant. “OEE can be used with financial metrics such as  return on capital employed (ROCE), to make decisions on whether to keep a plant open, close it, invest in it, or consolidate it with another operation,” he says.

One way to form the necessary consensus among managers is to conduct an analysis and determine what each plant’s maximum availability can be. Then calculating OEE is a matter of subtracting the inefficiencies that keep the plant from operating at this theoretical level. “The real availability as a percentage of maximum possible availability is multiplied with real speed as a percentage of maximum possible speed and multiplied again with total sellable product as a percentage of maximum possible sellable product,” says Pousette.

This suggests that management might need to let the plants calculate the three major factors for computing OEE—performance, quality and availability—in slightly different ways to get a truly “apples to apples” comparison among plants that make different products. It might even mean keeping two sets of numbers. “In cases where our software generates dual calculations for a plant, we have seen the ‘local’ number used for performance payments and comparison with historical data, and the corporate calculation used for comparisons across the enterprise,” says Barry Lynch, product manager at GE Fanuc Automation, in Charlottesville, Va.

To offer this flexibility, GE Fanuc developed its software to have both a “standard” and a custom OEE calculation. “The strength of doing it in the software, as opposed to through the programmable logic controller (PLC) or distributed control system, is that it allows modifying the calculations and formulas over time,” says Lynch. Management can adjust the calculations to accommodate new products, fluctuating demand, mergers and other changes in the business.‑

 

Automation vs. Clipboards

When advising companies on collecting the data, most automation vendors recommend giving serious thought to extracting it from the equipment automatically. “The amount of automation required depends on the industry and process you are measuring,” says Lynch. “If you are measuring OEE on a high-speed canning line making 2,000 cans per minute, for example, then, yes, you do need automation to count cans.”

Manual entry, on the other hand, is often acceptable on slow batch processes that work over several days. Moreover, no matter how smart the automation is, it usually needs some form of input from people. For example, most equipment capable of running more than one product will need a human being to tell it which product it is running. Then, when a problem halts production, the operator also might have to tell the machine’s controller what went wrong by selecting the cause from a menu. So the raw data for OEE calculations typically comes from a mixture of manual, semiautomatic and automatic means.

Nevertheless, erring on the side of automating has its benefits. “Many forms of electronic data collection devices and software packages exist today for delivering actionable information immediately,” says Jim Feltman, sales manager at Vorne Industries Inc., an Itasca, Ill.-based manufacturer of real-time monitoring products. “Typical data-collection techniques range from leveraging PLCs already on the equipment to using ‘bolt-on’ data collection devices that require just a couple of sensor inputs.”

Such automation solves the problem of receiving feedback a day, or even a week, after events have occurred. Controllers are always on the job, gathering and recording pertinent data minute-by-minute as the events happen. Filling out forms, spreadsheets or white boards manually simply cannot compete with the speed and detail with which automation can collect information and generate real-time process feedback. Automation eliminates any lag and the resulting reactionary approach to correcting problems that would otherwise sap manufacturing processes of their effectiveness, notes Feltman.

Collecting data automatically also eliminates human bias, both intentional and unintentional, that can mask problems. A good example of how bias can affect OEE is a bakery that Rockwell Automation helped to boost its equipment utilization. As Rockwell’s experts and the client’s engineering staff stood in front of the machinery that was the facility’s chief constraint, material stopped going into it. “We asked, how are you going to record that stoppage?” recalls Riley. “They said, we don’t record that stoppage because it was less than 15 seconds.” The bias was that those micro-stoppages were too short to really matter. In reality, however, their large number eroded OEE significantly.

Short stoppages are not the only events that might seem too insignificant to record or might even escape notice. Slight reductions in the speed of a process also are good at fooling people into ignoring them. Vorne’s Feltman urges his clients to resist the temptation. “Consider a process capable of running 5,000 pieces per hour, but only running 3,500,” he says. “Over an eight-hour shift, 12,000 pieces of salable product that should have been produced, weren’t.”

Once an accurate history is developed, management can find and fix the true causes of inefficiency. “People usually make up more of the unnecessary downtime than the machinery does,” says Rockwell’s Riley. Often, the data reveal that the biggest percentage of downtime was actually waiting for an operator to notice a jam, walk to the source of the problem, clear it and acknowledge the action in the controller. Armed with this information, management can determine whether to rectify the problem through training, applying some form of motivation, streamlining access to the machine, redistributing the workload, or adding people.

 

A Tale of Two Strategies

Two basic strategies exist for installing the automation necessary for collecting enough detail for performing such analyses. The first is to conduct a pilot program at one plant to prove the technology and methods before expanding its use to other plants. The second strategy is to do the necessary integration at all of the plants right away. The best strategy to use will depend on your organization.

“Some companies might not be ready for a massive organizational change, so it might be easier to define the requirements and integrate the automation at the single plant level,” explains Daniel Wilson, MES business consultant for Siemens Energy & Automation Inc., in Norcross, Ga.  “For one plant, you might just have to add a few sensors and some data collection equipment. But in a multi-plant environment, you’ve grown the amount of consulting, automation, and restructuring exponentially.”

In either case, however, management will still have to go through the pains of defining measurements that are useful and fair across several plants, as one of Wilson’s clients is in the process of learning to do now. Management has achieved its goal of improving throughput at the pilot plant by boosting the efficiency of its equipment by 2 percent, and wants to repeat the success at its other facilities. It is grappling with how to account for the differences in mix of products, level of manufacturing technology, and age of the equipment at each plant.

“They implemented [automation for calculating OEE in] the first plant to prove that the software works and that acting on OEE measurements can actually change behavior and affect performance,” says Wilson. “There are some inherent challenges in this approach, but they are going to end up being just as successful as someone who took a wider approach at the very beginning.” So despite the importance of technology for getting good numbers, the real trick to applying OEE across several plants is the method for ensuring that you’re comparing apples to apples. 

More in Control