OEE at Teva: Leveraging the Simplest KPI

Teva Parenteral Medicines is realizing the initial benefits of an overall equipment effectiveness program, with an eye towards less downtime and greater productivity sitewide.

By Paul Thomas, Senior Editor

Share Print Related RSS

Overall equipment effectiveness (OEE) is a measure of production uptime, a cornerstone of many Lean programs in that it ultimately aims to eliminate waste in the form of downtime. Its basic formula is well known: OEE = Availability x Performance x Quality.

But OEE projects have varied purposes, from identifying causes of downtime and wasted labor and materials, to driving customer service and regulatory compliance. By providing information on how well, and how often, production lines are working, OEE is a valuable tool for manufacturers to improve operations and, importantly, cut production and packaging costs.

Teva Parenteral Medicines, located in Irvine, California, began an OEE implementation in the spring of 2009, to get a better handle on how its production and packaging lines were performing, and what cost savings could be realized in doing so. TPM is Teva’s only sterile injectibles manufacturing facility in the U.S., with six vial and syringe filling lines and five packaging lines across 60,000 square feet of manufacturing space. There was no glaring problem in Irvine, just a lingering sense that lines were not performing optimally, says Duane Hiveley, TPM’s manager for controls engineering. And while Teva USA has embarked on a broad Lean initiative, this project was undertaken locally in Irvine as part of ongoing efforts to improve efficiency in operations;OEE, Hiveley knew, could provide real data on equipment performance and causes of downtime. As the old saying goes, you can’t improve what you can’t measure.

With the program in its infancy, Teva is just beginning to baseline its OEE and develop new performance standards. Some of the early benefits also include:

  • Instant visibility to hourly packaging rates for line operators
  • Instant visibility to shift performance for line supervisors
  • Greater operator awareness about line performance
  • Near real-time data summarization for production rates, performance, and downtime events
  • Increased data granularity over line events and equipment downtime events to account for and evaluate lost capacity 

A more significant milestone will come soon, when TPM will eliminate the manual reporting of downtime at the end of every run. This will not only eliminate errors, Hiveley notes, but allow data to be collected within a single source, requiring no aggregation.

Looking Back

Hiveley’s goal for his lines is an OEE of 65% to 70%, typical of world-class pharmaceutical facilities. This represents a significant break with the past, when OEE could not even be calculated accurately. Previously, the only information that Hiveley and colleagues had about their equipment was that collected between runs. Operators would be asked to reconcile what had happened over the last couple days, to jot down notes from memory about issues that had arisen and what actions were taken to overcome them. “Standards” were developed based not upon production goals and ideals, but upon average performance from past production runs.

It was challenging to maintain a concentrated effort for a project that was not deemed operations critical. “OEE is not something that you must have to run your lines,” Hiveley notes. TPM had done some limited pilot testing of OEE on a filling line in 2006, but a clear opportunity arose when industrial engineers from Teva’s global operations visited the Irvine plant and identified OEE as a need for their packaging lines. Hiveley now had a mandate and support of upper management to pursue an OEE system (OEE projects had proved successful in two of the company’s Israeli facilities.).

This coincided with TPM’s decision to install a newer version of its manufacturing metrics platform, Rockwell’s FactoryTalk, which includes OEE metrics among the suite’s offerings. Because TPM’s early pilot efforts relied on Rockwell and Teva has a longstanding relationship with the automation vendor, it did not consider other OEE solutions on the market.

The OEE project team consisted of Hiveley, Beechem Exum (associate director of packaging), an outside systems integrator, and several Rockwell consultants. Some of the key considerations before implementation included which line to focus on, who needed to be involved, and what training was required. There were also many decisions to make regarding data capture—how much granularity was needed? All of these decisions were made deliberately, Hiveley says. “Do as much up front as possible,” he says. (See Box: “Steps to OEE Success.”)

The Simplest KPI

In terms of granularity, the team kept its ambitions modest—to better understand what is happening during turnover from one run to the next, and what is happening during downtimes. “If a run is supposed to take 18 hours to run but it took 24, why? Pretty basic stuff,” he says.

 

Steps to OEE Success

The following best practices for starting an OEE project are courtesy of Teva’s Duane Hiveley:

  1. Focus on a single line or key product. Starting small carries low risk, and allows you to build on success.
  2. Solve a key business problem. Look for the “tall bars” that will deliver executive sponsorship.
  3. Demonstrate tangible results. Whether those results are tactical or strategic, it’s important to deliver visible payback immediately.
  4. Validate ease of implementation. Scalability and repeatability are key.

Scope the initial project for a 4-6 month completion cycle, Hiveley suggests, to maintain attention and produce fast results.

Other critical success factors:

  • Management commitment. Many downtime systems fail due to a lack of management support rather than technical problems, Hiveley believes. Managers must want to solve downtime issues in order to give it their full attention and oversee training and implementation.
  • Engineering. From line configuration to installation timing, engineers are critical to ensuring everyone understands the process and is on the same page.
  • Maintenance. OEE is often a KPI for maintenance effectiveness, Hiveley says. It’s a chance to improve existing preventative maintenance, as well as establish a measure of self-maintenance by well-trained operators.
  • Data Usage. Visibility and accountability provided by OEE can be a double-edged sword. Data should not be used to justify punishment or blame, Hiveley cautions. Rather, it should be used to empower employees to address downtime issues, and to reward excellence when improvements have been made.

 

“OEE is the simplest KPI,” Hiveley adds. It measures how much is made that can be sold vs. how much could have been made if everything is operating correctly. “It’s a simple number that has significant meaning buried in it.”

Hiveley and the implementation team had to remind themselves not to try to do too much with the software. OEE is not suited to process monitoring or troubleshooting, or batch or process reporting, and thus the software should not be expected to provide these functionalities. “You have to understand the limitations of the software,” he says. “You can’t try to mold it into an HMI or SCADA package.”

OEE can also serve as rallying point within an organization. “These kinds of systems provide a common platform for operators, engineers, and executives to communicate with,” he says, “instead of operations holding this information to themselves, and taking like a Siskel and Ebert approach—you know, thumbs up we’re doing good or thumbs down we’re not doing so good.”

Hiveley uses an example to illustrate: On some sections of the packaging lines at TPM, vials are manually inspected by QA personnel for major physical defects. One day, this inspector was pulled from the line due to priorities with another line. As a result, the line had to shut down for several hours due to personnel shortage.

OEE would call attention to this situation if it were a chronic problem. “With OEE, this kind of thing can be looked at as a pattern, as a common thing that is continually shutting down the line,” Hiveley explains. “It can spur discussion with other departments—not just within operations—and it will elevate the discussion with management.”

Implementation

The decision was made to launch OEE on a packaging line with high visibility and utilization. To minimize disruption, the solution needed to be “bolt on” and easily integrated with existing production operations. Photoelectric eyes connected to a PLC were strategically placed in locations to measure output (parts produced), as well as signal downtime—defined as whenever a machine has been stopped for one minute, says Hiveley. The collected data was made highly visible with customized content delivered via large LCD display screens accessible to line operators and accessible to others on the company Intranet (via an Ethernet connection to Microsoft SQL and OPC data servers).

Installation of the equipment and software was a matter of days, and the training and implementation “a matter of weeks, not months,” says Hiveley. Implementation cost was between $60,000 and $85,000: up to $10K for the servers, $15-25K for the PLCs, HMI’s and their integration, and $40-50K for the software and its configuration. (Future implementations should run approximately one-third of this cost, Hiveley says, due to efficiencies of scale and replication.)

There were few implementation challenges, Hiveley says, because it required no integration with the site’s manufacturing control system. “Once you intrude into those systems, you’ve got to worry about change control, and you may have to do revalidation,” Hiveley says. “And when you’re hooking up to legacy-type systems, you encounter a whole other set of challenges that will increase your cost and validation time.”

Again, simplicity was key. “One of the misperceptions with OEE is that you have to go down to the automation layer to get all the data out that you want,” Hiveley says.

Another common misconception is that, for OEE as with other systems, real-time data is best. What is real time? Hiveley asks. “For an OEE system, if the data you’re getting is 15 minutes or a half-hour old, likely its good enough,” he says. “We’re doing it better than that, on the order of every 5 or 10 minutes, which is certainly more than what we need. It’s certainly better than what we had, which was hours and days.”

Early Returns

Besides the operational benefits being realized, OEE is paying dividends as a motivational tool at Teva. Operators get a constant reminder of how well they’re doing. “It’s a constant pat on the back, or a reminder that you need to step it up,” Hiveley says. One plant-floor display screen indicates how much production is done during each shift. “It has sparked healthy competition between shifts,” he says.

There were concerns among operators that OEE could, as Hiveley puts it, make them feel like Big Brother was watching them. In response, he communicated from the start that OEE was a tool to empower employees, not to blame and punish them. Operators were also given the opportunity to provide feedback and fine-tune the system after the initial installation, helping to ease concerns.

Management has already recognized the value that OEE has brought to Irvine. “You know, for sure, where you’re at and that information can continually be produced,” Hiveley says. “You’re not expending more people and resources to get that information once it’s in place.”

Continuous improvement will be a key focus going forward. “I plan to get together with operations and upper management on a weekly basis and go through what’s been going on with this line and what information we’re getting from the system, just so we can have a hands-on approach to what the software is really doing for us,” Hiveley states.

Hiveley is in the process of developing a baseline documentation for an OEE system design, so that future implementations can be based upon common ideas and protocols, and so that consistency can be maintained should different project managers or systems integrators be involved.

The OEE data collected in Irvine will also allow TPM to benchmark itself against other Teva facilities. Granted, vial and liquid filling and packaging is different than solid dosage and other forms of production, but “OEE itself can be a common connection between plants,” Hiveley says.

 

Share Print Reprints Permissions

What are your comments?

You cannot post comments until you have logged in. Login Here.

Comments

No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments