Optimization

Visibility boosts efficiency, capacity

Pet food plant sees how to improve operations with automated data collection

By Paul Studebaker, editor in chief, Control

TechED banner

The United Pet Group (UPG) plant in St. Louis manufactures about 1.5 million pounds per week of small bird and pet animal feed, mostly bagged, but some molded in cakes and sticks.

"I've worked for the company for six years and have done a lot of equipment and process upgrades—anything to make the process better," Dave Seckel, maintenance electrician, UPG, told attendees of his session at Rockwell Automation TechED, in San Diego. For some time, "data collection has been a goal, but getting there is a bumpy road—a hard thing to sell. Now, with a little success here and there, things are starting to come around."

Also read: Learn about FactoryTalk Metrics

FactoryTalk Metrics helps manufacturers get a grip on data. "Most start with OEE—availability, throughput and quality—compared to the theoretical ideal. And they look at events related to downtime, so they can identify where their problems are and how to improve," said Wendy Armel, principal MES analyst, Stone Technologies.

UPG did it a little differently, by starting with labor efficiency. "Our main goal was to gain visibility," said Seckel. "We have a long history of manually collected data—how long it takes to run a lot, what faults occurred and the associated downtime. It's not very accurate."

New sheriff demands good data

The pivotal event occurred when a new plant manager arrived and wanted to get away from manual data collection. "Manual data accuracy is typically about 60%," said Armel. "Automated accuracy is usually 95%." This is because clock times are 100% accurate, but HMI entries may have errors.

"We found FactoryTalk Metrics provided the horsepower we needed to improve our efficiencies and reduce costs," said Seckel. "The software is intuitive and easy to use. Most of our plant is Rockwell Automation, so it's easy to talk to it."

UPG wanted help with the implementation and contacted Stone Technologies, which put them in touch with one of its successful clients so UPG could see how FactoryTalk Metrics works. Then UPG and Stone worked together on a pilot on UPG's S3 Bagging Line, where blended batches of seeds are packaged.

"We like to use a pilot to learn things like the terminology and plant priorities on a small scale, and develop a simple, repeatable solution with rapid payback," said Armel. "First, we talk with everybody—management, engineering, operations, maintenance—to see what they need and expect."

They start with a few basic reports, "the things they need to make decisions," Armel said. "Not fancy, not a lot of colors. If it won't improve the business, it's useless."

At UPG, the first report—labor hours—makes the project unique. The labor hours report links labor hours to each production order through HMI entries of available hours and crew size.

A line operations report adds the product SKU and "tells operations what their machine is doing," Seckel said. "It's near-real-time and highly accurate, not historical and approximate. Having visualization of what they're doing in real time gives them incentive to do better."

A last 10 line faults report shows current efficiency and what's gone wrong, such as bagger faults or cycle stops.

Small start gives rapid ROI

The initial implementation reduced overtime by 10% and increased labor efficiency 15% by reducing the number of line leaders in half. "Automated data collection means one line leader can now handle two lines," Seckel said. "We were able to move half the line leaders elsewhere in the plant."

Seckel recently added the application to a seed blender. "I configured the seed blender in less than a day and started collecting data," he said. "Based on what they had told me for estimated cycle times, our first OEE measurements were on the order of 180%. It's very interesting to compare what people think they run to what they actually run.

"In practice, automated OEE tends to be lower than manually collected OEE; it's real easy to push the pencil," said Seckel.

UPG's implementations were not without challenges. One was cultural—"to change people's mindsets, the tribal knowledge of how we've done things for years," Seckel said. The other was integrating data from proprietary scale systems, which was overcome by working with the scale OEM.

On the plus side, UPG can now push reports to management. "It gets them used to seeing it, and they don't have to come and get it," Seckel said.

The mechanic can see how the machines are doing and estimate when he should schedule preventative maintenance. "We'll automate that and tell him when the machine has done, say, 5,000 cycles and it's due for maintenance," Seckel added.

Above all, "we can see which operation or equipment is setting the pace of production and where to make improvements," Seckel said.

Formula for success

Armel agreed that cultural acceptance is critical. "Without it, having your technology at 100% is worthless," she said. "Give the people what they want and need, and talk to everybody; the one guy you don't talk to will bring the house down."

When developing the application, "find a common thread, use common terminology as much as you can, and create a simple, standard solution that's easy to explain and to expand to multiple applications in the plant," advised Armel. "Standard objects shorten the implementation time and transfer easily to other equipment and other areas."

To gain buy-in and rapid ROI, "use a solution that has a low out-of-the-box cost and start saving money right away," Armel said.

Start with a few key reports to learn more about the data. "Don't turn in 60 reports," Armel said. "Production overview gives OEE, and events let you see why you're not reaching the desired levels. Custom reporting is easily accomplished, and you can measure anything, but focus on the information that will help you satisfy business needs."

Then take advantage of having accurate data to improve operations. After all, "if you solve the wrong problem, you'll have two problems to solve," Armel said. "FactoryTalk Metrics lets you identify the right problems to solve."