Metal fabricating in sequence, by color

Continuous Improvement Manager Clay Buster (on left) and Plant Manager Mike Oliver stand by Hillphoenix’s laser cutting and panel bending manufacturing cell.

More overtime. Working another weekend, including Sunday. Spending too much time hunting through racks. Where is this part? Why isn’t this subassembly ready? More overtime. Another weekend at the factory.

That was life on the floor of Hillphoenix’s Centerville, Iowa, plant in 2014. Part of the company’s Specialty Products Division, the plant makes refrigerated display cases for grocery stores and other customers with temperature-critical applications. Every case is different. Each has different sheet metal and glass dimensions and thicknesses, shelf placement and look, refrigeration requirements, and more.

Up until several years ago, people at the plant pushed products to racks that surrounded assembly bays. The thinking was that racking more and more parts was just the nature of the beast. After all, the operation didn’t produce a consistent product. Build times for a display case varied from 10 hours to more than 350.

People never knew what was coming next, and they never knew where things stood in overall production. They didn’t have any idea of inventory levels plantwide. Not surprisingly given the circumstances, they focused on keeping their head above water and tried desperately to avoid yet another Sunday on the job. This, it was thought, was just the reality of a high-product-mix operation, particularly such a sales-driven operation that had a reputation for bending over backward for customers—right?

Over the past six years, the number of employees in Centerville has remained steady, but overtime plummeted, inventory turns doubled, lead times were cut in half, and sales grew by more than 65 percent.

Evidently, the people at Hillphoenix found a better way.

Time for Change

“We were tired of beating our head against the wall.”

So said Clay Buster, continuous improvement manager, who started at the Centerville plant six years ago. Back then it took 13 weeks for an order to ship, and people needed all seven days of the week to get many of those orders out the door. Some worked 17 consecutive weekends, and, to no one’s surprise, staff turnover continued to rise.

“We lost some really good people,” Buster recalled, adding that everyone knew the company needed to change if it were to remain competitive.

Where to Start?

This wasn’t an easy question. The company could have started with a little shop organization and standardized labeling, but that by itself wouldn’t solve the overarching problem. People could have pointed fingers at sales and “unrealistic” customer demand. But meeting that demand for customization was Hillphoenix’s competitive niche and, one could say, the reason for its existence.

Laser-cut parts are offloaded automatically to this table, then are transported to the panel bender.

Instead, people took a step back and started asking basic questions. The basics of lean manufacturing, including demand-pull part flow, sounded ideal. Downstream demand would trigger orders upstream. Nothing would be released until a unit could flow from one manufacturing step to the next, with little stopping in between, until it could finally be packaged and shipped to the customer.

But instead of preaching the merits of lean and force-feeding it to the staff, Buster and his team went to every department and asked questions. “It was about sitting down and trying to get people talking using the same language,” Buster said. “Do we believe in pull systems? What does a pull system mean to us? It will mean something different to people in sheet metal than to those in assembly. We had to have those fundamental discussions.”

These discussions helped Hillphoenix develop its current system, which adapts demand-pull manufacturing to a custom environment. The goal of any manufacturer is to shorten overall manufacturing time, and that’s difficult if you can’t see how jobs are flowing.

In a low-product-mix environment, workers can see how products flow based on a determined takt, the drumbeat of a level-loaded manufacturing facility, where every step takes the same amount of time.

The improvement team at Hillphoenix knew they couldn’t establish a standard takt. The fabrication and assembly times were just too varied and unpredictable. It’s a one-off environment. A customized display case isn’t likely to be made again. Each uses different grades of sheet metal, different cuts of vinyl, different glass dimensions and thicknesses, different refrigeration units—different everything.

Traditional single-piece part flow (in this case, “one display case” part flow) didn’t make sense either. Manufacturing times simply varied too greatly. Some simple cases took 10 hours to produce, others took five times as long or even longer. Produce a 350-hour case followed by a 10-hour case, and the 350-hour case would sit on the production line like a bump on a log or a clog in a pipe. Bottlenecks would be constant.

After extensive conversations with people on the floor, the improvement team began to see how lean could indeed be adapted. If given a group of jobs, teams in every department could work together to level-load the operation, uncovering efficiencies by sequencing tasks in specific ways. The teams still can’t work by a specific takt time, but grouping work together would allow people to decide how best to sequence the group of jobs to finish them as quickly as possible.

What Color’s Next?

Here is where Hillphoenix’s idea of “manufacturing by color” comes into the picture. As orders are released to the floor, they’re grouped sequentially into color groups, identified by the color of paper in job packets: purple, blue, yellow, tan, pink, and green. They know to tackle all purple jobs before working on any blue jobs. Everyone also keeps an eye on the colors throughout the plant. If people see a backup of purple ahead of them, they move up to help clear the bottleneck.

Following the tenets of demand-pull manufacturing, a color (that is, group of jobs) progressing forward triggers support processes into action. This includes subassemblies for refrigeration units, glass fabrication, and vinyl preparation. As a display case in the purple group moves forward at a certain stage of production, purple jobs in refrigeration are triggered into action. The timing is such that as the purple display case reaches the refrigeration installation station, the refrigeration subassemblies go right into the main display case assembly.

The triggers don’t occur just when an entire color is finished. Instead, every movement of a case triggers downstream processes into action. “If a case moves 2 feet, that triggers the framing kit,” Buster said. “It then moves 4 more feet, and that triggers us to make the refrigeration coil subassembly. When it gets to the refrigeration stage of the process, the value-added activity [of creating the coil subassembly] has been delayed just enough so that it marries perfectly with the product.”

The company has designated, tightly controlled areas for work-in-process (WIP), such as this one, just after sheet metal fabrication and before the initial framing operation for the display case. Note the blue and pink tags and job packets, which determine the sequence of release to the next operation.

The system operates with continually refined work-in-process (WIP) buffers to account for variation in production times. But those buffers are not so large that they require racks. In fact, look around and you won’t find a single rack in the entire facility.

“Using a FIFO [first-in, first-out] system before assembly, we ensure that all critical parts are available before a case build,” Buster said.

The system isn’t perfect. Mistakes happen. Perhaps it’s a problem with the bill of material; perhaps a refrigeration coil wasn’t ordered from the outside supplier. Whatever the mistake, a job is flagged, while the remaining jobs in a color progress forward to assembly. Again, the system is visual, so a flagged order is obvious to everyone in the factory, and it puts everyone on the same page when working to prevent reoccurrence.

Purchasing and Subassemblies

Buster added that this situation is a rarity, thanks largely to improvements in three areas: purchasing, subassembly, and machine capacity. For years the purchasing department had ordered everything from simple hardware to complex refrigeration coils. It was a massive task that took time and all too often opened the door for error and miscommunication. Some errors, like failing to order long-lead-time items, led to significant delay.

To remedy this, the company began interviewing vendors and adopting vendor managed inventory (VMI) for items in which it made sense to do so. This gave purchasing personnel time to focus on critical, expensive, and long-lead-time items.

When an item like a refrigeration coil arrives at Hillphoenix, it goes into a highly flexible subassembly cell. Production times are more predictable thanks to flexible fixtures and other techniques. The company also tries to place as many tasks in subassembly as they can. That way, the main assembly becomes simpler and, most important, more predictable.

Fabrication Capacity

The team also focused on machine capacity, specifically in the sheet metal department. For the pull system to work, the team knew they needed significant capacity in sheet metal cutting and bending. The machines needed to be flexible, too, considering that the “manufacture by color” demanded kit-based production.

Making things even more complex was the diverse material mix. The sheet metal mix consists of about 30 percent stainless steel; 4 percent aluminum; and the remaining carbon steel, galvanized as well as galvannealed.

The sheet metal department’s legacy resources, four turret punch presses feeding nine press brakes, were under strain. “For a pull system to work, we needed more capacity in capital equipment,” Buster explained. “You need to have available capacity in your machinery; otherwise the pull system is going to fail.”

To attain more flexible capacity, the company invested in a Salvagnini cutting laser with a sheet loading and part offloading system, along with a panel bender with automated part load and unload. Starting at the laser, parts are lifted from the nest and delivered to an offloading table. From there the operator scans the job packet bar code into the panel bender, which loads the program and moves the tools as the blank is loaded into the system. The result: zero setup time.

A display case has both sheet metal and vinyl components installed.

Even after being fed parts from the laser, the panel bender still has excess capacity. And the company makes maximum use of it, considering the unit uses universal tooling and requires no changeovers between jobs. As long as parts from the four punch presses meet certain part size and flange height requirements, they are sent to panel bending.

Software helps level-load the sheet metal department, which, as Buster explained, is a critical component to the entire system. Parts are nested in such a way that one color emerges from cutting in the most efficient way, with maximum material utilization—but again, all within the same color. The department does not nest parts in other color groups for a little better material utilization. It can’t. Manufacturing always occurs sequentially, one color after another, and one color isn’t released until the current color is finished.

What Schedule?

Like everywhere else, operators in the sheet metal department can’t work ahead in the schedule. They really can’t, not only because they need to stay “within the color,” but also because there is in fact no production schedule.

Instead, following the tenets of demand-pull manufacturing, downstream processes pull orders from upstream. To keep work flowing, operators who nest cases ahead of fabrication grab the next batch of ordered cases, always keeping three colors ahead; that is, they determine the jobs that will be in the three colors behind the most recent color released to the sheet metal department. This three-color buffer is large enough to ensure fabrication is never starved of work, but small enough so that the company can react to order changes or other unforeseen events.

“We generically know the life cycle [that is, total manufacturing time] of our product after we nest the cutting machines,” Buster said. “Orders are released to sheet metal fabrication only when assembly pulls cases into the build cycle. Once we nest, we’ve hit the metaphorical start button, and everyone starts producing in sequence by color. If you’re in welding, you weld in sequence. If you’re in subassembly, you work in sequence. If you’re in coil subassembly [in the refrigeration department], you build in sequence. Everybody just builds in sequence.”

And as described earlier, a display case’s movement into the FIFO lane as well as on to the main line triggers downstream processes into action. “Activities are based on reality, what’s really going on, not on a schedule,” Buster said.

As Buster explained, if the floor worked by a detailed schedule, departments would work to that schedule regardless of where everything else stood in the rest of the plant. This leads to bottlenecks, not to mention labor and space inefficiencies.

But with everything visual, the story changes. A worker can spot brightly colored job packets from 200 feet away. He sees that the work in front of him is purple while work downstream is blue—visual signals that tell him production is working as it should. If he sees that WIP carts in front of the next manufacturing step exceed a certain number, he knows that he needs to move downstream to help clear the bottleneck before it grows any larger. This, Buster said, is why cross training is so critical.

Visual cues tell people what to work on next. Once a color is finished, workers look downstream for bottlenecks. If one exists, they move to clear it. If none exist, they move on to the next color. When they get that next color, the team lead looks at the group of jobs within the color and decides how best to sequence them, so that everything as a group (not individual cases) is finished in the least amount of time possible.

“Least amount of time possible” pretty much sums up the company’s approach to continuous improvement.

Instead of being focused only on the job in front of them, everyone now is focused on velocity, or how fast jobs flow through the system.

A worker installs components in a salad bar case.

Quick changeover is part of it. So is smart fixturing, organized workspaces, accessible tools, shadow boards, controlled WIP, and all the rest. But all of it needs to be focused on increasing job velocity. This is how, in six years’ time, you get virtually the same number of people producing two-thirds more revenue.

“That,” Buster said, “is what makes us so competitive.”

Photos courtesy of Hillphoenix, Specialty Products Division, 641-437-1297, www.hillphoenix.com.

Salvagnini America Inc., 513-874-8284, www.salvagnini.com