Managing part flow in custom metal fabrication

Trying to get a handle on part flow? Look to your operators

POLCA, or paired-cell overlapping loops of cards with authorization, helps workers manage capacity. In this case, employees in bending cell B haven’t sent any cards back to the laser/punch operator saying they can accept more parts. So the laser/punch operator skips that cell until he receives a card saying that the cell can handle more parts again. Why can’t bending cell B accept more parts? Because it hasn’t received any cards from assembly cell B. In this case, cross-trained workers from bending cell B may assist those in assembly cell B to move product through the plant.

Managers at a custom fabricator want to organize machines into cells. They follow the logic that every part is cut and most parts require bending, spot welding, and hardware insertion. Why not put these three processes together? Would this not improve part flow?

To make this happen, managers examine the shop’s product mix and identify product families. Which products have similar requirements, manufacturing steps, and cycle times? A laser cutting system or punch press has a much shorter cycle time than a press brake, so to balance the line (or to get as close to “balanced” as practical), should the cell have a single cutting system feeding multiple press brakes?

“It depends, really. It’s a classic case in American manufacturing. We always like to focus on the solution, but we’re rarely successful unless we step back and define the problem.”

So said Bill Ritchie, president of Dayton, Ohio-based Tempus Institute. His consulting firm specializes in quick-response manufacturing (QRM), an improvement method for high-product-mix manufacturers, including job shops and contract manufacturers.

When people think about line balancing in custom fabrication, they usually think it’s impossible. A modern fiber laser cuts blazingly fast; meanwhile, press brake operators need to call up a program, change out tooling, and perhaps run some tryout parts to get a job started. How can both the laser and press brake possibly keep pace with anything remotely resembling a standard takt time?

According to Ritchie, scrutinizing press brake and laser cutting cycle times may be part of a larger part flow strategy, but it’s actually a small part, and it’s certainly not the place to start. Instead, he suggests starting with something that without which the business couldn’t exist: the customer.

Look at the Big Picture First

When Rajan Suri of the University of Wisconsin-Madison developed QRM, he focused on what he calls MCT, or manufacturing critical-path time. In his books, including It’s About Time, Suri goes into great detail about MCT. The concept involves purchasing decisions, front-office processing, and more, though it roughly equates to the “order-to-ship cycle.”

Shortening MCT increases a company’s capacity to produce products. A cell on the floor that shaves a day or two off of MCT is great, but what about the weeks the job spends in the front office? All that is part of MCT too. Streamlining part flow on the floor may look great, but if it still takes weeks to get an order ready for the shop, customers may not see the difference.

When it comes to improving part flow, Ritchie suggests going back one step further. “Again, we need to step back and define the problem,” he said. “What is it that we want to do? What is our strategy? What kind of business do we want to be? How will putting a cell together, and the resulting lead time reduction, meet the needs of the market, and where we want to go? And what are the opportunities?”

Say a fabricator makes the initiative to “lean out” a portion of the shop, and managers choose to build a cell for a product family of similar parts for several customers in the telecom business. Customers continue to put on pricing pressure, so the fab shop’s managers are hunting for cost savings.

People start moving machines, surrounding each laser with a handful of press brakes, hardware insertion machines, and spot welders. Parts flow smoothly through the cell and go on to powder coating. A few days are shaved off lead time, and everything looks great.

Then a few longtime telecom customers start putting on even more pricing pressure. They start shopping around, and the fabricator loses work to a competitor, though managers can’t for the life of them figure out how that competitor is making money on the job. The cells continue to work great—when they have parts to produce, that is.

“It’s unfortunate that so many lean events are essentially tactical,” Ritchie said. “But first, you really need to think about the overarching strategy within the business. How is the business going to grow? Are we going to [implement improvement strategies] for a product that the market doesn’t value, or where we don’t have any real advantage? Or are we going to tackle something that is consistent with our thoughts on what our strategies are for improving and growing the business?”

Consider again the fabricator serving those telecom customers. Looking at the big picture, managers might rethink their strategy. Do we want to continue serving telecom customers? Or do we want to become more competitive in subassemblies for the transportation sector? Which markets take advantage of the shop’s strengths, and among those, which of them has the greatest potential for growth?

These questions start to uncover what Suri calls a focused target market segment, or FTMS. Note that in his book It’s About Time, Suri says that defining an FTMS involves not only external customers and prospects—where the sales team sees the greatest potential—but also internal customers, such as the people performing downstream manufacturing steps.

An FTMS doesn’t have to include products sold only to a particular sector. It can, for instance, include products that may be for different customers and industries, but they all share similar physical attributes and manufacturing steps. Internal customers may be engineers and other personnel in the front office who would benefit from grouping certain complex jobs (those with a lot of drawing revisions, bill of material changes, etc.) together in an FTMS.

Suri sums it up in his book this way: “You begin by looking for a situation where there is a clear opportunity for benefit through lead time reduction.”

In the previous example, the fabricator may have rethought its strategy and identified an FTMS to include product families that serve not just telecom but also transportation customers, where sales sees the most growth potential over the next few years.

The Nitty-Gritty of Cell Design

With the FTMS and product families identified, managers move ahead with improving flow. Ritchie said that some sheet metal fabrication operations have benefited from designing cells not necessarily for laser cutting or punching, but for downstream manufacturing steps.

He recalled one company that designed cells in which several high-speed lasers fed about four cells, each with a press brake, hardware insertion machine, and spot welder. The setup made sense for the fabricator’s circumstance. Managers considered that the lasers’ cycle time was much faster than the bending and hardware processes downstream. A couple of lasers could feed all cells at once and still keep pace.

Ritchie added, though, that the best machine arrangement really depends on the shop, its people, and mix of products. Machine cycle time isn’t the only consideration. Other factors play a role, including communication.

Say a shop designs a cell with a laser/punch combo system, two press brakes, and two hardware machines. The laser/punch is so fast that it’s not running all the time. If the shop were measuring overall equipment effectiveness (OEE), the laser/punch uptime would be abysmal; it’s idle because operators are helping the brake and hardware operators process parts through the cell.

So why is the laser/punch combo machine in the cell? It turns out that for this product family, edge and hole quality in certain blanks is critical. The brake operator receives the first batch of 10 blanks, looks at the edges, then walks over to the laser/punch operator and tells him about a problem: A punch needs to be replaced. In this situation, the communication between the cell’s team members saved days of lead time.

As always, the best solution depends on the application and situation. Regardless, in this scenario, if the shop had focused solely on machine uptime, Ritchie said that it probably wouldn’t have designed the cell this way, and lead time reduction would never have been realized.

About Utilization

If a machine has low OEE, it’s spending a good part of a shift not producing parts. Traditionally, that means the machine “isn’t making money.”

Ritchie chuckled at this. “I’ve had arguments with people around the country about this. People love data. They love OEE and utilization metrics.”

He added that the reason behind a low OEE score is important. If a machine isn’t producing parts when it needs to produce parts, there’s a problem. Raw stock may not be available; an operator may not be present; changeovers may take forever—all that is worthy of scrutiny. But if operators need to leave their machines to help others process products, that may be a very good thing.

Put another way, if jobs are flowing, an idle machine isn’t a problem. If a laser operator is running extra parts people don’t need just to increase machine utilization, that’s a problem.

Ritchie added that this also applies to material utilization, when a laser programmer looks ahead in the schedule to fill parts on the sheet. More often than not, the time spent managing those extra parts costs more than managing sheet remnants and living with a little less material utilization.

Line Balancing

Is it really possible to “balance the line” at a custom fabricator, where machine cycle times, setups, routings, material handling requirements, and other factors change every day?

According to Ritchie, such a high-product-mix operation can indeed balance its line, but not by having machines running product all day at a constant takt time. It’s not about machines running at all, in fact. It’s about having products flow through the operation at a consistent rate, set by market demand. Again, depending on the product mix, machines may be idle, but with a balanced line, products should be moving. So how can a shop make this happen?

“Let’s say when you create a QRM cell, you have five people running eight machines,” Ritchie said. “Those people represent your line balancing—the people who figure out the best way for the product to flow through the cell and the entire business.”

According to QRM, people do this using several scheduling and capacity-planning tools. Consider again the two lasers feeding four bending-hardware-spot welding cells. The planner gives each laser operator a schedule of jobs that need to be completed, each feeding one cell after the other—A, B, C, and D—but this schedule isn’t set in stone.

Operators in each cell receive blanks and begin to process work through the cells. Each machine in the cell isn’t scheduled individually. Instead, cross-trained operators work together to decide how best to process the work through the cell’s press brake bending, hardware insertion, and spot welding.

Because this is a high-product-mix operation, some jobs take longer than others. To account for this variation, Suri developed a method he calls POLCA, or paired-cell overlapping loops of cards with authorization. It’s a kanban-like system, though not for specific products, but instead for capacity.

Here’s how POLCA works. Employees in the bending cells send cards (each representing one cart, empty pallet, or otherwise defined amount of work) to the laser operator, telling him that they can accept more work. If the laser operator runs out of cards, he knows to skip a nest of parts destined for cell B and start work on cell C. He continues to supply parts to the other cells, and starts making parts for cell B only after he receives cards from cell B telling him that it can accept parts (see figure).

The system helps control the amount of work-in-process between operations. Of course, starving the shop of needed parts is never a good thing. Some WIP buffer is needed between the laser and bending cells to account for variation; the number of cards operators can work with determines the amount of WIP. For instance, five cards can represent a WIP buffer of five pallets. Still, as Ritchie explained, the more reliable an operation becomes with proper machine maintenance, material staging, streamlined setups, and the like, the less WIP it may need.

The less WIP you have, the more efficiently work flows through the shop, the less cash you have tied up in inventory, and the less time it takes to deliver quality parts and products—which, of course, is the whole point of improving part flow in the first place.

Tempus Institute, 937-630-3035, tempusinstitute.com

Center for Quick Response Manufacturing, 608-262-4709, qrm.engr.wisc.edu

Dealing With Variability: Inventory, People, and Machines

As the late Dick Kallage explained at his LeanFab seminars, organized by the Fabricators & Manufacturers Association International, custom fabrication companies by their nature must deal with a lot of variability. The trick is to reduce that variability as much as possible with documented procedures, proactive machine maintenance, cross training, and streamlined changeovers.

Still, in the job shop world, some variability will remain. As Kallage has said, “You can absorb that variability with time, inventory, or capacity.”

These days competitive manufacturers rarely can absorb variability with more time. This leaves inventory and capacity. Traditionally, manufacturers have chosen inventory.

As Bill Ritchie, president of Tempus Institute, explained, they do this partly because of how the generally accepted accounting principles (GAAP) treats inventory, as an asset to be sold, even though selling that inventory (especially a large amount of it) isn’t guaranteed.

They also do this because producing extra parts on a machine reduces the number of changeovers and helps short-term utilization metrics and overhead absorption—again, even though the parts aren’t needed immediately and may never be sold.

At a fab shop, turning certain pieces of raw stock into something that will be sold is almost a sure thing. Sometimes a healthy raw stock buffer, particularly for those items that can’t be obtained immediately, can help a custom fabricator deal with variability in demand. “But I always look at WIP and finished-goods inventory with suspicion,” Ritchie said.

Regardless, inventory is still just a number on the balance sheet. “If you’ve got $2 million in WIP, take a million out of that and buy a machine or hire people. Now you’ve added capacity and throughput capability, and it all came out of the same balance sheet.”