Our Sites

How manufacturers uncover flow through operational excellence

How to make the most of monument machines

To the uninitiated, custom sheet metal fabrication can seem pretty straightforward. Lasers and punches cut parts that are then sorted for downstream deburring, bending, welding, and so on. Every part is cut, most parts are bent, many have hardware inserted, many are powder coated, and some go through assembly. Shouldn’t a fab shop be able to define value streams that help get the product flowing?

Many fab shops have cells dedicated to specific customers and product families. But the rest of the shop is full of shared resources grouped in departments, from the front office to the shipping dock. And for good reason. The company probably produces thousands of different parts with varying cycle times and processing requirements, which is a tough environment for cellular manufacturing.

The whole point of cellular manufacturing, with its carefully measured takt times and dedicated equipment, is to make manufacturing more predictable. And with operational excellence’s autonomous, self-healing, visual flow, anyone—any employee on or off the shop floor, any visiting customer, even a visitor who knows nothing about manufacturing—should be able to look at work in the shop and know whether it’s on time or late.

Picture a cart with work-in-process (WIP) staged before a cell in a first-in, first-out (FIFO) lane with three squares; one square can hold one cart. There’s a green square on the floor, labeled “normal.” Behind that is a yellow square and, finally, a red square for “abnormal.” Normal flow is good, abnormal flow is bad. If there is a cart in the red square, everyone intuitively knows that abnormal flow exists, and when trained workers see it, they know what corrective action to take before it snowballs into a larger problem. There are no production meetings, no alarm bells. Employees simply see a problem and fix it, returning everything to normal flow.

This is a central tenet to Kevin Duggan’s approach to continuous improvement. Duggan is the president and founder of the Institute for Operational Excellence, a firm that promotes an improvement method that focuses not on waste, but on flow. During a December webinar (and in previous books), Duggan defined operational excellence as “when each and every employee can see the flow of value to the customer and fix that flow before it breaks down.”

The idea is to create a self-healing system in which front-line workers need not call supervisors who in turn organize heated production meetings. Instead, workers can identify the abnormal flow and correct it on their own, freeing managers to work on offense—that is, top-line growth.

There’s a lot to this improvement technique, but one aspect that might be of particular interest to custom fabricators is its approach to shared resources.

A typical custom fabricator is full of shared resources. They’re everywhere. In fact, outside certain manufacturing cells dedicated to high-volume products (often dedicated to one or two major customers), everything is shared, from upfront estimating and planning to cutting, bending, welding, powder coating, and shipping.

If a fabricator can identify product or process families, certain previously shared resources could become dedicated. But some resources are monuments, designed for batch production and quite often physically difficult or impractical to move. Think of heat-treating ovens and powder coat lines.

So how does a fabricator manage these? During his December webinar, Duggan outlined his approach. He clarified that this is not the automotive model for handling shared resources, as that approach works off forecasts. When the production mix of products is different than what customers want, you get car dealerships with excess inventory offering cars at zero percent financing.

Process families include machines dedicated to producing a specified range of products. In this example, those products then flow to areas dedicated to specific product families.

Regardless, the questions presented here follow deductive reasoning. Some companies may well answer “no” to some of them. Still, Duggan said, it’s important to ask the questions; otherwise, a shop might miss uncovering unforeseen opportunities.

Boiled down, the method that follows exploits commonalities in mixed-mode production to define what “normal flow” is. If normal flow isn’t accurately defined, employees can’t identify abnormal flow and take corrective action—making a self-correcting system that’s core to Duggan’s entire approach.

Question 1: Can we identify product families, and can we extend them?

Improvement consultants suggest different ways to identify product families. When it comes to Duggan and his operational excellence technique, product families involve parts that are or could be run on dedicated resources.

“What companies tend to do is try to look at their [product-process matrix] grid and try to figure out their product families,” Duggan said. “But that’s virtually impossible. So to achieve a good flow, we first look only at where we can dedicate equipment [to certain products]. That is where we make a decision of where a product family is.”

A fabricator might identify potential (or perhaps current) cells or other processes of the factory dedicated to specific customers or products. And that’s fine, but what about the “long tail” of low-volume and custom products, the bread and butter of most fab shops?

Here is where “extending the families” comes into play. Using a value stream map and product-process matrix, a fabricator can identify processes that could be dedicated to certain products. The basic idea is to dedicate as many machines as possible to specific product families, all the “steady runners” in the product mix. With this, a shop can calculate machine loading more easily based on available capacity and number of shifts. This begins to create greater predictability and regularity around the flow of products through the shop.

Question 2: Can we create process families with dedicated equipment?

If a fabricator can’t answer “yes” to question No. 1 for its entire product mix (and most custom fab shops probably can’t), it can consider the potential of what Duggan calls process families. Each process family consists of a variety of parts that flow through dedicated equipment.

In his webinar, Duggan described one process family as a group of various parts that all flow through sawing, deburring, and drilling, before flowing to a range of assembly cells with equipment dedicated to specific product families. “In this case, the sawing, deburring, and drilling process can move work in single-piece-flow fashion to feed different product family assembly areas farther downstream,” Duggan said.

He added that to do this requires machines to have the right tools and fixtures to handle the full variety of work flowing through them. A brake operator, for instance, needs a press brake with a bed length, sufficient tonnage, and tooling crib large enough to handle all the jobs flowing through.

Setup times need to be quick too, flow through the processes needs to be visual, and machines need to be well-maintained. Smooth flow can’t survive in an environment rife with reactive maintenance.

A sequenced FIFO adapts the first-in, first-out concept for use with shared resources.

Duggan added that there’s a lot more math to this, and most of it is based on what he calls branch takt, which is a calculated number that tells how frequently a part needs to be produced at a shared resource to meet customer demand. Branch takt is the sum of all the demand on a shared resource divided by the rate of customer demand for those products.

“We call it a ‘branch takt’ because it’s a branch of many product families,” Duggan explained. “So it’s a composite takt of all the parts that [the process family] makes. And the closer the branch takt time is to our cycle times, the better we need to be on our setups and maintenance, as there’s less time to recover if things go wrong.”

Suppose a saw, grind, and deburr process has been identified as a process family. In other words, these three machines are now dedicated to processing jobs for a specific product family. The cycle time (setup, run time, and handling) of all the parts flowing through sawing never exceeds 10 minutes, grinding never exceeds 12 minutes, and deburring never exceeds 13 minutes. If the branch takt at each process was calculated to be 15 minutes, then we know each process would be able to keep up with demand.

A process family also requires what Duggan calls an EPEI, or “every part, every interval.” Precisely calculating intervals requires some detailed equations, but essentially, identifying intervals involves determining how long it takes to produce all the part numbers within a family.

Calculating the interval involves subtracting the total time needed to produce the parts (run time plus a variability factor) from the available working time. The time left over is for changeovers. “We then see if the targeted interval for the machine can be supported with the changeover time available,” Duggan said. “If it can’t, we try to reduce the changeover so the targeted interval can be supported. If, after aggressive changeover reductions are made, the targeted interval still cannot be achieved, then we may need to increase the interval at the shared resource.

“The smaller we can make the interval, the better the flow we have through the shared resource,” Duggan added, explaining that within a process family, the process with the largest interval wins—that is, it dictates the lead time of the entire process family.

Question 3: Can we create flow through a true shared resource?

Some operations in fabrication really can’t be dedicated to any product or process family. These might be batch processes like powder coating and heat treating, or perhaps a shared production resource like quality assurance. Duggan called these “true shared resources.”

Drawing from the lean toolbox, would first-in, first-out (FIFO) work? According to Duggan, not really. “FIFO is like pingpong balls going through a pipe. They always come out in the same order, and the pipe is only so big. Once you fill it, that’s it.”

Think of a full FIFO lane in front of a shared resource like powder coating. If the FIFO lane is full, by that logic, the upstream processes should stop producing until the FIFO lane partially clears. This makes sense in a dedicated line, where upstream processes would just fabricate excess WIP if they kept producing.

But for a shared resource, it’s not that simple. Putting parts from all product families into the same FIFO lane could cause tremendous variability in flow times—and, of course, the whole point is to shorten that flow time as much as possible.

Imagine if a job is held up in a forming cell. Various parts from other families file into the FIFO lane, then the late product from forming arrives. Now you have an already late product stuck behind a long line of other products in the FIFO lane. The supervisor could push the late job to the front of the line, but then that job makes other jobs—those that were once on time—behind schedule as well.

As Duggan explained, strategic sequencing with multiple FIFO lanes makes all the difference. It’s called a sequenced FIFO lane. In this situation, a shared resource would have multiple FIFO lanes next to each other, each of them feeding the shared process. The number of lanes depends on flow velocity and cycle times of the shared resource (which again is where some math comes into play). Often, there is one FIFO lane per product family feeding the shared resource, but the basic idea is to use multiple FIFO lines next to each other to achieve smooth, predictable flow through the shared resource.

In the current example, the late product from forming would flow to a dedicated FIFO lane for that product family. A standard sequence is developed to cycle through all the FIFO lanes, ensuring each type of job (one per FIFO lane) gets equal treatment on the shared resource and achieving what Duggan calls a “guaranteed turnaround time.”

“When you have a true shared resource in your operation, one of the things we’re trying to achieve is a guaranteed turnaround time,” Duggan said. “If a part enters any of the FIFO lanes at, say, 10 a.m., the parts will be available at 3 p.m.—guaranteed, no matter which lane it’s in. This makes your shared resources like FedEx.”

Additional standard work is also developed to determine what to do when a FIFO lane is full. “Do we continue cycling through the FIFO lanes normally? Do we give the full FIFO lane an extra turn, given that it’s in an abnormal condition? The exact answer will depend on a number of factors, but the idea is that there’s a predefined, standard response that is executed by the employees when abnormal flow occurs,” Duggan said.

Duggan further explained that the methods outlined here are just scratching the surface. The process of developing flow entails rigorous value stream mapping based on lean principles (not on free-flowing brainstorming sessions). Moreover, determining the guaranteed turnaround time requires some calculations and analyses that include variations in the shared resource’s cycle time for all the products flowing through. The shared resource also needs a process of determining what to work on next, usually achieved through a “mix indicator” that shows the order in which jobs are processed at the shared resource.

Regardless, the basic concept is to build a self-correcting system, where everyone can catch abnormal flow before it snowballs into a larger problem. For instance, how will workers know that the guaranteed turnaround time at a shared resource will be met? They’ll know if they place parts in a portion of the FIFO lane painted green, they will all be processed within the guaranteed turnaround time. If the parts are far enough back in the FIFO lane, it might be in an area painted yellow or red. That’s abnormal flow; workers know immediately that it’s broken and, ultimately, take corrective action to fix it. From here the cycle of self-healing continues.

Information for this article is based on “Untangling flow in shared processes, monument equipment, and job shops,” a webinar presented in December 2018 by Kevin Duggan, founder of the Institute for Operational Excellence, 1130 Ten Rod Road, Suite A-202, North Kingstown, RI 02852 401-667-0117, instituteopex.org.

About the Author
The Fabricator

Tim Heston

Senior Editor

2135 Point Blvd

Elgin, IL 60123

815-381-1314

Tim Heston, The Fabricator's senior editor, has covered the metal fabrication industry since 1998, starting his career at the American Welding Society's Welding Journal. Since then he has covered the full range of metal fabrication processes, from stamping, bending, and cutting to grinding and polishing. He joined The Fabricator's staff in October 2007.