Tangled Mess

Many of the world’s factories are designed in ways detrimental or unhealthy to product quality. The typically complex and lengthy flow times multiply process variables, which can obscure and delay the discovery of defects and nonconformities. This can thwart efforts to trace the origins of problems, improve processes and make them more efficient.

Two deficiencies stand out at many factories: Crisscrossed process-to-process pathways can snarl audit trails, and too many product varieties running on too few production lines or cells can lengthen plant-to-customer lead times. Both deficiencies delay discovery of defects and leave a cold trail of causes.

These deficiencies are quite common and exist even in well-regarded factories. Rectifying these flaws in factory design and equipment may be viewed as within the purview of production engineering and operations management. But those personnel rarely address these matters, in large part because they’re not aware of quality implications.

Why? How could quality not be of foremost concern to engineering and operations—and throughout the enterprise? One obvious explanation is the proverbial silo system, which may come into play when an organization treats quality as a separate organizational function.

Another explanation related to the silo syndrome has to do with an organization’s financial policies. High hurdles, set in the name of capital cost frugality, sometimes prevent the purchase of additional equipment that could improve quality and lower overall costs.

There’s a way to deal with these issues. Regulatory agencies can include factory design for quality in their standards and policies, which will bring attention to the matter. In turn, the quality community must:

- Broaden its standards to include matters related to factory equipment and organization.
- Upgrade its concepts of good factory design with particular emphasis on quality in engineering and operations areas.

To further address factory design flaws that adversely affect quality, an organization must be willing to dissect and analyze its processes to find stumbling blocks such as entangled flow paths, long flow times and limited knowledge or resources for shifting mindsets and changing preconceived notions about quality, production and operations.

One-to-one responsibility chains
Crisscrossing process flows occur when two or more production associates grab and process parts from a common source—such as a moving conveyor—and then feed the work forward to another phase of work or set of hands. Among those hands, some might be unreliable and prone to causing defects, while others are unclean or germ-ridden, which is particularly a problem in the food processing, healthcare and safety products industries.¹

In cases of automated production, which ideally should minimize the chances of human contamination or mishap, too many mechanical processors might grab and work on parts, which then can confound identity and sources of nonconformances. Figures 1 and 2 show the growing disarray and opportunities for quality problems as crisscrossing extends through stages of manufacturing.

Figure 1

Figure 2

There are four groups of three production associates in Figure 1. Assemblers in groups one, two and four work from wide conveyor belts using parts coming from external sources. Final assemblies go to any one of the three testers in the third cluster.

Figure 2 shows the same scene as Figure 1 but without the symbols for incoming parts and with arrows showing the assorted pathways through the four process stages. The total number of flow paths is 54. With that many flow paths, the sources of any cases of bad workmanship tend to be obscured.

The difficulty of pinpointing the fault when nonconformities are discovered may invite carelessness while degrading motivation to do the job right. Efforts to fix defects and prevent mishaps are hampered because quality assurance (QA) would need to consider root causes along 54 different pathways. Even if the 12 people are replaced with 12 machines, such as robotic assemblers, automatic testers and robotic packaging, the number of flow paths is the same large number: 54.

But untangling these process flows doesn’t need to be difficult. The method in Figure 3 shows how to convert from a single line with wide belts and many flow paths to three work cells, each with a narrow channel conveyor or just a work table, and one-to-one flows. The number of flow paths is reduced from 54 to three—each of the three cells being a single-flow path.

Figure 3

Beginning with the assembler next to the icon labeled "parts" and ending with the third packaging associate, the workflow in each cell is from person to person. The effect is similar to that of a bucket brigade: Motivation is high not to fumble or drop a bucket and spill water, so quality at the source is built in.

In the cellular mode, the teams usually are loath to tolerate a slacker or someone prone to errors, unclean or sick. Instead of requiring supervisory intervention, peer pressure may get the offender to shape up or find other employment.

"Cells create responsibility centers where none existed before," said Deborah Davis of Baxter Healthcare, describing cells at Baxter’s plants in Mexico. "Procrastination, finger-pointing and excuses fade. The stage is set for conversion to a culture of continuous improvement."²

Additional longer-term quality benefits can result from the cross-training feature that’s standard practice in cellular operations. Figure 3 shows the possibilities of cell associates learning from rotating through the eight positions in the workflow. Each cell associate can learn how a small nonconformity at one step might develop into serious quality problems down the line and beyond. More importantly, members acquire a sense of the full multistep process, just as a quality or process engineer studying and troubleshooting the process would.

With that whole-process awareness, cell members may gain sufficient understanding to resolve many quality problems themselves before miscues quickly multiply. Cell members also may learn to recognize and record root causes, and be able to participate effectively with QA personnel and process improvement teams in larger problem-solving efforts.

In an alternative scenario, machines can be organized into cells. The number of machine operators may range from one operator tending each machine to one operator tending all machines, perhaps in sequential cycles. Regardless of the number of operators in the cell, the suggestions about cross-training, high motivation for quality work and complete process awareness still apply. In the case of a fully automated cell—containing no operators—these benefits fall to the technical support team assigned.

So, how do we sum up the proper design of these types of assembly lines? Manufacturers should strive for process routings segmented into one-to-one flow paths—a single, fixed next-process destination at each step in the flow. One-to-one chains compress responsibilities for results and pinpoint targets for quality and process improvement.

This quality principle of one-to-one chains can apply to many industries and most critically to food, healthcare and safety products in which crisscrossing flow paths can confuse the chain of responsibility and open the door to life-threatening product nonconformities.

Too many products, too little time
Another common factory design that can adversely affect quality occurs when organizations run too many product varieties in too few production facilities.
Sizeable inventories and long flow times are guaranteed. Large inventories provide hiding places for defects and nonconformities. Extended flow times introduce more conditions and variables that confound efforts to isolate root causes.

One way of coping is to segment production and production support resources into value streams, also known as product families or customer families. Figure 4 is an example in which the processing of 100 SKUs is segmented into four value streams labeled A, B, C and D:

- The three A’s are very high-volume products with stable demand.
- The seven B’s are newer products involving advanced technologies.
- The 20 C’s are high-profit SKUs undergoing continual material and process upgrades.
- The 70 D’s are mature, lower-volume items.

Figure 4

Partitioning the 100 SKUs this way simplifies quality control and QA because each value stream involves far fewer conditions and variables than if the 100 were mixed. Partitioning also simplifies production scheduling and control, and delivery-date promising—a matter of higher-quality information and product delivery between the company and the customer. This discussion also can apply to made-to-order job shops in which 100 customer orders, instead of 100 SKUs, are production items competing for run time on just four value streams.

The value-stream organization shown in Figure 4 is, indeed, an improvement—less congestion and complexity-—than with 100 mixed SKUs. Still, with just four productive units, only four products can enter or finish production at any one time, while the other 96 wait. Typically, any given product is run in fairly large lots because with so few production lines, it won’t be scheduled again for days or weeks. Those days or weeks equate to large inventories, which can magnify the effect of undetected defects. The long lead times can leave cold, contaminated audit trails.

Organizations tend to deal with the large production-lot issue in two primary ways:

Invest in equipment to speed production, which can introduce problems that offset some gains.
Employ devices and methods that enable quick changeover from one SKU to the next and allow many more SKUs to be produced and sent to market in a given period.
This best practice—quick changeover—is always desirable and beneficial, but it’s rarely sufficient. Referring to Figure 4, there are still only four production facilities for 100 SKUs, so production will still be considerably out of sync with customer use: long and variable lead times, inventories high on some items and low on others, and often-buried quality issues.

Re-equip for quick response
Figure 5 provides a possible solution: raising the number of productive units from four to 20. While on the surface this solution seems too costly, if done right—accounting for quality and other benefits—it may prove cost effective.

Figure 5

Under the scheme shown, each of the three A’s gets its own dedicated line or cell. Production of A1, A2 and A3 is closely synchronized with customers’ use rates, so inventories are low and quality trace-backs are short and straightforward.

Moreover, a dedicated line or cell is likely to gain enough attention over time to become relatively problem-free with less likelihood of nonconforming outputs. If automated, it may be equipped with fail-safe and vision-checking devices. If manual, experienced operators may acquire sensory capabilities that detect slight variations from correct, standard operations.

Similarly, some of the B, C and D SKUs may be produced on dedicated lines or cells, while others may be produced in semidedicated cells on a rotating schedule. For the 100 SKUs, channel inventories may be reduced by an average factor of five so defects and their root causes can be found five times more quickly.

There are some notable examples of production facilities equipped with large numbers of lines or cells, as shown Figure 5. Every day at the Ariens plant in Brillion, WI, 35 cells assemble every seasonal product sold, including mowers and snow-removal equipment.³ At AmorePacific’s cosmetics plant in Suwon, Korea, assembly is organized into 23 compact cells making 23 different cosmetic SKUs concurrently.⁴

These examples, though, are exceptions. The vast majority of factories are configured with a small number of productive units. Each SKU is made infrequently, and long lead times plague discovery of defects and other problems, as well as the the ability to address them.

Many organizations may be reluctant to add more production lines or cells because they assume the cost is prohibitive. In some industries, it may be. For example, given the mass and weight of automobiles and the bulky, expensive automation equipment involved, having assembly equipped and organized into multiple value streams for simultaneous assembly of numerous car types and models is not feasible.

That limitation does not apply to the manufacture of many other products, which in comparison to automobiles are modest in size, weight and production technology. The five-fold increase in number of cells shown in Figure 5 is most likely reflective of relatively simple, low-cost equipment.

Slower speeds, shorter lines
In cases of many SKUs and few production lines, there is usually relentless pressure for greater output on the existing lines. Here’s what can happen:

Operations personnel run more shifts per week at speeds often exceeding rated equipment specifications and straining workforce capabilities. Typical negative results are frequent jam-ups, breakdowns, and off-spec or damaged product.
Maintenance personnel, instead of being focused on prevention activities, spend most of their time fighting fires and, sometimes, cleaning up messes left from jam-ups. Additional shifts per week further cut into preventive maintenance, so assembly line downtimes become more frequent and severe. QA personnel have their hands full checking for reworkable product and attempting to determine root causes.
Larger production engineering staffs continually work over the long haul on equipment upgrades: speedier and more complex feeders, mixers, assemblers, inverters, cappers, scales, packers and computer controllers.
Processors that used to be single stations now become dual stations—and later even four or six stations.
Conveyors, once single channeled and short, widen and lengthen so products six—even 10—abreast are sped forward to successive multistation cappers and wrappers. At high speeds, for example, ever lighter, thinner-skinned cans, bottles and packages are problematic, creating spectacular accidents and quality messes when imperfections of product, container or equipment occur.^{5, 6}
The return on these investments in high speed may not be good. In terms of benefits, there may be more output and somewhat shortened lead times. But on the cost side, there are never-ending years of expenditures on advanced technologies, growing process complexity with unpredictability, increasing difficulties determining causes of random line stoppages and nonconforming product, and high maintenance.

On top of everything, the central imbalance remains: more SKUs than production units to make them. Production, therefore, is only weakly tied to actual customer demand, lead times are still overly long, stock shortages of some products and excesses of others downstream become common, and product quality in the distribution system is too often compromised by exceeded sell-by dates.

The solution to these issues? Factories can use another quality principle on sufficiency of productive units. Increase the ratio of productive units (production lines, cells, teams and machines) to product types. Production-to-use times can be shortened and less cluttered. Nonconformities can be found and addressed more quickly before undetected defects can multiply and while causal evidence is fresh. Incidences of products beyond their sell-by date become rare.

Moreover, with greater numbers of productive units, some may be dedicated to and perfected for a single key product type, a single customer or families of highly similar items. These dedicated units can result in:

- Quality levels driven toward zero defects.
- Relieved pressures to push output rates beyond rated capacities.
- Production equipped with multiple simple, low-cost productive units operated relatively slowly with better controls on quality.

Institutional, conceptual implications
When grocery stores are ordered to clear their shelves of lettuce, sprouts or meats found with E. coli, salmonella or mold, why does it seem to take so long to find the sources of contamination? When medicines are found with strange odors or spots, why does it seem to take so long to pinpoint the cause of the problem? When tens of thousands of cars are recalled for gasoline tank leaks or problems with accelerators, what’s the delay in finding the answers?

Maybe there are "too many hands in the pot" or, more formally, crisscrossing flow paths that are plaguing the food and other industries. For example, slow response to finding the causes of vehicles’ leaky fuel tanks and accelerator problems is closely related to their production in mass quantities weeks or months in advance of sale and use. Manufacturers are forced to produce in mass quantities because they have far too few productive resources to produce their many SKUs concurrently with short customer lead times and audit trails.

Learning resources for manufacturers appear scarce, too. In production engineering and operations circles, there seems to be a lack of awareness or respect for quality implications. In reviewing 10 books that seemed to be likely resources on these related topics, all included discussions on factories and equipment, and they featured best manufacturing practices, including the Toyota Production System and lean manufacturing. Each book contained only short index listings for "quality," and connections between quality and factory organization, equipment and operation were scarce.

In addition, many organizations treat concerns about the costs for additional productive units as separate issues, and they don’t weigh the costs against product quality costs and costs of poor quality service along the downstream value chain. Some in the quality community also may be at fault for narrow vision—a little q rather than a big Q quality outlook.7 In short, the silo syndrome intervenes.

There’s no mystery when it comes to breaking down organizational silos. It’s a matter of multifunctional deliberation, problem solving and decision making. Most well-run organizations know how to do this, but seemingly, too few actually do it. Designing better factories to improve in-process and outgoing quality with quicker, more effective corrective action must be carried to multifunctional meetings. QA may serve well as the carrier of the message.

There is ample room for more targeted approaches. QA might lead a Six Sigma project to correct the ills of factory design and equipment that degrade quality along the value chain. QA might implement the two principles of one-to-one flow paths and sufficiency of productive units, which already may be accounted for in various quality standards and policies.

With a suitably strong push from the various quality entities, the two principles may influence production engineering and operations enough so they become incorporated into their best practices of factory design and equipment.

References 
Elizabeth Dwoskin, "Your Dinner Has Been Touched by Multitudes," Bloomberg BusinessWeek, Aug. 25, 2011, pp. 29-30.
Rajan Suri, Quick Response Manufacturing, Productivity Inc., 1998, p. 97.
Robert W. Hall, "Vigorous Locally, Competitive Globally," Target, No. 2, 2004, pp. 7-16.
Author’s personal visit to AmorePacific’s cosmetics plant in Suwon, Korea, on Dec. 3, 2004.
Richard J. Schonberger, Best Practices in Lean Six Sigma Process Improvement: A Deeper Look With Telling Evidence From the Leanness Studies, Wiley, 2008, pp. 189-191.
Richard J. Schonberger, "Can Lean Find Its Way in Packaged Goods?" Target, May 19, 2011, pp. 19-24.
Big Q and little q are terms used to contrast the difference between managing for quality in all business processes and products (big Q) and managing for quality in a limited capacity—traditionally only in factory products and processes (little q). "Quality Glossary," Quality Progress, June 2007, p. 41.

Article & Image Credits: QP