Work Cells Can Apply in Process Plants

Work Cells Can Apply in Process Plants

Think flow, not function.

Peter L. King
 

Despite the successful application of work cells in discrete parts assembly plants, lean practitioners working in process plants may be reluctant to implement work cells because of difficulties relocating the massive pieces of equipment typically used. They miss a golden opportunity to improve flow, simplify changeovers, and increase throughput. As reflected in a successful lean application in a DuPont synthetic rubber manufacturing operation, cellular manufacturing can be practical in a process operation. It usually generates even more benefit than it does in discrete manufacturing.

By process operations, I mean plants that transform materials through chemical and physical processes to make paint, synthetic rubber, salad dressings, toothpaste, and the like. These operations differ from discrete assembly processes, which produce automobiles, refrigerators, cell phones, and computers (traditionally the prime application area for lean).
 

Deep Roots
Of all of the improvement tools in the lean toolbox, cellular manufacturing is one of the most powerful. It enables smaller lot production, more visible flow, quality improvements, reduced work in process (WIP), shorter leadtimes, and simplified pull replenishment systems.
 

Figure 1. A U-shaped cell.

The concept of work cells dates back more than 200 years. In the early 1800s, the Portsmouth Block Mills used work cells to produce blocks and pulleys for the British navy. The operation produced 130,000 pulleys annually, substantially reducing traditional cost and improving quality.

As manufacturers moved away from the traditional functional layout (with an area or shop dedicated to each craft or type of processing) toward cellular concepts, they discovered cellular manufacturing’s many benefits. Cells enable small-lot flow, sometimes one-piece flow. Flow becomes visual and therefore easier to manage and improve. Since a part processed on one machine is immediately fed to the next operation, quality feedback is immediate. Flow within a cell and flow from cell to cell can be better synchronized, enabling reduced WIP.

An additional advantage: Because a given cell processes only a subset of the entire parts catalog, with similar processing requirements, simpler and faster machine setups emerge.

Cell Application in Process Industries

	Figure 2. Typical process industry equipment footprint — functional configuration.

Massive process equipment (often referred to as “monuments”) deters widespread cellular manufacturing in process operations. Many plant managers perceive that equipment size and complex process interconnections make rearranging operations into U-shaped or L-shaped patterns impractical. (See Figure 1.) A paper forming machine, for example, is approximately 30’ x 150’, weighs many tons, and has process piping and hydraulic and pneumatic lines connected to it. Relocating this “monument” can cost several million dollars. Process vessels used to polymerize resins in paint manufacturing (typically sized for 5,000- to 10,000-gallon batches) have many process interconnections; relocating them would require similar funding.

A few process companies implementing lean learned years ago how to realize cellular manufacturing’s benefits without actual equipment relocation. By creating virtual work cells, they may achieve even greater performance gains.
 

Figure 3. Typical process industry flow patterns.

Typical Configurations
A generic configuration typical of many process industry plants is depicted in Figure 2. It has a small number of key processing steps and a few machines, tanks, or reaction vessels in parallel at each step. The parallel machines are quite similar. Often a specific material can be processed by any one of them.

This equipment configuration pattern often characterizes production of synthetic rubbers, plastic and paper sheet goods, automotive paints, staple fibers, plastic pellets, and a variety of other materials made in process plants. Such plants usually require this array of equipment to handle high production volumes. Practical equipment size limitations prohibit the design of a single machine or vessel large enough to process the required full throughput.

Process companies value this configuration’s flexibility. If a batch of material leaves Step 1, and one of the Step 2 machines is down for maintenance, other machines may be available to process the material. Flow paths typically resulting from this situation are shown in Figure 3. This pattern exploits the system’s inherent flexibility, generally with more negative than positive consequences.

If the step-to-step processing sequence is not well-synchronized, material can’t flow directly from one step to the next. It must be temporarily put into some type of storage. In a sheet goods process, for example, formed rolls typically do not flow directly to the heat treating or the slitting step. They move to an automatic roll storage system for later retrieval and transport to the next process step, creating large WIP storage. Operators find challenges visualizing and managing the unsynchronized flow or implementing pull.

Quality suffers for two reasons. As in parts assembly, significant time lapses between each process step prevent timely discovery of defects or out-of-spec material, making all intermediate material suspect. Even with the simple-looking arrangement depicted in Figure 3, 192 (4 x 4 x 3 x 4) possible flow path combinations exist. Because no two machines or vessels produce exactly the same product, that provides 192 different ways that process variabilities can add up. A Statistical Process Control (SPC) specialist would tell you that you don’t have a process, you have 192 different processes. With so many variables, root cause analysis of product defects becomes very difficult.

Because operators believe that there are alternate paths available whenever a piece of equipment fails, appropriately maintaining the equipment receives less attention. Over time, equipment performance as measured by uptime or Overall Equipment Effectiveness (OEE) deteriorates.
 

	Figure 4. Grouping into virtual work cells.

Overcoming Barriers
Cellular manufacturing could overcome most or all of these problems. As noted, because of equipment size and interconnectedness, rearranging it into U- or L-shaped arrangements is impractical. But if you realize that most of the benefits of cells can be attained without rearrangement, a process plant can reap the benefits by managing flow in a cellular fashion, without any equipment relocation. The key is to think in terms of flow rather than function.

The basic concept is straightforward: Start by grouping all materials into families requiring similar process conditions. Identify the process equipment required by each family. Then, instead of creating a work cell by rearranging the equipment, create virtual work cells by defining the acceptable flow patterns.

Figure 4 shows what this arrangement would look like. No equipment would be moved. Operators would define and follow new, more limited flow patterns.

This virtual work cell concept’s advantages are similar to those described for assembly processes. Understanding, visualizing, and managing flow is easier. As flow becomes more continuous, with far less material transported to storage, WIP and material handling decrease. Quality improves because feedback is much more immediate. With only four possible flow paths instead of the previous 192, product variability dramatically decreases.

Recognize that the numbers don’t always work out perfectly, but reasonable compromises giving most of the benefit can usually be found. In the case shown, since there are only three machines at Step 3, one must be shared between Cells 2 and 3. If that machine didn’t have enough capacity to process the total throughput of the two cells, sharing flow with machines 1 and 3 at Step 3 would be required. Even with this compromise, only six path combinations result, far better than the original 192.
 

Figure 5. Synthetic rubber manufacturing configuration.

Synthetic Rubber Production Facility
This virtual cell concept, applied to a Neoprene manufacturing operation, brought immediate and dramatic improvements. Capacity increased, leadtimes and inventory decreased, and yield improved.

Figure 5 depicts the Neoprene process at DuPont’s Louisville, KY, plant. It begins with three tanks where monomers and other ingredients were weighed so that the correct quantities could be fed into one of six polymerization kettles. After the batch polymerization was completed, the polymer was fed to one of three tanks for a chemical process called emulsion stripping. The stripped emulsion was then stored in tanks (not shown) and later cast onto one of six freeze rolls where the emulsion solidified into a thin sheet. The sheet was then gathered into a rope and cut into pellets, later bagged on one of three bagging lines, and palletized for shipment to the finished product warehouse.

Customers of this synthetic rubber used it to make high-pressure hoses, industrial belting, and gaskets and seals for refrigerator doors and car trunk lids.
 

	Figure 6. Synthetic rubber virtual cell configuration.

The product lineup encompassed three major families: type F, type J, and type R. Volume distribution among the families was in the approximate ratio of 45 percent F, 35 percent J, and 20 percent R. Each family comprised eight to 15 individual grades.

The pre-cellular scheduling process:

• Production needs for the coming month, by specific grade within each family, were set in a monthly sales and operations planning (S&OP) process.
• The plant production scheduler determined the grade with the most immediate customer due date, say grade J-43.
• The entire production facility was set up to make grade J-43. Production continued until the full monthly requirement for J-43 had been produced.
• The scheduler then determined the grade with the next most immediate customer due date, say F-6.
• The entire facility would be reconfigured to produce F-6. The monthly requirement of F-6 was produced.
• This scheduling process was repeated until all production requirements for the month had been met.

Operating in this manner created a number of problems:

• Each process area ran independently, with its own area supervisor, and with little coordination between areas. Flow was therefore nonsynchronous. Batches often waited for hours in the storage tanks between steps, and often got out of sequence. Asset productivity suffered as a result.
• Because flow patterns were confusing, and because there was no single individual responsible for overall flow, batches occasionally got pumped to the wrong place and then “ditched” (sent to the waste sewer).
• Large quantities of WIP usually resided in the storage tanks as a result of the flow discontinuities.
• Since each grade was produced only once per month, large quantities of finished product inventory had to be maintained in the warehouse.
• Transitions from one family to another were complex and time-consuming.
• After major transitions were mechanically complete, and the process restarted, achieving viscosity within customer specifications (the most important product characteristic) took several hours. Significant yield losses (material waste) resulted. The time required to reach required conditions also negatively impacted asset productivity.
• Implementing a pull replenishment system was nearly impossible because of flow discontinuity.

Synthetic Rubber Virtual Work Cells
Designing a cellular configuration for the rubber process overcame these problems. (See Figure 6.) The numbers worked out remarkably well in this case. Because there are three major product families, it seemed logical to create three virtual cells. And because there are either three or six pieces of equipment at each significant process step, dividing the equipment into cells was a straightforward change.

Even though product demand wasn’t equally distributed across families, with Type F representing 45 percent of the total demand, productivity improvement resulting from virtual cell implementation enabled even the Type F cell to gain enough capacity for producing to takt (the rate of customer demand).

Results from the cellular plan were dramatic. The most significant benefits resulted from reducing yield losses while getting back within viscosity specifications after a changeover. In the old scheme, it took several hours (perhaps four or more) to reach desired conditions after a change from F to J, J to R, or R to F. Because each new cell produced only grades within a type, target viscosity could be reached rapidly, typically within two to four minutes. This change not only improved yield, it improved asset productivity.

The organizational structure also changed from a horizontal alignment to a vertical one, based on flow rather than on function. Instead of an area supervisor for each process step, a flow manager for each cell is responsible for all equipment within that cell and for synchronizing flow within the cell.

Specific benefits recorded by the business:

• Scrapped material decreased by 28 percent.
• Finished product variability, measured by standard deviation of viscosity of first grade product, decreased by 15 percent.
• Leadtime through the complete process decreased by 28 percent.
• Average transition time declined from eight hours to three hours.
• Average time to reach aim viscosity targets dropped from five hours to five minutes.
• Usable capacity increased by several million pounds per year.
• Finished product inventory decreased 50 percent, a significant change.

These improvements required no equipment relocation or significant equipment modifications. Some process piping was removed, and some valves were locked out so that the cellular boundaries couldn’t be crossed inadvertently. Added changes included lines painted on the floor to designate cell assignment and signs hung over each process vessel or machine to indicate its cell. These slight costs were miniscule compared to the resulting benefits.

Applying cellular concepts to the manufacture of industrial fibers, plastic films, automotive paints, and paper-like sheet resulted in similarly impressive improvement levels.

Where the equipment configuration lends itself to cellular manufacturing, it should be designed and implemented before considering pull replenishment systems. Pull is far easier to put in place with the small number of flow paths required by a cellular arrangement compared to the large number of flow paths possible with a functional layout.

Summary
Viewed in terms of Taiichi Ohno’s seven wastes, often cited in lean literature, flow based on a traditional functional layout can cause most of the seven:

1. Overproduction: Like the synthetic rubber case, a functional layout generally leads to longer campaigns and thus excess production.
2. Inventory: Long campaigns create excess inventory.
3. Transportation: Inventory must be stored somewhere; in process industry plants, it can be in a fairly remote area.
4. Waiting: As described above, product changes can be quite long on equipment running the full range of products, causing longer wait times.
5. Defects: Product changes from one major family to another can result in significant yield loss on startup.
6. Processing: Yield losses don’t disappear on their own; someone must load them into a container for transportation to a waste disposal area. Moving them there is another cause of transportation waste. In some cases, yield losses can be reworked to become first grade, or be blended in small percentages with good product, both of which are examples of wasteful processing.

The only waste that will not generally be caused by a functional layout is the seventh, operator movement.

Cellular manufacturing can reduce or eliminate all of these wastes. It also addresses what is often called “the eighth waste,” the waste of human creativity and potential. If operators, mechanics, and schedulers operating the process on a daily basis are included on the cell design team, their intelligence, experience, and creativity are put to good use.

Anyone in a process operation who has dismissed cellular manufacturing as impractical or irrelevant has simply not looked at cellular manufacturing from the right perspective. The key is to think flow, not function.

Peter L. King has extensive experience in the process industries, including 40 years with DuPont Company applying lean and other manufacturing systems improvements to a wide variety of product types. He has consulted with food, carpet, lubrication oils, and apparel companies. King is a past president of the Process Industry Division of IIE. He is president of Lean Dynamics LLC, www.LeanDynamics.us. King wrote Lean for the Process Industries – Dealing With Complexity, Productivity Press, 2009. Portions of this article were drawn from the book, with permission from the publisher.