But, workcells have been around long before Toyota adopted them. Early workcells were most often seen in "job-shop" environments where volume was low and product variety was high. Their appearance and operation was quite different than the Toyota version and for good reasons.

As Toyota perfected their production system and migrated it through their company and their suppliers, they developed many rules, axioms and edicts about how their cells should look and function. This was necessary for spreading the knowledge quickly. It was effective because the volume and variety throughout Toyota and their supplier base was similar.

Now that Lean Manufacturing is spreading to other industries, the Toyota approach to cell design often creates problems. When the product-process mix does not fit the Japanese formula one of two things usually occurs:

1. Cellular Manufacturing is abandoned because people cannot see how it can work using the standard approaches.
2. The formulae and slogans are applied blindly with sub-optimum (and sometimes disastrous) results.

I have been designing workcells for about 25 years. Long before JIT, WCM, DFT, Lean and the various other acronyms. Out of this experience has come a structured approach to workcell design that fits virtually any process and product. The approach is based on fundamental, universal principles, sound engineering analysis and appreciation for the human side of these often-delicate socio-technical systems. The results are workcells that are appropriate for their products, markets, processes, people and environments. It is documented in my two books and many articles, seminars and presentations.

The approach has four major tasks:

1. Select The Products
2. Engineer The Process
3. Plan The Infrastructure
4. Design The Layout

#1 Select The Products
Here we examine the entire range of products that a factory manufactures and determine which products should have dedicated workcells and which products should be part of "families" that are appropriate for a "Group Technology" workcell.

When the products and processes are few, as in automotive industries, this task is simple and straightforward. It requires only experience and common sense, a few hours and a few meetings. We call this "Intuitive Grouping." In the Toyota literature, little attention is paid to this task because elaborate analyses are unnecessary.

But how do you sort out the product families when there are hundreds, thousands or even tens of thousands of different products? And, suppose these many products can use any number and combination of dozens or hundreds of machines, processes or components and in different sequences? In such situations, we must use more sophisticated methods. Production Flow Analysis, pioneered by John Burbidge, is one technique for sorting out more complex product-process matrices. Coding and Classification is another.

Whatever methods are used, at the end of the Product Selection task, we should have a list of products and/or product families. Each product or family will have a workcell for its manufacture.

#2 Engineer The Process
We generally begin this task with an existing or proposed process for a particular workcell. A detailed process chart or Value Stream Map is the best representation. First, we streamline and simplify the process. Then, we modify it for cellular manufacturing. For example, do we need smaller-scale equipment?

Next, we need estimates of time for each step in the process. We need the time that each step requires for a machine, a time that each step requires a person and times for each setup or changeover. With simple Toyota-like cells, Takt time will suffice. In more complex situations, all of the above information, perhaps for each product, must be estimated. This is a lot of work and can get a bit complicated. However, it is necessary for a sound design. From these times we calculate:

• How many machines/workstations we need
• How many people we need
• A reasonable range for lot sizes

#3 Design The Infrastructure
Infrastructure supports the processes within a workcell but does not participate in the process or change the product. It includes scheduling methods, material handling, supervision, motivation, work balance, training and work assignments.

Infrastructure is often the hidden part of a workcell. Because it is not immediately visible, it tends to be ignored. However, it is just as important as the process engineering and product selection. When it comes to infrastructure, there are many available methods and techniques that are not widely known. It is easy for designers to come up with sub-optimum solutions as a result of their failure to consider the many available possibilities.

The Toyota history and emphasis in Lean Manufacturing tends to hide the full range of infrastructure options. Since many of us learned Toyota methods and approaches, we may not know about other approaches for non-Toyota situations. Work balance is a good (but not the only) example.

Balancing Workcell
Most Toyota cells have a single product or a highly similar family. The process is linear, sequential and identical for all products in the family. Because of this, the most appropriate balance methods are Inherent Balance and Circulation (Chaku-Chaku). With Inherent Balance, we attempt to break the work into micro-tasks and assign micro-tasks to operators in a way that the total work times are identical (or nearly so). With Circulation, operators follow a single part or batch through the cell and perform every task.

Balancing cells where process sequences, batches or times differ from product to product requires different methods. These other methods include excess capacity and floating balance. Our web page, Balancing Workcells, examines this issue and shows the various methods for balancing. In addition, you may download the paper on this topic that I delivered to the 2000 conference of the Institute of Industrial Engineers. It addresses work cell balance decisions in more detail. Here are two considerations for balancing workcells:

1. Separate people from machines. Imbalance in the operator's work is a serious problem. Imbalance among machines or workstations is often inconsequential and frequently advantageous. Besides, people and machines have different characteristics and can use different methods.
2. Recognize two forms of balance: Static and Dynamic. Static balance refers to the average work content of operations over periods of hours or days. Dynamic balance refers to the momentary piece-to-piece balance. If some process steps are not consistent in their work times, a cell can be statically balanced but highly imbalanced dynamically. This leads to differing solutions to the balance problem.

#4 Plan The Layout
If the first three tasks are done well, the layout is usually straightforward and easy. On the other hand, if the first three tasks have been neglected, layout can be a nightmare as unresolved issues surface. These unresolved questions force changes, cause dissension and promote confusion.

One consideration in workcell layouts is the shape of the cell. Toyota-type cells favor the U-shape. Indeed, the U-shape cell has many advantages when the process is consistent, linear and sequential.

However, U-shaped layouts are often impractical for more complex cells. Here, the layouts take many and varied forms. They must accommodate multiple process sequences, differing batches and multiple scheduling approaches.

Summary
Workcell design, like most engineering designs, is itself a process (or should be). It consists of a series of logical steps culminating in a viable set of design documents that show the layout of a workcell and specify the methods of operation. This design includes the process, the arrangement and supporting infrastructure.

As in other forms of engineering designs, copying of another design, superficial analysis or incomplete knowledge results in sub-optimum solutions and is often dangerous.

Quarterman Lee ("Q") started his career at the fountainhead of Lean Manufacturing, Ford Motor Company. He has worked in foundries, paper mills and a window manufacturing plant in positions from Engineer to Plant Manager. Since 1977 he has been consulting, training and writing. Mr. Lee has authored two books and hundreds of articles and programs. Mr. Lee is President of Strategos, Inc.

For more detailed information you may visit our website at www.strategosinc.com. You may also download a free copy of Chapter 4 from Mr. Lee's forthcoming second edition of "Facilities and Workplace Design".